# AI-Powered Real-Time Recommender > Source: https://www.recombee.com > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. Trusted by 10,000+ sites and apps Apply AI ## Empower your Platform with AI Personalization Explore a feature-rich Recommender & Search solution tailored for your unique use case. Video AI Personalization ## Maximize Engagement With Real‑Time Video Recommendations Surface the right content at the right moment with recommendations that **adapt in milliseconds**. Increase **watch time, completion, subscriptions, and retention** across every device. [Explore More ->](https://www.recombee.com/domains/video) Use Cases Examples ![Last Chance](https://www.recombee.com/img/use-cases/video/last-chance.png) ![Trending In Your Country](https://www.recombee.com/img/use-cases/video/trending-in-your-country.png) ![Fully Personalized Homepage](https://www.recombee.com/img/use-cases/video/fully-personalized-homepage.png) ![Top Genres For You](https://www.recombee.com/img/use-cases/video/top-genres-for-you.png) ![Editors' Picks For You](https://www.recombee.com/img/use-cases/video/editorial-picks-for-you.png) Fully Personalized Homepage ## Personalization Powerhouse helps you make a unique impression through a personalized user experience. [![Recommendations](https://www.recombee.com/img/homepage-new/recommendations.svg)Recommendations->](https://www.recombee.com/features/recommendations-search) [![Search](https://www.recombee.com/img/homepage-new/search.svg)Search->](https://www.recombee.com/features/recommendations-search#personalized-semantic-search) [Perfect Flexibility](https://www.recombee.com/features/full-control-with-scenario-settings) [Quick & Easy Integration](https://www.recombee.com/features/integration) [![Analytics & Insights](https://www.recombee.com/img/homepage-new/analytics-and-insights.svg)Analytics & Insights->](https://www.recombee.com/features/real-time-analytics-insights) Why Recombee ## Top-Notch Technology Supporting Your Business ### Full Control #### Follow your product vision by setting specific behavior for each box with recommendations. ![Full Control](https://www.recombee.com/img/homepage-new/schema-full-control.svg) * Choose the **behavior of the model**, what can be recommended, and what shall be boosted * **Express your custom filters** and boosters using our flexible ReQL language * Use our **AI ReQL Assistant** to create any rules with ease [Explore Full Control ->](https://www.recombee.com/features/full-control-with-scenario-settings) ### Easy Integration #### Follow your product vision by setting specific behavior for each box with recommendations. ![Easy Integration](https://www.recombee.com/img/homepage-new/schema-easy-integration.svg) * **Use SDK** in your programming language to call our **REST API** * **Integrate** using an HTML **low-code solution** * Take advantage of our [extensive documentation->](https://docs.recombee.com) * Get help from our **support team** whenever needed. [Explore Integration ->](https://www.recombee.com/features/integration) ### Cutting Edge AI & Research #### Benefit from unique features and inventions developed by our data scientists. ![Cutting Edge AI & Research](https://www.recombee.com/img/homepage-new/schema-cutting-edge-ai-and-research.svg) * Our data scientists frequently contribute to the most **prestigious conferences** in the field * We transform the **latest innovations** from our data scientists and the scientific community into **product solutions** * We ensure the highest quality of our recommendations through automatic **hyper-parameter optimization** [Explore AI & Research ->](https://www.recombee.com/research) ### Scalability #### Thanks to our horizontally scalable infrastructure, we recommend over a billion products, videos, articles, and other pieces of content every day. ![Scalability](https://www.recombee.com/img/homepage-new/schema-scalability.svg) * We can handle over **30k recommendations per second** * We support large catalogs with **tens of millions of items** * We ensure **low network latency** to your servers and customers by having multiple data centers strategically situated across the globe. [Explore Scalability ->](https://www.recombee.com/how-it-works/performance-at-scale) #### Follow your product vision by setting specific behavior for each box with recommendations. * Choose the **behavior of the model**, what can be recommended, and what shall be boosted * **Express your custom filters** and boosters using our flexible ReQL language * Use our **AI ReQL Assistant** to create any rules with ease [Explore Full Control ->](https://www.recombee.com/features/full-control-with-scenario-settings) ![Full Control](https://www.recombee.com/img/homepage-new/schema-full-control.svg) ![DAZN](https://www.recombee.com/img/testimonials/dazn-logo.svg) ![Christoph Haas](https://www.recombee.com/img/testimonials/dazn.png) ### Christoph Haas EVP Product & Platform Engineering at DAZN At DAZN, being the Global Home of Sports means delivering the right matches, highlights, and moments to audiences in 200+ markets - bringing fans even closer to the live game. That’s why we’ve teamed up with Recombee to personalize experiences at scale. Their tech enables us to connect each viewer on any device with the right game or clip in real time through flexible solution built for growth. This partnership sets the pace for a smarter, more connected global sports experience. [Case Studies ->](https://www.recombee.com/case-studies/) How It Works ## A Few Steps to Personalization ![How it works](https://www.recombee.com/img/homepage-new/schema-hiw.svg) ## From Developers to Developers The recommendation engine is provided for multiple programming languages by [RESTful API ->](https://docs.recombee.com/api) [API Clients ->](https://docs.recombee.com/api_clients) [No-Code Widgets ->](https://docs.recombee.com/no-code-widgets) [Widget SDK ->](https://docs.recombee.com/widget-sdks) npm install recombee-js-api-client [API Client](https://github.com/recombee/js-api-client) ``` const client = new recombee.ApiClient('database-id', dbPublicToken); // Send a view of item 'item_x' by user 'user_42' client.send(new recombee.AddDetailView('user_42', 'item_x')); // Get 5 recommended items for user 'user_42'. Recommend only items which haven't expired yet. const recommended = await client.send( new recombee.RecommendItemsToUser('user_42', 5, { filter: "'expires' > now()" }), ); ``` ### Read & Watch [Docs ->](https://docs.recombee.com) [YouTube ->](https://www.youtube.com/@recombee) ## Shape the Future Start Your Free 30-Day Trial Today. * Increase KPIs * Improve User Experience * Maximize Profits * Boost Customer Loyalty [Start Free ->](https://admin.recombee.com/sign-up) ![](https://www.recombee.com/img/homepage-new/m.png) --- # Real-Time Content-Based Recommendation Engine > Source: https://www.recombee.com/content-recommendations > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. ![](https://www.recombee.com/img/bg/polygon-green-xs.svg) ![](https://www.recombee.com/img/bg/polygon-green-2-md.svg) Content Recommendations # Improve User Engagement and Retention with Content Recommendations Personalize the user experience through real-time recommendation engine. Organize your content in the most relevant way for individual users based on their behavior and preferences. [Sign Up for Free](https://admin.recombee.com/sign-up) ## Your Personalized Content [Videos**Explore**](https://www.recombee.com/domains/video) [Articles**Explore**](https://www.recombee.com/domains/articles-news-media) [Podcasts**Explore**](https://www.recombee.com/domains/music-podcasts) [Music**Explore**](https://www.recombee.com/domains/music-podcasts) 01 ### Personalized for a User After receiving all the user’s data including his historical behavior, organize content in a unique, personalized way. ![Personalized for a User](https://www.recombee.com/img/products/personalized-for-a-user.png) 02 ### Recently Interacted With Display the content the user recently interacted with. ![Recently Interacted With](https://www.recombee.com/img/products/recently-interacted-with.png) 03 ### Similar Content Show content similar to one the user had already interest in, based on common content attributes. ![Similar Content](https://www.recombee.com/img/products/similar-content.png) 04 ### Liked This, Try That Recommend the content to users with similar tastes and habits based on their previous behavior. ![Liked This, Try That](https://www.recombee.com/img/products/liked-this-try-that.png) 05 ### Most Popular Content Show your most engagement and trending content. ![Most Popular Content](https://www.recombee.com/img/products/most-popular-content.png) 06 ### Latest Content Display the newest, recently added or released content. ![Latest Content](https://www.recombee.com/img/products/latest-content.png) 07 ### Personalized Search Enhance your online experiences with tailored search results. ![Personalized Search](https://www.recombee.com/img/products/personalized-search-content.png) ## Core Technology ### Adapting to your Data Personalization based on the Collaborative and Content-based filtering algorithms. ### Dynamically Retrained Models Real-time Personalization for your individual user at every point. ### Accelerated Integration Quick Integration through our well documented and easy to use APIs, SDKs. ### AI-powered A/B Testing To keep maximal KPIs at any time, AutoML AI is applied to optimize the algorithm ensembles. ### Advanced Business Rules Our solutions enable quick and easy addition of any Business Rules through boosters or filters. ### Real AI Inside Formation of Deep Neural Networks helps to predict the next action based on the historical behavior. ## How Recombee Works ![How Recombee Works Schema](https://www.recombee.com/img/products/how-it-works-content.svg) ## Success Stories [_![Haydn Strauss](https://www.recombee.com/img/customers/unfiltered-media-group.png)_Haydn StraussCOO at Unfiltered Media Group](#customer-unfiltered-media-group) [_![Meindert van der Meulen](https://www.recombee.com/img/customers/showmax.png)_Meindert van der MeulenHead of Strategy at Showmax](#customer-showmax) [_![Petr Kelin](https://www.recombee.com/img/customers/mafra.png)_Petr KelinManager at MAFRA, a.s.](#customer-mafra) [_![Jan Lajka](https://www.recombee.com/img/customers/ftv-prima.png)_Jan LajkaChief Data Officer at FTV Prima](#customer-ftv-prima) 50% Increase in Click Through Rate "Prior to Recombee, we used a general recommendation algorithm based on popularity and date published. Since moving our recommendation system to Recombee, we’ve seen a **50% increase in click-through across our 5 media brands** (millions of readers per month). Recombee was **easy to integrate, test, and deploy within just a couple of hours.**" _![Haydn Strauss](https://www.recombee.com/img/customers/unfiltered-media-group.png)_Haydn StraussCOO at Unfiltered Media Group ![Unfiltered Media Group](https://www.recombee.com/img/logos/unfiltered-media-group.png) [Read Case Study](https://www.recombee.com/case-studies/unfiltered-media-group) Higher Engagement "Recombee is capable of **scaling the service** and keeps pace with our **rapid growth.** Constant innovation and proactive development of new features makes our collaboration smooth and pleasant." _![Meindert van der Meulen](https://www.recombee.com/img/customers/showmax.png)_Meindert van der MeulenHead of Strategy at Showmax ![Showmax](https://www.recombee.com/img/logos/showmax.png) [Read Case Study](https://www.recombee.com/case-studies/showmax) 40% higher CTR of suggested articles "We conducted A/B testings of multiple recommendation engines to find the best content personalization solution. Out of all solutions, only **Recombee outperformed our internal read-next recommendations** of news articles. After long-lasting A/B testing, Recombee achieved **40% higher CTR of suggested articles,** which ultimately led to deployment of the solution to most of our news sites (iDNES, Lidovky, Expres)." _![Petr Kelin](https://www.recombee.com/img/customers/mafra.png)_Petr KelinManager at MAFRA, a.s. ![MAFRA, a.s.](https://www.recombee.com/img/logos/mafra.png) [Read Case Study](https://www.recombee.com/case-studies/mafra) 34% Increase in Video views on VOD platform "Recombee allows us to modify how we want the recommendations to behave across our platforms in very specific use cases. The implementation of Recombee on the VOD platform prima+ helped us to increase video views by 34% and ad views by 73%, resulting in a significant rise in advertising revenue. Another major success has been the growth of recirculation, which is our top priority on online magazine platforms. We are currently discussing extending their recommendations also to our emails. Highly valued partnership!" _![Jan Lajka](https://www.recombee.com/img/customers/ftv-prima.png)_Jan LajkaChief Data Officer at FTV Prima ![FTV Prima](https://www.recombee.com/img/logos/ftv-prima.png) [Read Case Study](https://www.recombee.com/case-studies/ftv-prima) ## Our Customers ![The Telegraph](https://www.recombee.com/img/logos/the-telegraph.svg)![9GAG](https://www.recombee.com/img/logos/9gag.svg)![Audiomack](https://www.recombee.com/img/logos/audiomack.svg)![Unfiltered Media Group](https://www.recombee.com/img/logos/unfiltered-media-group.png)![Stingray](https://www.recombee.com/img/logos/stingray.svg)![mafra](https://www.recombee.com/img/logos/mafra.png)![prima](https://www.recombee.com/img/logos/ftv-prima.png)![Pathé Thuis](https://www.recombee.com/img/logos/pathe-thuis.svg)![Showmax](https://www.recombee.com/img/logos/showmax.svg) and 1000+ other sites and apps. --- # Pricing > Source: https://www.recombee.com/pricing > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. Pricing # AI Personalization Plans That Scale With You **Start Unlimited 30-Days Trial.** After the trial expires, choose to upgrade or continue with a Free Plan [PlansFind the right plan for your traffic and usage.Explore Plans ↓](#pricing-plans) [Platinum PlanUnlock a fully tailored personalization platform built around your business.Explore Platinum ↓](#platinum) **Plan Overview** Choose plan FreeStandardProPremiumPlatinum [Calculate Price](#calculate) Free Forever $0 [Start Free](https://admin.recombee.com/sign-up) Standard Starting at $99 /mo [Start Free](https://admin.recombee.com/sign-up) Pro Starting at $1,699 /mo [Start Free](https://admin.recombee.com/sign-up) Premium Starting at $4,499 /mo Usage Interactions < 100,000/mo 100,000/mo 5,000,000/mo ≥ 20,000,000/mo Recom. Requests < 100,000/mo 100,000/mo 5,000,000/mo ≥ 20,000,000/mo Active Users < 20,000/mo 20,000/mo 1,000,000/mo ≥ 4,000,000/mo Catalog Items < 20,000 20,000 1,000,000 ≥ 4,000,000 Extra Traffic Unit $0.70 / $0.25 $0.20 Custom Features Content Recommendations Models Product Recommendations Models User Recommendations Models Personalized Search Models Real-Time Analytics Instantly Performing Solution Ready to use filters and boosters ReQL with AI Code Assistant Constraints Item Segmentations No-Code Widget Search Widget SSO Semantic Search Add-on Semantic Segmentations Add-on Support Customer Support Standard L1 Advanced L1 Premium L1 Mail Availability Guarantees Account Manager Video Calls Slack Channel Advanced Integration Support Data Science Support ## Platinum Built for visionary platforms. Engineered to scale. Leverage Recombee’s advanced personalization technology and expertise to deliver tailored experiences, support complex use cases, and drive measurable impact at scale. ### Innovative Solutions Crafted With Your Unique Vision in Mind * R&D-driven Innovations * Advanced custom use cases & complex recommendation scenarios * High-volume scaling up to billions of monthly recommendation requests & interactions * Optimized performance during traffic spikes and high-demand periods * Comprehensive support and expert guidance Trusted by Industry Leaders ![DAZN](https://www.recombee.com/img/customers-domains/dazn.svg)![The Telegraph](https://www.recombee.com/img/customers-domains/the-telegraph.svg)![Audiomack](https://www.recombee.com/img/customers-domains/audiomack.svg)![BeatStars](https://www.recombee.com/img/customers-domains/beatstars.svg)![FTV Prima](https://www.recombee.com/img/customers-domains/ftv-prima.svg) ## Calculate Price Estimate The price is calculated on the maximum value of the plan limit. Interactions/mo 0 /0 Recom. Requests/mo 0 /0 Active Users/mo 0 /0 Catalog Items 0 /0 Pro Plan $1,699.00/mo Custom price [Start Free](https://admin.recombee.com/sign-up) Interactions 0 Recom. Requests 0 Active Users 0 Catalog Items 0 Extra Traffic Unit 0 [See Plan Features and Support](#pricing-plans) --- # Recommendation Engine for Real-time Personalization > Source: https://www.recombee.com/product > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. Product # Recommendation System Providing Recommendations in Real-Time Why waste time and money on the development of your own recommender system, if you can use the most advanced engine tailored by data scientists. ![](https://www.recombee.com/img/product/main.png) ![](https://www.recombee.com/img/product/raas.png) ## Real-Time Recommendation as a Service Our API/SDK returns the recommendations immediately after user’s action at your website or in your app in less than 200 ms [**More About Technology**](https://www.recombee.com/technology) ## Real-Time Analytics & Insights Insights offer various predefined and fully customizable reports to track recommended items and how users interact with these recommendations. By gaining a deep understanding of user behavior, you can fine-tune your recommendation engine to match your product vision and maximize your KPIs. [**Explore Insights**](https://www.recombee.com/features/real-time-analytics-insights) [**Documentation**](https://docs.recombee.com/insights) ![](https://www.recombee.com/img/product/analytics-insights.png) ![](https://www.recombee.com/img/product/ui.png) ## User Interface You can check the key performance indicators in near real-time using our intuitive graphical user interface (GUI). [**Explore Admin UI**](https://www.recombee.com/admin-ui) ## Scalable Solution We can deliver the service no matter how large is the traffic at your website, being it a million or billion monthly page views, without any compromises to the quality. What is more, our system is capable of delivering more than 30,000 recommendations per second. [**More About Technology**](https://www.recombee.com/technology) ![](https://www.recombee.com/img/product/scaleability.png) ![](https://www.recombee.com/img/product/domain.png) ## Domain Independence Our system is successfully used to recommend movies, news, cultural events, books, songs, job advertisements, or products in E-Commerce. It is easily adaptable to any other domain. [**Where to Use**](https://www.recombee.com/where-to-use) [**Success Stories**](https://www.recombee.com/case-studies) ## Easy to Customize You can customize our recommender for your business using rules in our simple and highly innovative filtering/boosting language (ReQL). Filters and boosters can be managed independently for each recommendation request. [**More About ReQL**](https://docs.recombee.com/reql) ![](https://www.recombee.com/img/product/reql.png) ![](https://www.recombee.com/img/product/item-segmentations.png) ## Item Segmentations Native support for products and content hierarchy. Get personalized recommendations of the top categories, brands, genres, tags, artists, or any custom group of items that can be defined based on the properties of the items in your catalog. For example, you can specify individual homepage rows and ask Recombee for the best order of the rows for a particular user based on their interaction history. [**More About Item Segmentations**](https://docs.recombee.com/segmentations) ## Research and Improvements Sophisticated algorithms form the core of our recommendation engine. Algorithms are continuously managed and improved by the Artificial Intelligence. Our team is comprised of data scientists with 15+ years of AI and Machine Learning experience. We also conduct progressive machine-learning research. [**More About Technology**](https://www.recombee.com/technology) ![](https://www.recombee.com/img/product/research-improvements.png) ## Get the Most out of Your Website with Recombee Increase Conversion Rate Higher Profits Increase Loyalty Increase Retention Improve User Experience Increase Lifetime Value Reduce Churn and Confusion AI Driven A/B Testing ## Quick and Easy Integration into Your Environment The recommendation engine is provided via a RESTful API, No-Code Widgets and SDKs for multiple programming languages. [No-Code WidgetsThe easiest way how to get recommendations into your site.](https://docs.recombee.com/no-code-widgets) [DocumentationExplore our documentation and guides on how to start personalizing fast.](https://docs.recombee.com/) ![](https://www.recombee.com/img/codes/icon-javascript.svg)JavaScript ![](https://www.recombee.com/img/codes/icon-python.svg)Python ![](https://www.recombee.com/img/codes/icon-ruby.svg)Ruby ![](https://www.recombee.com/img/codes/icon-java.svg)Java ![](https://www.recombee.com/img/codes/icon-nodejs.svg)Node.js ![](https://www.recombee.com/img/codes/icon-android.svg)Android ![](https://www.recombee.com/img/codes/icon-ios.svg)iOS ![](https://www.recombee.com/img/codes/icon-php.svg)PHP ![](https://www.recombee.com/img/codes/icon-net.svg).NET ![](https://www.recombee.com/img/codes/icon-go.svg)Go ![](https://www.recombee.com/img/codes/icon-rest.svg)REST * [_![](https://www.recombee.com/img/codes/icon-javascript.svg)_JavaScript](#code-javascript) * [_![](https://www.recombee.com/img/codes/icon-python.svg)_Python](#code-python) * [_![](https://www.recombee.com/img/codes/icon-ruby.svg)_Ruby](#code-ruby) * [_![](https://www.recombee.com/img/codes/icon-java.svg)_Java](#code-java) * [_![](https://www.recombee.com/img/codes/icon-nodejs.svg)_Node.js](#code-nodejs) * [_![](https://www.recombee.com/img/codes/icon-android.svg)_Android](#code-android) * [_![](https://www.recombee.com/img/codes/icon-ios.svg)_iOS](#code-ios) * [_![](https://www.recombee.com/img/codes/icon-php.svg)_PHP](#code-php) * [_![](https://www.recombee.com/img/codes/icon-net.svg)_.NET](#code-net) * [_![](https://www.recombee.com/img/codes/icon-go.svg)_Go](#code-go) * [_![](https://www.recombee.com/img/codes/icon-rest.svg)_REST](#code-rest) ```javascript const client = new recombee.ApiClient('database-id', dbPublicToken); // Send a view of item 'item_x' by user 'user_42' client.send(new recombee.AddDetailView('user_42', 'item_x')); // Get 5 recommended items for user 'user_42'. Recommend only items which haven't expired yet. const recommended = await client.send( new recombee.RecommendItemsToUser('user_42', 5, {filter: "'expires' > now()"}) ); ``` [See Recombee API Client for JavaScript on Github](https://github.com/recombee/js-api-client) ```ruby client = RecombeeClient.new('database-id', secret_token) # Send a view of item 'item_x' by user 'user_42'. Create user and/or item if it doesn't exist yet. client.send(AddDetailView.new('user_42', 'item_x', 'cascadeCreate' => true)) # Get 5 recommended items for user 'user_42'. Recommend only items which haven't expired yet. recommended = client.send(RecommendItemsToUser.new('user_42', 5, 'filter' => "'expires' > now()")) ``` [See Recombee API Client for Ruby on Github](https://github.com/Recombee/ruby-api-client) ```java RecombeeClient client = new RecombeeClient("database-id", secretToken); // Send a view of item "item_x" by user "user_42". Create user and/or item if it doesn't exist yet. client.send(new AddDetailView("user_42", "item_x").setCascadeCreate(true)); // Get 5 recommended items for user "user_42". Recommend only items which haven't expired yet. RecommendationResponse recommended = client.send( new RecommendItemsToUser("user_42", 5).setFilter("'expires' > now()") ); ``` [See Recombee API Client for Java on Github](https://github.com/recombee/java-api-client) ```python client = RecombeeClient('database-id', secret_token) # Send a view of item 'item_x' by user 'user_42'. Create user and/or item if it doesn't exist yet. client.send(AddDetailView('user_42', 'item_x', cascade_create=True)) # Get 5 recommended items for user 'user_42'. Recommend only items which haven't expired yet. recommended = client.send(RecommendItemsToUser('user_42', 5, filter="'expires' > now()")) ``` [See Recombee API Client for Python on Github](https://github.com/recombee/python-api-client) ```javascript const client = new recombee.ApiClient('database-id', secretToken); // Send a view of item 'item_x' by user 'user_42'. Create user and/or item if it doesn't exist yet. client.send(new rqs.AddDetailView('user_42', 'item_x', {cascadeCreate: true}), callback); // Get 5 recommended items for user 'user_42'. Recommend only items which haven't expired yet. const recommended = await client.send( new rqs.RecommendItemsToUser('user_42', 5, {filter: "'expires' > now()"}) ); // Supports both Promises and callbacks ``` [See Recombee API Client for Node.js on Github](https://github.com/Recombee/node-api-client) ```kotlin val client = RecombeeClient(databaseId = "yourDatabaseId", publicToken = dbPublicToken,) // Send a view of item 'item_x' by user 'user_42' client.send(AddDetailView("user_42", "item_x")); // Get 5 recommended items for user 'user_42'. Recommend only items which haven't expired yet. val result = client.sendAsync(RecommendItemsToUser("user_42", count = 5, filter = "'expires' > now()")) result.onSuccess { response: RecommendationResponse -> // Show recommendations } ``` [See Recombee API Client for Android on Github](https://github.com/Recombee/kotlin-api-client) ```swift let client = RecombeeClient(databaseId: "yourDatabaseId", publicToken: dbPublicToken, region: .euWest) try await client.send(AddDetailView(userId: "user_42", itemId: "item_x")) let result = try await client.send(RecommendItemsToUser(userId: "user_42", count: 5, filter: "'expires' > now()")) result.recomms.forEach { print($0.id) } ``` [See Recombee API Client for iOS on Github](https://github.com/Recombee/swift-api-client) ```php $client = new Client('database-id', $secret_token); // Send a view of item 'item_x' by user 'user_42'. Create user and/or item if it doesn't exist yet. $client->send(new Reqs\AddDetailView('user_42', 'item_x', ['cascadeCreate' => true])); // Get 5 recommended items for user 'user_42'. Recommend only items which haven't expired yet. $recommended = $client->send(new Reqs\RecommendItemsToUser('user_42', 5, ['filter' => "'expires' > now()"])); ``` [See Recombee API Client for PHP on Github](https://github.com/Recombee/php-api-client) ```csharp var client = new RecombeeClient("database_id", secretToken); // Send a view of item "item_x" by user "user_42". Create user and/or item if it doesn't exist yet. client.Send(new AddDetailView("user_42", "item_x", cascadeCreate: true)); // Get 5 recommended items for user 'user_42'. Recommend only items which haven't expired yet. var recommended = client.Send(new RecommendItemsToUser("user_42", 5, filter: "'expires' > now()")); ``` [See Recombee API Client for .NET on Github](https://github.com/Recombee/net-api-client) ```go client, _ := recombee.NewRecombeeClient("databaseId", secretToken) // Send a view of item 'item_x' by user 'user_42', creating user/item if they don't exist client.NewAddDetailView("user_42", "item_x").SetCascadeCreate(true).Send() // Get 5 recommended items for user 'user_42', filtering out expired items recommendRes, _ := client.NewRecommendItemsToUser("user_42", 5).SetFilter("'expires' > now()").Send() // Print recommended item IDs for _, rec := range recommendRes.Recomms { fmt.Println(rec.Id) } ``` [See Recombee API Client for .NET on Github](https://github.com/Recombee/net-api-client) ```ruby # Send a view of item 'item_x' by user 'user_id'. Create user and/or item if it doesn't exist yet. POST https://rapi.recombee.com/database_id/detailviews/ Data: {'userId': 'user_42', 'itemId': 'item_x', 'cascadeCreate': true} # Get 5 recommended items for user 'user_42'. # Recommend only items which haven't expired yet (filter: 'expires' > now()). GET https://rapi.recombee.com/database_id/recomms/users/user_42/items/?count=5&filter=%27expires%27%3Enow() ``` [See more API documentation](https://docs.recombee.com/getting_started) --- # Product Recommendation Engine Tailored by Data Scientists > Source: https://www.recombee.com/product-recommendations > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. ![](https://www.recombee.com/img/bg/polygon-green-xs.svg) ![](https://www.recombee.com/img/bg/polygon-green-2-md.svg) Product Recommendations # Maximize your Revenue and Customer Satisfaction with Product Recommendations Personalize the product offer for every individual customer. Utilize real-time personalization for a better user experience using our content and collaborative filtering based recommender system. [Sign Up for Free](https://admin.recombee.com/sign-up) ## Your Personalized Product Offer [E-commerce**Explore**](https://www.recombee.com/domains/e-commerce) [P2P Marketplaces**Explore**](https://www.recombee.com/domains/p2p-marketplaces) 01 ### You May Also Like / Others Also Bought Cross-sell / Up-Sell by showing products mainly based on collaborative filtering and onsite user behavior. ![You May Also Like / Others Also Bought](https://www.recombee.com/img/products/you-may-also-like-others-also-bought.png) 02 ### Frequently Bought Together Show your users complementary products based on their buying behavior. ![Frequently Bought Together](https://www.recombee.com/img/products/frequently-bought-together.png) 03 ### Recommended for You Show your users new product offers based on individual user’s preferences and product similarities. ![Recommended for You](https://www.recombee.com/img/products/recommended-for-you.png) 04 ### Personalized Search Display the most relevant product offers for each individual user to satisfy its specific needs and save time. ![Personalized Search](https://www.recombee.com/img/products/personalized-search.png) 05 ### Local Deals Trending Near You Show personalised product offer to your user by using their geo-location. ![Local Deals Trending Near You](https://www.recombee.com/img/products/local-deals-trending-near-you.png) 06 ### Popular Right Now / Trending Now Show the most trending products based on your user purchases and general trend. ![Popular Right Now / Trending Now](https://www.recombee.com/img/products/popular-right-now-trending-now.png) 07 ### Recently Interacted With Display the products the user recently interacted with. ![Recently Interacted With](https://www.recombee.com/img/products/recently-interacted-with-product.png) ## Core Technology ### Adapting to your Data Personalization based on the Collaborative and Content-based filtering algorithms. ### Dynamically Retrained Models Real-time Personalization for your individual user at every point. ### Accelerated Integration Quick Integration through our well documented and easy to use APIs, SDKs. ### AI-powered A/B Testing To keep maximal KPIs at any time, AutoML AI is applied to optimize the algorithm ensembles. ### Advanced Business Rules Our solutions enable quick and easy addition of any Business Rules through boosters or filters. ### Real AI Inside Formation of Deep Neural Networks helps to predict the next action based on the historical behavior. ## How Recombee Works ![How Recombee Works Schema](https://www.recombee.com/img/products/how-it-works.svg) ## Success Story [_![Daniel Uhm](https://www.recombee.com/img/customers/slickdeals.png)_Daniel UhmProduct Manager at Slickdeals](#customer-slickdeals) [_![Dennis Ostner](https://www.recombee.com/img/customers/kunzmann.png)_Dennis OstnerHead of E-commerce at Robert Kunzmann GmbH & Co.](#customer-kunzmann) [_![Vincent van Leeuwen](https://www.recombee.com/img/customers/reliving.png)_Vincent van LeeuwenCo-Founder & CTO/CPO at Reliving](#customer-reliving) 30% Increase in CTR to Affiliate Links "Placing recommendations on our homepage was a huge success: **70%+ higher product detail page views and 30%+ higher clickthroughs.** The Recombee team is a great partner in helping solve our unique use cases, and we look forward to continue working with them." _![Daniel Uhm](https://www.recombee.com/img/customers/slickdeals.png)_Daniel UhmProduct Manager at Slickdeals ![Slickdeals](https://www.recombee.com/img/logos/slickdeals.png) [Read Case Study](https://www.recombee.com/case-studies/slickdeals) 14% Increase in conversion rate "Recombee is an amazing recommendation engine which we use for personalizing different parts on our website, including homepage, product detail page, and search. With their solution, we managed to increase our conversion rate by 14% and shopping cart volume by 8%. A great partnership and looking forward to improving our customer journey even more." _![Dennis Ostner](https://www.recombee.com/img/customers/kunzmann.png)_Dennis OstnerHead of E-commerce at Robert Kunzmann GmbH & Co. ![Robert Kunzmann GmbH & Co.](https://www.recombee.com/img/logos/autohaus-kunzmann.png) [Read Case Study](https://www.recombee.com/case-studies/autohaus-kunzmann) 37% Increase in Place Bid "Instead of the lengthy and costly process of building an in-house personalization solution, we seamlessly implemented Recombee's AI-powered recommender engine to improve our services. Applying their tailored scenarios and boosters, we have registered steady improvements ever since, with a **6% increase in 'Add to Cart' and a 37% increase in 'Place Bid'.** Thanks to their easy and intuitive integration, Recombee was an obvious choice from the range of personalization solutions." _![Vincent van Leeuwen](https://www.recombee.com/img/customers/reliving.png)_Vincent van LeeuwenCo-Founder & CTO/CPO at Reliving ![Reliving](https://www.recombee.com/img/logos/reliving.png) [Read Case Study](https://www.recombee.com/case-studies/reliving) ## Our Customers ![Slickdeals](https://www.recombee.com/img/logos/slickdeals.svg)![Segundamano](https://www.recombee.com/img/logos/segundamano.png)![itison](https://www.recombee.com/img/logos/itison.png)![Design Group](https://www.recombee.com/img/logos/design-group.svg)![Reliving](https://www.recombee.com/img/logos/reliving.png)![konga.com](https://www.recombee.com/img/logos/kongacom.png) and 1000+ other sites and apps. --- # What Makes Our Recommendation Technology Unique > Source: https://www.recombee.com/technology > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. Recombee Technology # Recommendation Engine Built on the Most Innovative Algorithms Hi-tech recommender at your service ![recombee-technology](https://www.recombee.com/img/technology/recombee-technology.svg) Unlike traditional rule-based personalization systems, Recombee’s AI driven solution **reflects real-time changes and complexity of user behavior online** and thus enables you to **personalize 1:1** improving your user experience and KPI’s. Our solution analyzes user interactions and behavior online as well as product attributes and generates recommendations which are more likely to spark the interest of the customer. At Recombee, we conduct research in collaboration with academia, which keeps us at the leading edge of innovation. We utilize **deep-learning** and **collaborative filtering**, as well as **content-based algorithms** (such as **image** and text processing algorithms) to ensure the most accurate content for all visitors. What is more, our solution is **real-time**, meaning that our models are **adapting after every interaction of the users.** ![Collaborative Filtering](https://www.recombee.com/img/technology/collaborative-filtering.svg) ## Collaborative Filtering Models based on behavioral patterns First type of the models Recombee adds to its model ensembles are Collaborative Filtering models, which are built from collected user-item interactions, such as detail-views or purchases. By analyzing behavioral patterns across the whole userbase, the recommendations are based on extracting interactions similarities between users, items, or both. “Similar users also liked” or “others also purchased” are both examples of CF-based recommendations. Recombee uses following CF models: Matrix Factorization, Nearest Neighbor methods, and Association Rules. ![Content-Based](https://www.recombee.com/img/technology/content-based.svg) ## Content-Based Models based on attributes of items and users We also use Content-Based models, which estimate similarities between items or users by analyzing the provided property values. For example, two items can be considered similar by having similar categorization, name, text descriptions, etc. Various models are used to process different type of attribute data. These models are especially useful in cold - start situations when there’s not enough interaction data yet(brand new item or user). ![Deep Learning](https://www.recombee.com/img/technology/deep-learning.svg) ## Deep Learning Models combining interaction and attribute data together Following the cutting-edge research in the field, Recombee offers models based on neural networks and deep autoencoders to build the recommendations. Such models are able to consider at once all the data provided in the given context. The models build AI-based understanding of concepts hidden in the data. ![Specialized Models](https://www.recombee.com/img/technology/specialized-models.svg) ## Specialized Models Models reflecting specific business-cases and product needs Diversification models (recommending variety of different items), popularity-based models (long-term or trending), reminder models or periodicity-based models (based on repeating behavior in user-item interactions), are also part of Recombee. These come from vast amount of experience that Recombee team gained during years in business, applying the systems to hundreds of different use-cases. ![Image Processing](https://www.recombee.com/img/technology/image-processing.svg) ## Image Processing Models based on analyzing images and visual similarity Recombee can process product images to extract similarities based on visual style. This allows e.g. recommending items which are visually similar to those liked by a user in the past. Advanced models based on top of convolutional neural networks are used for that. Multiple images (such as photos taken from different angles) can be provided to further improve the performance. ![AI-Based Model Optimization](https://www.recombee.com/img/technology/ai-based-model-optimization.svg) ## AI-Based Model Optimization Automated searching for proper hyperparametrization The models in production need to continuously adapt to the changes in the environment such as different seasons or holidays. When put to production, the models adapt based on collected feedback. Knowing whether the recommendations really led to user actions allows Recombee to tune both the structure and the hyperparameters of the deployed model ensemble. ![Large Language Models](https://www.recombee.com/img/technology/llm.svg) ## Large Language Models Models based on semantic understanding Another type of model included in Recombee's ensembles are Large Language Models (LLMs), which excel in understanding language and semantics. These models can capture subtle patterns and relationships in text, providing more accurate and context-aware recommendations. This capability is particularly valuable in applications like Semantic Search, where LLMs enhance the relevance of search results by understanding both user intent and the content of items. Recombee leverages various LLM approaches to further enrich its recommendation capabilities. ## Technology Stack [![Puppet](https://www.recombee.com/img/techstack/puppet.svg)](https://www.puppet.com/) [![Kubernetes](https://www.recombee.com/img/techstack/kubernetes.svg)](https://kubernetes.io/) [![Docker](https://www.recombee.com/img/techstack/docker.svg)](https://www.docker.com/) [![Aerospike](https://www.recombee.com/img/techstack/aerospike.svg)](https://www.aerospike.com/) [![Grafana](https://www.recombee.com/img/techstack/grafana.svg)](https://grafana.com/) [![Elasticsearch](https://www.recombee.com/img/techstack/elasticsearch.svg)](https://www.elastic.co/) [![GitLab](https://www.recombee.com/img/techstack/gitlab.svg)](https://about.gitlab.com/) [![Sentry](https://www.recombee.com/img/techstack/sentry.svg)](https://sentry.io/) [![Icinga](https://www.recombee.com/img/techstack/icinga.svg)](https://icinga.com/) [![PostgreSQL](https://www.recombee.com/img/techstack/postgresql.svg)](https://www.postgresql.org/) [![Apache Kafka](https://www.recombee.com/img/techstack/apachekafka.svg)](https://kafka.apache.org/) [![Vault](https://www.recombee.com/img/techstack/vault.svg)](https://www.vaultproject.io/) [![TensorFlow](https://www.recombee.com/img/techstack/tensorflow.svg)](https://www.tensorflow.org/) [![Prometheus](https://www.recombee.com/img/techstack/prometheus.svg)](https://prometheus.io/) [![Flux](https://www.recombee.com/img/techstack/flux.svg)](https://fluxcd.io/) [![Cilium](https://www.recombee.com/img/techstack/cilium.svg)](https://cilium.io/) [![Jira](https://www.recombee.com/img/techstack/jira.svg)](https://www.atlassian.com/software/jira) [![NVIDIA CUDA](https://www.recombee.com/img/techstack/nvidiacuda.png)](https://developer.nvidia.com/cuda-zone) [![MinIO](https://www.recombee.com/img/techstack/minio.svg)](https://min.io/) [![Thanos](https://www.recombee.com/img/techstack/thanos.svg)](https://thanos.io/) [![nginx](https://www.recombee.com/img/techstack/nginx.svg)](https://www.nginx.com/) [![ClickHouse](https://www.recombee.com/img/techstack/clickhouse.svg)](https://clickhouse.com/) [![Timescale](https://www.recombee.com/img/techstack/timescale.svg)](https://www.timescale.com/) [![Keras](https://www.recombee.com/img/techstack/keras.png)](http://keras.io) --- # Examples of Personalized Recommendations > Source: https://www.recombee.com/where-to-use > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. ![](https://www.recombee.com/img/bg/polygon-green-xs.svg) ![](https://www.recombee.com/img/bg/polygon-green-2-md.svg) Where to use # Chosen Examples of Recombee Recommendation Engine Recombee recommender engine can be utilized in a variety of areas including movies, music, news, books or e-commerce products. Our data scientists help you to set up the right scenarios to fit your domain specific use cases. ## Homepage ### Provide users with **personalized experience** from the very beginning. * Recommendations tailored to each specific user * Also applicable to anonymous users * Improve user experience and satisfaction * Decrease the time spent searching for relevant content ![](https://www.recombee.com/img/where-to-use/homepage.png) ## Product Detail Page ### **Increase basket size and/or conversion rates** with personalized product suggestions. * Related product/item recommendations applicable not only to e-commerce but also to gaming, e-learning, job boards, etc. * Show your customers products they might like the most * Can be used cross-sell, up-sell * Different editable scenarios ![](https://www.recombee.com/img/where-to-use/scr.png) ## Read Next/Watch Next ### **Increase time spent and advertising revenue** with recommendations on relevant content. * CTR increase * Natural Language Processing based recommendations * Show your users newly released content based on each user’s preference * Satisfy and amaze your viewers and readers without closing them into a bubble ![](https://www.recombee.com/img/where-to-use/next.png) ## Email/Push Notifications ### **Increase the click through rate by 30% (CTR)** with personalized e-mail marketing. * Bring users back to your website * Boost order value and profit margin * Reduce Purchasing Cycle ![](https://www.recombee.com/img/where-to-use/newsletter.png) ## Full-Text Search ### Improve **relevance of search results and save customers’ time.** Apply boosters of certain products in the search ranking to maximize revenue. * Combination of Search Engine and Machine Learning * Results based on search query full text matching and user’s interaction data and metadata * Narrow searches to specific items to save time * Higher purchase rate and satisfaction * Aids in the selection process for undecided customer * Support for typo corrections * Support for variety of languages, including non-European languages [How to Apply Personalized Search](https://docs.recombee.com/getting_started#getting-started-search) ![](https://www.recombee.com/img/where-to-use/personalized-search.png) ## Domains Video [**Read More**](https://www.recombee.com/domains/video) E-commerce [**Read More**](https://www.recombee.com/domains/e-commerce) Music, Podcasts [**Read More**](https://www.recombee.com/domains/music-podcasts) Articles, News, Media [**Read More**](https://www.recombee.com/domains/articles-news-media) Real Estate [**Read More**](https://www.recombee.com/domains/real-estate) P2P Marketplaces [**Read More**](https://www.recombee.com/domains/p2p-marketplaces) Deal Aggregators [**Read More**](https://www.recombee.com/domains/deal-aggregators) Job boards, HR, Networking [**Read More**](https://www.recombee.com/domains/jobs-boards-hr-networking) Travel, Trips [**Read More**](https://www.recombee.com/domains/travel-trips) Apps Cultural events E-learning Discussion forums Platforms --- # Configurable Recommender with Real-Time Analytics > Source: https://www.recombee.com/admin-ui > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. Admin UI # Recommendation Software With Built-In Analytics Explore performance metrics and configure recommendations to reflect your personalization needs. Use simple and user-friendly interface designed for all your team members. ![](https://www.recombee.com/img/admin-ui/main.png) ## KPI Dashboard ### Leverage Advanced Insights With Customizable Dashboard Monitor performance of recommendations in time. Pin chosen KPI displays to create your own dashboard. * Recommendation funnels * Number of interactions * Metrics comparison * Per scenario metrics ![](https://www.recombee.com/img/admin-ui/02.png) ## Scenarios & Business Rules ### Manage Your Recommendations Tailor your recommendations using specific models and business rules. Use your own specific filters and boosters or the ones predefined by our data scientists. * **Scenario** customization. [Read more](https://docs.recombee.com/scenarios) * **Business rules** \- Filters and Boosters. [Read more](https://docs.recombee.com/scenarios#business-rules) * **Recombee Library** contains predefined Business rules ![](https://www.recombee.com/img/admin-ui/03.png) ## Catalog and Interactions ### Explore Uploaded Data Check quality of your data sent to Recombee. Enjoy smooth integration and precise management of database items/users interactions. * **Smooth and flawless integration** for developers * **Insights** for marketing department’s desicion making ![](https://www.recombee.com/img/admin-ui/04.png) ## And More... ![](https://www.recombee.com/img/admin-ui/product-feed.png) ### Product Feed Use Google Merchant, Custom XML/JSON/CSV, Heureka product feed or Atom/RSS to import items into item catalog. The feed is periodically crawled and item catalog updated. ![](https://www.recombee.com/img/admin-ui/collaboration.png) ### Collaboration Choose the databases or the organizations you want to share with the Marketing, Finance, Development teams or collaborators. --- # Features > Source: https://www.recombee.com/features > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # Features Explore features built for full control over recommendations and search. [![Recommendations & Search](https://www.recombee.com/img/features/recommendations-search.png)](https://www.recombee.com/features/recommendations-search) ## [Recommendations & Search](https://www.recombee.com/features/recommendations-search) Recombee’s real-time recommendation engine adapts instantly to user interactions and content updates, delivering personalized experiences as they happen. [Real-Time Recommendations](https://www.recombee.com/features/recommendations-search#real-time-recommendations) [Segmentations](https://www.recombee.com/features/recommendations-search#segmentations) [Fully Personalized Homepage](https://www.recombee.com/features/recommendations-search#fully-personalized-homepage) [Personalized Infinite Feed](https://www.recombee.com/features/recommendations-search#personalized-infinite-feed) [Personalized Semantic Search](https://www.recombee.com/features/recommendations-search#personalized-semantic-search) [Semantic Segmentations](https://www.recombee.com/features/recommendations-search#semantic-segmentations) [Cutting-Edge AI & Research](https://www.recombee.com/features/recommendations-search#cutting-edge-ai-research) [![Real-Time Analytics & Insights](https://www.recombee.com/img/features/real-time-analytics-insights.png)](https://www.recombee.com/features/real-time-analytics-insights) ## [Real-Time Analytics & Insights](https://www.recombee.com/features/real-time-analytics-insights) Track engagement with detailed, customizable reports available in our Admin UI, offering insights into how users interact with recommendations and your platform. [Real-Time Data](https://www.recombee.com/features/real-time-analytics-insights#real-time-data) [Library of Insights](https://www.recombee.com/features/real-time-analytics-insights#library-of-insights) [Custom Insights](https://www.recombee.com/features/real-time-analytics-insights#custom-insights) [A/B Testing](https://www.recombee.com/features/real-time-analytics-insights#ab-testing) [![Full Control with Scenario Settings](https://www.recombee.com/img/features/full-control-with-scenario-settings.png)](https://www.recombee.com/features/full-control-with-scenario-settings) ## [Full Control with Scenario Settings](https://www.recombee.com/features/full-control-with-scenario-settings) Customize recommendations with Recombee’s flexible Scenario Settings to perfectly align content with your product strategy. [Scenario](https://www.recombee.com/features/full-control-with-scenario-settings#scenario) [Logic](https://www.recombee.com/features/full-control-with-scenario-settings#logic) [Filters](https://www.recombee.com/features/full-control-with-scenario-settings#filters) [Boosters](https://www.recombee.com/features/full-control-with-scenario-settings#boosters) [Constraints](https://www.recombee.com/features/full-control-with-scenario-settings#constraints) [A/B Testing](https://www.recombee.com/features/full-control-with-scenario-settings#ab-testing) [![Business Rules](https://www.recombee.com/img/features/business-rules.png)](https://www.recombee.com/features/business-rules) ## [Business Rules](https://www.recombee.com/features/business-rules) Craft custom rules to highlight specific categories, brands, editor-picked content, local deals, and more—delivering diverse recommendation boxes and email campaigns. [ReQL](https://www.recombee.com/features/business-rules#reql) [AI Assistant](https://www.recombee.com/features/business-rules#ai-assistant) [![Integration](https://www.recombee.com/img/features/integration.png)](https://www.recombee.com/features/integration) ## [Integration](https://www.recombee.com/features/integration) Integrate easily with a well-structured REST API and SDKs, complemented by catalog feed processing, No-Code widgets, and ready-made integrations with key platforms like Segment. [Easy Integration](https://www.recombee.com/features/integration#easy-integration) [API SDKs](https://www.recombee.com/features/integration#api-sdks) [Widget SDKs](https://www.recombee.com/features/integration#widget-sdks) [No-Code Widgets](https://www.recombee.com/features/integration#no-code-widgets) [Catalog Feeds](https://www.recombee.com/features/integration#catalog-feeds) [Support](https://www.recombee.com/features/integration#support) [![Scalability](https://www.recombee.com/img/features/scalability.png)](https://www.recombee.com/features/scalability) ## [Scalability](https://www.recombee.com/features/scalability) Built as a high-performance, real-time distributed system, Recombee scales to support even the most demanding environments. [Scalable Infrastructure](https://www.recombee.com/features/scalability#scalable-infrastructure) [Secure Platform](https://www.recombee.com/features/scalability#secure-platform) --- # Recommendations & Search | Features > Source: https://www.recombee.com/features/recommendations-search > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [Features](https://www.recombee.com/features) [Recommendations & Search](https://www.recombee.com/features/recommendations-search) * [Real-Time Recommendations](#real-time-recommendations) * [Segmentations](#segmentations) * [Fully Personalized Homepage](#fully-personalized-homepage) * [Personalized Infinite Feed](#personalized-infinite-feed) * [Personalized Semantic Search](#personalized-semantic-search) * [Semantic Segmentations](#semantic-segmentations) * [Cutting-Edge AI & Research](#cutting-edge-ai-research) [Real-Time Analytics & Insights](https://www.recombee.com/features/real-time-analytics-insights) [Full Control with Scenario Settings](https://www.recombee.com/features/full-control-with-scenario-settings) [Business Rules](https://www.recombee.com/features/business-rules) [Integration](https://www.recombee.com/features/integration) [Scalability](https://www.recombee.com/features/scalability) Recommendations & SearchReal-Time Analytics & InsightsFull Control with Scenario SettingsBusiness RulesIntegrationScalability Recommendations & Search # Recommendations & Search ![](https://www.recombee.com/img/features/recommendations-search.png) ## Real-Time Recommendations Recombee functions as a dynamic real-time recommender, instantly adapting to new interactions—whether it’s a content view, product purchase, like, or dislike—and any content changes, such as adding new items or modifying attributes. This capability is particularly advantageous for platforms with fast-moving content, including [News Sites](https://www.recombee.com/domains/articles-news-media), [Deal Aggregators](https://www.recombee.com/domains/deal-aggregators), and [P2P Marketplaces](https://www.recombee.com/domains/p2p-marketplaces). Additionally, the real-time responsiveness of Recombee effectively addresses the cold-start challenge for new users. **From their very first interaction,** users receive **personalized recommendations,** which only become more refined as additional data is collected. ## Segmentations: Recommending Categories, Brands, Tags, Artists, and More While traditional recommender systems focus on suggesting individual pieces of content or products, Recombee goes further by offering recommendations for groups of related items, **known as Item Segments.** ![](https://www.recombee.com/img/features/segmentations.png) **Segmentation Examples** Categories/Genres Actors Brands Tags Artists or albums based on songs Individual homepage rows This comprehensive approach enhances personalization across a wider array of use cases than conventional systems allow. You can create dedicated sections like _"Your Favorite Brands"_ or _"Similar Artists,"_ and even rearrange UI elements to highlight the most relevant options for each user. This optimization significantly enhances their experience, exemplified by features like the [Fully Personalized Homepage](#fully-personalized-homepage). Item Segmentations are easily defined in the Admin UI using your existing catalog data. Segmentation can be based on a single attribute, such as category, or a combination of multiple attributes, leveraging our innovative [ReQL query language](https://www.recombee.com/features/business-rules#reql) for added flexibility. [Read more about Item Segmentations](https://www.recombee.com/blog/recombee-item-segmentations) [Read the Item Segmentations Documentation](https://docs.recombee.com/segmentations) ## Fully Personalized Homepage Deliver the ultimate personalized experience with the Fully Personalized Homepage. This innovative solution enables **complete personalization of both the content within each section and the order in which they appear.** For example, a user who loves thrillers will see a homepage that is uniquely tailored to their interests, showcasing different rows and videos compared to someone who prefers comedies. This bespoke approach ensures that each user enjoys a distinct experience that reflects their individual tastes. Recombee provides the essential tools to achieve this level of personalization: * [Item Segmentations:](#segmentations) Customize the order of rows and sections based on user preferences. * [Composite Recommendations:](https://docs.recombee.com/scenarios#composite-recommendations) Native support for returning both the row header (e.g., a category or genre) and its contents in a single API response. Supports scenarios such as [_Because You Watched_](https://docs.recombee.com/recipes/video/fully-personalized-homepage/because-you-watched) as well. * [Batch Requesting:](https://docs.recombee.com/api#batch) Obtain recommendations for all rows simultaneously, complete with automatic deduplication to avoid content overlap. [Explore the Fully Personalized Homepage Documentation](https://docs.recombee.com/recipes/video#fully-personalized-homepage) to learn more about creating tailored experiences for your users. ## Personalized Infinite Feed Engage your audience with a captivating Personalized Infinite Feed that dynamically adapts to their unique tastes. ![](https://www.recombee.com/img/features/personalized-infinite-feed.png) With Recombee, you can [seamlessly integrate](https://docs.recombee.com/api#recommend-next-items) an infinite scrolling experience, allowing a continuous stream of content that loads new results as users scroll. For TikTok- and Instagram-like **short-content experiences**, Recombee offers [out-of-the-box models](https://docs.recombee.com/recipes/video/feed/swiping-feed) optimized for swiping feeds, helping users discover relevant content in fast-moving, engagement-driven environments. This real-time adaptability makes Recombee a strong fit for platforms with **user-generated content (UGC)**, where content catalogs change quickly and user intent evolves with every interaction. Powered by advanced ensembles, Recombee keeps even the most dynamic feeds fresh, relevant, and highly engaging. Discover how this works in practice by checking out the [9GAG Case Study](https://www.recombee.com/case-studies/9gag). ## Personalized Semantic Search Deliver intent-driven, context-aware search results even when search terms don’t directly match the content. ![](https://www.recombee.com/img/features/personalized-semantic-search.png) Recombee’s search combines **full-text query** matching with machine learning on **user interaction** data to deliver precise, highly personalized results. For enhanced accuracy, our Premium feature, **Semantic Search, uses a large language model (LLM) to interpret the intent and context of user queries.** This approach goes beyond keyword matching by capturing the deeper semantic meaning of each query, allowing the system to retrieve results that align with the user’s true intentions. With Semantic Search, **users receive more accurate, context-aware results**—even if their query terms don’t directly match the content in the database. For example, they can search for "movies about racing" or "eco-friendly jackets for extreme cold," and discover precisely relevant recommendations. ## Semantic Segmentations Automatically uncover micro-categories, genres, and topics — and put them to work in recommendations and analytics. Recombee’s AI automatically discovers niche clusters of items based on their attributes and behavior. A large language model (LLM) then assigns these clusters **clear, meaningful names.** These **Semantic Segments** unlock powerful use cases: * **Automated discovery experiences** – generate dynamic homepage rows of micro-genres or trending topics, refreshed in real time. * **Actionable insights** – enrich analytics dashboards with semantic categories to reveal what users truly engage with. **Examples**: * In video, dynamically create homepage rails of micro-genres tailored to audience interests. * In news, automatically surface homepage sections around emerging topics of the day. ## Cutting-Edge AI & Research Stay ahead with a recommendation engine backed by world-class research, proven results, and continuous innovation. ![](https://www.recombee.com/img/features/cutting-edge-ai-and-research.png) Recombee’s machine learning stack combines deep learning, collaborative filtering, large language models, reinforcement learning, and transformer architectures like [**beeFormer**](https://www.recombee.com/blog/introducing-beeformer-a-framework-for-training-foundational-models-for-recommender-systems), purpose-built for delivering top-tier recommendations. Behind it all is **RecombeLab**, our in-house research division. We actively contribute to the scientific community by publishing cutting-edge work and supporting PhD students. Recombee-supported PhD researchers are encouraged to publish and share their algorithms publicly, helping advance the broader field of machine learning and recommendation systems. Our research has been featured at top conferences such as **RecSys**, **ICML**, and the **ACM Web Conference**, and we continuously push these advancements into production to ensure your system stays at the forefront of personalization technology. [Read more about our Research](https://www.recombee.com/research) ## Explore More [![Real-Time Analytics & Insights](https://www.recombee.com/img/features/real-time-analytics-insights.png)Real-Time Analytics & Insights](https://www.recombee.com/features/real-time-analytics-insights) [![Full Control with Scenario Settings](https://www.recombee.com/img/features/full-control-with-scenario-settings.png)Full Control with Scenario Settings](https://www.recombee.com/features/full-control-with-scenario-settings) --- # Accelerated Integration to Popular Platforms Using Our Plugins > Source: https://www.recombee.com/integrations > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. Integrations and plugins # Accelerated Integration to Popular Platforms Using Our Plugins Developed a custom plugin? Contact us for collaboration and listing your plugin. ### Official Plugins ![Recombee for Keboola](https://www.recombee.com/img/integrations/keboola.svg) Data Operations Platform [More ->](https://docs.recombee.com/keboola) ![Recombee for Segment](https://www.recombee.com/img/integrations/segment.svg) Customer Data Platform [More ->](https://docs.recombee.com/segment) ### Plugins Developed By Community ![Recombee for WordPress](https://www.recombee.com/img/integrations/wordpress.svg) Content Management System [More ->](https://www.wpsolr.com/feature-recombee/) ![Recombee for Drupal](https://www.recombee.com/img/integrations/drupal.svg) Content Management System [More ->](https://www.drupal.org/project/recombee) ![Laracombee](https://www.recombee.com/img/integrations/laravel.png) PHP Framework [More ->](https://github.com/amranidev/laracombee) ![Recombee for Kentico](https://www.recombee.com/img/integrations/kentico.svg) Content Management Platform [More ->](https://github.com/Kentico/xperience-module-recombee) ![Recombee for Kentico Kontent.](https://www.recombee.com/img/integrations/kentico-kontent.svg) Content Management System [More ->](https://github.com/Kentico/kontent-example-integration-recombee) ![n8n](https://www.recombee.com/img/integrations/n8n.svg) Workflow Automation Platform [More ->](https://github.com/tawfekov/n8n-nodes-recombee-api) ## Implement Your Own Integration Using Our SDKs [_![JavaScript](https://www.recombee.com/img/codes/icon-javascript.svg)_](https://github.com/recombee/js-api-client) [_![Python](https://www.recombee.com/img/codes/icon-python.svg)_](https://github.com/Recombee/python-api-client) [_![Ruby](https://www.recombee.com/img/codes/icon-ruby.svg)_](https://github.com/Recombee/ruby-api-client) [_![Java](https://www.recombee.com/img/codes/icon-java.svg)_](https://github.com/Recombee/java-api-client) [_![Node.js](https://www.recombee.com/img/codes/icon-nodejs.svg)_](https://github.com/Recombee/node-api-client) [_![Android](https://www.recombee.com/img/codes/icon-android.svg)_](https://github.com/Recombee/kotlin-api-client) [_![iOS](https://www.recombee.com/img/codes/icon-ios-white.svg)_](https://github.com/Recombee/swift-api-client) [_![PHP](https://www.recombee.com/img/codes/icon-php.svg)_](https://github.com/Recombee/php-api-client) [_![.NET](https://www.recombee.com/img/codes/icon-net.svg)_](https://github.com/Recombee/net-api-client) [_![Go](https://www.recombee.com/img/codes/icon-go.svg)_](https://github.com/Recombee/go-api-client) [_![REST](https://www.recombee.com/img/codes/icon-rest.svg)_](https://docs.recombee.com/getting_started) [**Documentation**](https://docs.recombee.com/) [**API Reference**](https://docs.recombee.com/api) --- # AI-Driven Recommendation Engine Suitable for Every Industry > Source: https://www.recombee.com/specialized-recommendations > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. ![](https://www.recombee.com/img/bg/polygon-green-xs.svg) ![](https://www.recombee.com/img/bg/polygon-green-2-md.svg) Specialized Recommendations # Tailored Recommendations for Your Industry Recombee finds a solution for every industry. Our recommendation engine personalizes and delivers content based on your individual user taste and preferences, saving their time. [Sign Up for Free](https://admin.recombee.com/sign-up) ## Flexible and Robust Solutions to Fit Your Industry ### P2P Marketplaces Utilize user-generated data including text descriptions and images for precise recommendations of fast-changing content. [**Explore**](https://www.recombee.com/domains/p2p-marketplaces) ### Real Estate Find the best fit per individual user by matching the attributes of your user and offering. [**Explore**](https://www.recombee.com/domains/real-estate) ### Job Portals Suggest the best fit for every individual client using our sequence prediction techniques. [**Explore**](https://www.recombee.com/domains/jobs-boards-hr-networking) ### Travel Increase the number of bookings by personalising the individual onsite user experience. [**Explore**](https://www.recombee.com/domains/travel-trips) ### Dating & Networking Recommend the best matching options to users based on their past behavior, interests, geolocation and other parameters. ### Have a Specific Case? ## Core Technology ### Adapting to your Data Personalization based on the Collaborative and Content-based filtering algorithms. ### Dynamically Retrained Models Real-time Personalization for your individual user at every point. ### Accelerated Integration Quick Integration through our well documented and easy to use APIs, SDKs. ### AI-powered A/B Testing To keep maximal KPIs at any time, AutoML AI is applied to optimize the algorithm ensembles. ### Advanced Business Rules Our solutions enable quick and easy addition of any Business Rules through boosters or filters. ### Real AI Inside Formation of Deep Neural Networks helps to predict the next action based on the historical behavior. ## How Recombee Works ![How Recombee Works Schema](https://www.recombee.com/img/products/how-it-works-specialized.svg) ## Success Story 3x more conversions from users who engage with recommendations "Our developers love Recombee documentation as well as quick and valuable technical support. We see Recombee’s **recommendationsToUser** algorithm as a great option to start offering a personalized experience on our site." _![Marco Alvarez](https://www.recombee.com/img/customers/segundamano.png)_Marco AlvarezProduct Manager at Segundamano ![Segundamano](https://www.recombee.com/img/logos/segundamano.png) [Read Case Study](https://www.recombee.com/case-studies/segundamano) ## Our Customers ![Zumper](https://www.recombee.com/img/logos/zumper.png)![itison](https://www.recombee.com/img/logos/itison.png)![Sabai99](https://www.recombee.com/img/logos/sabai99.png)![Romer](https://www.recombee.com/img/logos/romer.png)![Midland Realty](https://www.recombee.com/img/logos/midland-realty.png) and 1000+ other sites and apps. --- # Business Rules | Features > Source: https://www.recombee.com/features/business-rules > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [Features](https://www.recombee.com/features) [Recommendations & Search](https://www.recombee.com/features/recommendations-search) [Real-Time Analytics & Insights](https://www.recombee.com/features/real-time-analytics-insights) [Full Control with Scenario Settings](https://www.recombee.com/features/full-control-with-scenario-settings) [Business Rules](https://www.recombee.com/features/business-rules) * [ReQL](#reql) * [AI Assistant](#ai-assistant) [Integration](https://www.recombee.com/features/integration) [Scalability](https://www.recombee.com/features/scalability) Recommendations & SearchReal-Time Analytics & InsightsFull Control with Scenario SettingsBusiness RulesIntegrationScalability Business Rules # Custom Filtering and Boosting Rules With [Filters](https://www.recombee.com/features/full-control-with-scenario-settings#filters) and [Boosters](https://www.recombee.com/features/full-control-with-scenario-settings#boosters), you can highlight specific categories, brands, editor-picked content, local deals, and more, helping you deliver varied recommendation boxes and targeted email campaigns. ![](https://www.recombee.com/img/features/business-rules.png) While our predefined rules cover a wide range of scenarios, you might find a unique requirement that calls for a custom filtering or boosting rule. No problem, **Recombee’s ReQL language** is built to handle those situations with ease. ## ReQL ReQL is a powerful query language specifically designed for expressing Filters and Boosters based on the attributes of both items (content or products) and users. Its flexibility allows you to craft intricate rules tailored to your needs. ![](https://www.recombee.com/img/features/business-rules-reql.png) ReQL is feature-rich, boasting functions for geo-filtering, processing user interactions, and more. [Comprehensive documentation](https://docs.recombee.com/reql) on ReQL is available to help you harness its full potential. Within the Admin UI, you’ll find a user-friendly ReQL editor and validator that allows you to preview the outcomes of your Filters and Boosters. ### Example Queries Consider this straightforward filter: ```python "thriller" in 'genres' ``` This filter ensures that only thriller movies are recommended. The following example uses user-specific content suitability: ```python if context_user["is_kid_profile"] then 'is_kids_friendly' else true ``` This filter makes sure only kid-friendly content gets recommended to users flagged as minors. ## AI Assistant Our AI Assistant is on hand to support you in crafting these queries. The AI Assistant can **turn natural language requests into working ReQL** code. For example, if you type: _“Allow only premium content from the user’s country,”_ the AI Assistant will generate the following ReQL filter: ```python 'premium' and context_user["country"] == 'country' ``` It can also break down existing rules in plain terms, so you can understand and tweak them when needed. ## Applying the Rules Once you create a rule, it is seamlessly added to your Filters and Boosters library, making it [easy to apply to any Scenario](https://www.recombee.com/features/full-control-with-scenario-settings). That way, once your rules are in place, you can adjust parameters like how recent the content is or how close it is to the user, without requiring technical expertise. ## Explore More [![Full Control with Scenario Settings](https://www.recombee.com/img/features/full-control-with-scenario-settings.png)Full Control with Scenario Settings](https://www.recombee.com/features/full-control-with-scenario-settings) [![Integration](https://www.recombee.com/img/features/integration.png)Integration](https://www.recombee.com/features/integration) --- # Full Control with Scenario Settings | Features > Source: https://www.recombee.com/features/full-control-with-scenario-settings > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [Features](https://www.recombee.com/features) [Recommendations & Search](https://www.recombee.com/features/recommendations-search) [Real-Time Analytics & Insights](https://www.recombee.com/features/real-time-analytics-insights) [Full Control with Scenario Settings](https://www.recombee.com/features/full-control-with-scenario-settings) * [Scenario](#scenario) * [Logic](#logic) * [Filters](#filters) * [Boosters](#boosters) * [Constraints](#constraints) * [A/B Testing](#ab-testing) [Business Rules](https://www.recombee.com/features/business-rules) [Integration](https://www.recombee.com/features/integration) [Scalability](https://www.recombee.com/features/scalability) Recommendations & SearchReal-Time Analytics & InsightsFull Control with Scenario SettingsBusiness RulesIntegrationScalability Scenarios # Full Control with Scenario Settings Take charge with Scenario Settings, offering a suite of customization options to align recommendations with your **product vision.** ![](https://www.recombee.com/img/features/full-control-with-scenario-settings.png) ## Scenario Each placement for recommendations, referred to as a **Scenario,** can be uniquely tailored to meet diverse needs. **Here are a few examples** of what you can create: * **New Releases For You:** Present a curated selection from recently added content. * **Local Favorites:** Highlight popular content from the user’s country or area. * **Category Highlights:** Showcase personalized picks from the user's favorite category. All these configurations can be easily managed through our intuitive Admin UI, making customization a breeze. Let’s explore the individual options! ## Logic The Logic setting defines the **ensemble of machine learning models** applied to each specific Scenario. ![](https://www.recombee.com/img/features/logic.png) We offer a diverse array of Logics, each fine-tuned for distinct use cases and tailored to particular domains. Selecting the right Logic is simple and ensures optimal performance for your needs. For instance, you could choose a Logic designed for: * Recommending Similar Products * Optimizing "Watch-Next" Suggestions * Curating a Personalized News Feed Many of these Logics come with adjustable parameters, allowing you to fine-tune their behavior—for example, deciding whether to recommend content that users have already watched. [See List of Logics](https://docs.recombee.com/recommendation_logics) ## Filters Filters enable you to specify which content or products are **eligible for recommendations** within a given Scenario. ![](https://www.recombee.com/img/features/filters.png) **For example, you can restrict recommendations to:** * Articles published within the last 7 days * Promoted deals * Apartments located within 10 miles of the user * Kid-friendly content for minor users Our **library of predefined Filters** addresses many common needs, allowing for immediate application. If you don’t find a specific filter in our library, you can [easily create a custom one using our flexible ReQL language.](https://www.recombee.com/features/business-rules) Like Logic and Boosters, Filters can also be passed via the API on a per-request basis, making them ideal for dynamic scenarios based on the user’s current selections. ## Boosters Boosters enable you to **bias the recommendation engine toward your specific business goals.** ![](https://www.recombee.com/img/features/boosters.png) **For example, you can prioritize recommending:** * Newly added content * Curated selections handpicked by editors * Higher-end alternatives for upselling * Deals tailored to the user’s location Similar to Filters, we provide a library of predefined Boosters, and [custom rules can be defined using ReQL](https://www.recombee.com/features/business-rules). ## Constraints Constraints allow you to **manage the diversity** of recommended content or products effectively. ![](https://www.recombee.com/img/features/constraints.png) **For example, you can set limits such as:** * No more than two items from each category * No more than 50% of items from a single brand * Only one product per parent product ID ## A/B Testing A/B Testing helps you validate changes to your recommendation strategy using real user behavior. A/B Testing helps you validate changes to your recommendation strategy using real user behavior. Run an experiment within any Scenario and automatically split traffic between the current configuration—the Control—and one or more Variants. Test different Logics and their settings, Filters, Boosters, and Constraints, then measure their impact on metrics such as Click-Through Rate, Conversion Rate, Watch Time, Revenue, or any custom metric defined in [Insights analytics](https://www.recombee.com/features/real-time-analytics-insights). Clear reports show improvement over the Control, statistical significance, and the probability that each Variant will outperform the others—so you can roll out optimizations with confidence. ## Explore More [![Real-Time Analytics & Insights](https://www.recombee.com/img/features/real-time-analytics-insights.png)Real-Time Analytics & Insights](https://www.recombee.com/features/real-time-analytics-insights) [![Business Rules](https://www.recombee.com/img/features/business-rules.png)Business Rules](https://www.recombee.com/features/business-rules) --- # Integration | Features > Source: https://www.recombee.com/features/integration > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [Features](https://www.recombee.com/features) [Recommendations & Search](https://www.recombee.com/features/recommendations-search) [Real-Time Analytics & Insights](https://www.recombee.com/features/real-time-analytics-insights) [Full Control with Scenario Settings](https://www.recombee.com/features/full-control-with-scenario-settings) [Business Rules](https://www.recombee.com/features/business-rules) [Integration](https://www.recombee.com/features/integration) * [Easy Integration](#easy-integration) * [API SDKs](#api-sdks) * [Widget SDKs](#widget-sdks) * [No-Code Widgets](#no-code-widgets) * [Catalog Feeds](#catalog-feeds) * [Support](#support) [Scalability](https://www.recombee.com/features/scalability) Recommendations & SearchReal-Time Analytics & InsightsFull Control with Scenario SettingsBusiness RulesIntegrationScalability Integration # Easy Integration Recombee makes integration a breeze with its well-structured **REST API** and **SDKs**. Coupled with tools like catalog feed processing, **No-Code widgets**, and pre-built integrations with platforms like [Segment](https://docs.recombee.com/segment), we streamline the entire integration process for your development teams. ![](https://www.recombee.com/img/integration-schema.svg) [Read the documentation](https://docs.recombee.com/) ## API SDKs Take advantage of SDKs available in multiple programming languages, supporting server-side, client-side, and hybrid integrations. ![](https://www.recombee.com/img/features/sdks.png) Recombee offers API SDKs in multiple programming languages, ensuring effortless implementation and maintenance for both server-side and client-side integrations, or a combination of both. ![](https://www.recombee.com/img/codes/icon-javascript.svg)JavaScript ![](https://www.recombee.com/img/codes/icon-python.svg)Python ![](https://www.recombee.com/img/codes/icon-ruby.svg)Ruby ![](https://www.recombee.com/img/codes/icon-java.svg)Java ![](https://www.recombee.com/img/codes/icon-nodejs.svg)Node.js ![](https://www.recombee.com/img/codes/icon-android.svg)Android ![](https://www.recombee.com/img/codes/icon-ios.svg)iOS ![](https://www.recombee.com/img/codes/icon-php.svg)PHP ![](https://www.recombee.com/img/codes/icon-net.svg).NET ![](https://www.recombee.com/img/codes/icon-go.svg)Go ![](https://www.recombee.com/img/codes/icon-rest.svg)REST * [_![](https://www.recombee.com/img/codes/icon-javascript.svg)_JavaScript](#code-javascript) * [_![](https://www.recombee.com/img/codes/icon-python.svg)_Python](#code-python) * [_![](https://www.recombee.com/img/codes/icon-ruby.svg)_Ruby](#code-ruby) * [_![](https://www.recombee.com/img/codes/icon-java.svg)_Java](#code-java) * [_![](https://www.recombee.com/img/codes/icon-nodejs.svg)_Node.js](#code-nodejs) * [_![](https://www.recombee.com/img/codes/icon-android.svg)_Android](#code-android) * [_![](https://www.recombee.com/img/codes/icon-ios.svg)_iOS](#code-ios) * [_![](https://www.recombee.com/img/codes/icon-php.svg)_PHP](#code-php) * [_![](https://www.recombee.com/img/codes/icon-net.svg)_.NET](#code-net) * [_![](https://www.recombee.com/img/codes/icon-go.svg)_Go](#code-go) * [_![](https://www.recombee.com/img/codes/icon-rest.svg)_REST](#code-rest) ```javascript const client = new recombee.ApiClient('database-id', dbPublicToken); // Send a view of item 'item_x' by user 'user_42' client.send(new recombee.AddDetailView('user_42', 'item_x')); // Get 5 recommended items for user 'user_42'. Recommend only items which haven't expired yet. const recommended = await client.send( new recombee.RecommendItemsToUser('user_42', 5, {filter: "'expires' > now()"}) ); ``` [See Recombee API Client for JavaScript on Github](https://github.com/recombee/js-api-client) ```ruby client = RecombeeClient.new('database-id', secret_token) # Send a view of item 'item_x' by user 'user_42'. Create user and/or item if it doesn't exist yet. client.send(AddDetailView.new('user_42', 'item_x', 'cascadeCreate' => true)) # Get 5 recommended items for user 'user_42'. Recommend only items which haven't expired yet. recommended = client.send(RecommendItemsToUser.new('user_42', 5, 'filter' => "'expires' > now()")) ``` [See Recombee API Client for Ruby on Github](https://github.com/Recombee/ruby-api-client) ```java RecombeeClient client = new RecombeeClient("database-id", secretToken); // Send a view of item "item_x" by user "user_42". Create user and/or item if it doesn't exist yet. client.send(new AddDetailView("user_42", "item_x").setCascadeCreate(true)); // Get 5 recommended items for user "user_42". Recommend only items which haven't expired yet. RecommendationResponse recommended = client.send( new RecommendItemsToUser("user_42", 5).setFilter("'expires' > now()") ); ``` [See Recombee API Client for Java on Github](https://github.com/recombee/java-api-client) ```python client = RecombeeClient('database-id', secret_token) # Send a view of item 'item_x' by user 'user_42'. Create user and/or item if it doesn't exist yet. client.send(AddDetailView('user_42', 'item_x', cascade_create=True)) # Get 5 recommended items for user 'user_42'. Recommend only items which haven't expired yet. recommended = client.send(RecommendItemsToUser('user_42', 5, filter="'expires' > now()")) ``` [See Recombee API Client for Python on Github](https://github.com/recombee/python-api-client) ```javascript const client = new recombee.ApiClient('database-id', secretToken); // Send a view of item 'item_x' by user 'user_42'. Create user and/or item if it doesn't exist yet. client.send(new rqs.AddDetailView('user_42', 'item_x', {cascadeCreate: true}), callback); // Get 5 recommended items for user 'user_42'. Recommend only items which haven't expired yet. const recommended = await client.send( new rqs.RecommendItemsToUser('user_42', 5, {filter: "'expires' > now()"}) ); // Supports both Promises and callbacks ``` [See Recombee API Client for Node.js on Github](https://github.com/Recombee/node-api-client) ```kotlin val client = RecombeeClient(databaseId = "yourDatabaseId", publicToken = dbPublicToken,) // Send a view of item 'item_x' by user 'user_42' client.send(AddDetailView("user_42", "item_x")); // Get 5 recommended items for user 'user_42'. Recommend only items which haven't expired yet. val result = client.sendAsync(RecommendItemsToUser("user_42", count = 5, filter = "'expires' > now()")) result.onSuccess { response: RecommendationResponse -> // Show recommendations } ``` [See Recombee API Client for Android on Github](https://github.com/Recombee/kotlin-api-client) ```swift let client = RecombeeClient(databaseId: "yourDatabaseId", publicToken: dbPublicToken, region: .euWest) try await client.send(AddDetailView(userId: "user_42", itemId: "item_x")) let result = try await client.send(RecommendItemsToUser(userId: "user_42", count: 5, filter: "'expires' > now()")) result.recomms.forEach { print($0.id) } ``` [See Recombee API Client for iOS on Github](https://github.com/Recombee/swift-api-client) ```php $client = new Client('database-id', $secret_token); // Send a view of item 'item_x' by user 'user_42'. Create user and/or item if it doesn't exist yet. $client->send(new Reqs\AddDetailView('user_42', 'item_x', ['cascadeCreate' => true])); // Get 5 recommended items for user 'user_42'. Recommend only items which haven't expired yet. $recommended = $client->send(new Reqs\RecommendItemsToUser('user_42', 5, ['filter' => "'expires' > now()"])); ``` [See Recombee API Client for PHP on Github](https://github.com/Recombee/php-api-client) ```csharp var client = new RecombeeClient("database_id", secretToken); // Send a view of item "item_x" by user "user_42". Create user and/or item if it doesn't exist yet. client.Send(new AddDetailView("user_42", "item_x", cascadeCreate: true)); // Get 5 recommended items for user 'user_42'. Recommend only items which haven't expired yet. var recommended = client.Send(new RecommendItemsToUser("user_42", 5, filter: "'expires' > now()")); ``` [See Recombee API Client for .NET on Github](https://github.com/Recombee/net-api-client) ```go client, _ := recombee.NewRecombeeClient("databaseId", secretToken) // Send a view of item 'item_x' by user 'user_42', creating user/item if they don't exist client.NewAddDetailView("user_42", "item_x").SetCascadeCreate(true).Send() // Get 5 recommended items for user 'user_42', filtering out expired items recommendRes, _ := client.NewRecommendItemsToUser("user_42", 5).SetFilter("'expires' > now()").Send() // Print recommended item IDs for _, rec := range recommendRes.Recomms { fmt.Println(rec.Id) } ``` [See Recombee API Client for .NET on Github](https://github.com/Recombee/net-api-client) ```ruby # Send a view of item 'item_x' by user 'user_id'. Create user and/or item if it doesn't exist yet. POST https://rapi.recombee.com/database_id/detailviews/ Data: {'userId': 'user_42', 'itemId': 'item_x', 'cascadeCreate': true} # Get 5 recommended items for user 'user_42'. # Recommend only items which haven't expired yet (filter: 'expires' > now()). GET https://rapi.recombee.com/database_id/recomms/users/user_42/items/?count=5&filter=%27expires%27%3Enow() ``` [See more API documentation](https://docs.recombee.com/getting_started) [Read SDKs documentation](https://docs.recombee.com/api_clients) ## Widget SDKs Display recommendations on your site with ease, while keeping full control over how they look and feel. ![](https://www.recombee.com/img/features/widget-sdks.png) Recombee’s Widget SDKs are a collection of **customizable, pre-built components** that let you **easily embed personalized recommendations and search** into your website or app. They speed up development by offering ready-to-use components for carousels, feeds, grids, and quick search. These widgets handle the heavy lifting for you, from fetching recommendations to managing loading states and reacting to user interactions. The widgets are built for flexibility, so your team can **fully adapt them** to match your site’s design using **custom CSS** or your **own rendering templates**. They’re also **fully responsive**, so users get a seamless experience on any screen size or device. The SDKs are **optimized for modern front-end frameworks like React**, making it simple to build rich, fast, and personalized user experiences without sacrificing performance or design consistency. [Explore Widget SDKs](https://docs.recombee.com/widget-sdks/) ## No-Code Widgets For those seeking a no-code option, Recombee provides an intuitive way to integrate recommendations and search functionality directly into your website. ![](https://www.recombee.com/img/features/integration-html-widgets.png) Our "WYSIWYG (What You See Is What You Get)" editor in the Recombee Admin UI allows you to design and customize widgets to fit both desktop and mobile layouts. **Available Widget Types** ![](https://www.recombee.com/img/features/integration-html-widget-grid.png)Grid ![](https://www.recombee.com/img/features/integration-html-widget-carousel.png)Carousel ![](https://www.recombee.com/img/features/integration-html-widget-feed.png)Infinite Feed ![](https://www.recombee.com/img/features/integration-html-widget-quick-search.png)Quick Search Once you've configured your widget, deployment is straightforward: simply **copy the provided embed code and paste it into your website.** Whether for recommendations or a search box, the widget will seamlessly integrate into your site. ![](https://www.recombee.com/img/features/integration-html-widgets-embed.png) You can also enhance its appearance further by applying your own CSS for a perfect match with your website’s style. [Read No-Code Widgets Documentation](https://docs.recombee.com/no-code-widgets) ## Catalog Feeds Catalog synchronization with Recombee can be achieved through the API or by using catalog feeds. Popular formats like **Google Merchant feed, RSS feed, custom XML, CSV, and JSON** are supported out of the box. Plus, feed processing settings are **easy to configure** directly in the Recombee Admin UI. [Read Catalog Feeds Documentation](https://docs.recombee.com/catalog_feeds) ## Support ![](https://www.recombee.com/img/features/support.png) [Documentation](https://docs.recombee.com/): Recombee offers **comprehensive public documentation** to assist with your integration and usage needs. **Email Support**: For any questions or further clarification, our skilled support team is readily available at [support@recombee.com](mailto:support@recombee.com). **Video Calls and Slack Support:** Available exclusively for Pro and Premium plans, our team is just a video call or Slack message away, offering tailored support and guidance. ## Explore More [![Business Rules](https://www.recombee.com/img/features/business-rules.png)Business Rules](https://www.recombee.com/features/business-rules) [![Scalability](https://www.recombee.com/img/features/scalability.png)Scalability](https://www.recombee.com/features/scalability) --- # Real-Time Analytics & Insights | Features > Source: https://www.recombee.com/features/real-time-analytics-insights > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [Features](https://www.recombee.com/features) [Recommendations & Search](https://www.recombee.com/features/recommendations-search) [Real-Time Analytics & Insights](https://www.recombee.com/features/real-time-analytics-insights) * [Real-Time Data](#real-time-data) * [Library of Insights](#library-of-insights) * [Custom Insights](#custom-insights) * [A/B Testing](#ab-testing) [Full Control with Scenario Settings](https://www.recombee.com/features/full-control-with-scenario-settings) [Business Rules](https://www.recombee.com/features/business-rules) [Integration](https://www.recombee.com/features/integration) [Scalability](https://www.recombee.com/features/scalability) Recommendations & SearchReal-Time Analytics & InsightsFull Control with Scenario SettingsBusiness RulesIntegrationScalability Real-Time Analytics & Insights # Real-Time Analytics & Insights Unlock the power of data with Real-Time Analytics & Insights. The Insights section within the Admin UI provides a range of predefined and fully customizable reports, allowing you to monitor how users engage with recommendations and your platform. This transparency enables **close tracking of the recommender's performance,** user engagement metrics, and organic interactions. ## Real-Time Data Experience the benefits of data updated in near real-time, so you can swiftly assess the impact of any changes made to Scenarios or their settings. ## Library of Insights ![](https://www.recombee.com/img/features/insights-recombee-library.png) Explore our curated library of visualizations, designed for common analytical use cases. These visualizations are ready to use, while still offering flexibility for adjustments and deeper exploration. ## Custom Insights ![](https://www.recombee.com/img/features/insights-custom-insight.png) **Easily create a variety of visualizations and report types,** including stacked and line charts, as well as tables. Agile data slicing, selective filtering, and compound metrics calculations enable tailored insights. In harmony with Recombee's robust customization features, the Insights module empowers you to **focus on your unique KPIs,** enabling informed and strategic decision-making. [Explore More About Insights](https://www.recombee.com/blog/insights-the-next-level-of-analytics-in-recombee-ui) ## A/B Testing ![](https://www.recombee.com/img/features/insights-ab-testing.png) Use Insights metrics in A/B Tests to compare recommendation strategies and identify the best-performing configuration. [Read More](https://www.recombee.com/features/full-control-with-scenario-settings#ab-testing) ## Explore More [![Recommendations & Search](https://www.recombee.com/img/features/recommendations-search.png)Recommendations & Search](https://www.recombee.com/features/recommendations-search) [![Full Control with Scenario Settings](https://www.recombee.com/img/features/full-control-with-scenario-settings.png)Full Control with Scenario Settings](https://www.recombee.com/features/full-control-with-scenario-settings) --- # Scalability | Features > Source: https://www.recombee.com/features/scalability > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [Features](https://www.recombee.com/features) [Recommendations & Search](https://www.recombee.com/features/recommendations-search) [Real-Time Analytics & Insights](https://www.recombee.com/features/real-time-analytics-insights) [Full Control with Scenario Settings](https://www.recombee.com/features/full-control-with-scenario-settings) [Business Rules](https://www.recombee.com/features/business-rules) [Integration](https://www.recombee.com/features/integration) [Scalability](https://www.recombee.com/features/scalability) * [Scalable Infrastructure](#scalable-infrastructure) * [Secure Platform](#secure-platform) Recommendations & SearchReal-Time Analytics & InsightsFull Control with Scenario SettingsBusiness RulesIntegrationScalability Scalability ## Scalability Recombee is a powerful, real-time distributed system designed for high performance and unparalleled scalability. ![](https://www.recombee.com/img/features/scalability.png) It is built to deliver on these key objectives: * **Handle High Volume:** Effortlessly process **over 30,000 recommendation requests per second,** making it ideal for high-traffic websites. * **Adapt Quickly:** Support **real-time model updates** and incremental training, ensuring rapid adaptation to emerging trends and catalog changes. * **Maintain Availability:** Ensure fault tolerance and sustain **high service availability,** keeping your recommendations always on point. * **Secure Your Data:** Provide robust **security and data protection** at every level, safeguarding your information with the highest standards. [Read more about the architecture and used technologies](https://www.recombee.com/how-it-works/performance-at-scale) ## Explore More [![Integration](https://www.recombee.com/img/features/integration.png)Integration](https://www.recombee.com/features/integration) [![Recommendations & Search](https://www.recombee.com/img/features/recommendations-search.png)Recommendations & Search](https://www.recombee.com/features/recommendations-search) --- # How It Works > Source: https://www.recombee.com/how-it-works > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. How It Works [Overview](https://www.recombee.com/how-it-works) [Data & Results](https://www.recombee.com/how-it-works/data-and-results) [AI & Machine Learning](https://www.recombee.com/how-it-works/ai-and-machine-learning) [Performance at Scale](https://www.recombee.com/how-it-works/performance-at-scale) OverviewData & ResultsAI & Machine LearningPerformance at Scale Learn # How Recombee Works Recombee recommendation engine was from the very beginning designed as a modern, secure, real-time, and horizontally scalable distributed cloud system. This robust architecture has led to its success, enabling us to serve hundreds of businesses and reach hundreds of millions of end users globally. ![](https://www.recombee.com/img/how-it-works/main.png) ## Dive Into the Details [![](https://www.recombee.com/img/how-it-works/data-and-results.png)Data & ResultsLearn about the types of data Recombee processes to tailor your users' experiences.Explore ->](https://www.recombee.com/how-it-works/data-and-results) [![](https://www.recombee.com/img/how-it-works/ai-and-machine-learning.png)AI & Machine LearningUncover the cutting-edge technologies we use to power your personalized recommendations.Explore ->](https://www.recombee.com/how-it-works/ai-and-machine-learning) [![](https://www.recombee.com/img/how-it-works/performance-at-scale.png)Performance at ScaleFind out how we ensure rapid delivery of recommendations, regardless of traffic volume.Explore](https://www.recombee.com/how-it-works/performance-at-scale) --- # AI & Machine Learning | How It Works > Source: https://www.recombee.com/how-it-works/ai-and-machine-learning > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. How It Works [Data & Results](https://www.recombee.com/how-it-works/data-and-results) [AI & Machine Learning](https://www.recombee.com/how-it-works/ai-and-machine-learning) [Performance at Scale](https://www.recombee.com/how-it-works/performance-at-scale) Data & ResultsAI & Machine LearningPerformance at Scale Learn # Next-Generation Machine Learning for Recommendations and Search From foundational algorithms to cutting-edge AI, Recombee's research-driven approach combines advanced machine learning techniques to deliver high accuracy and performance at scale for both recommender systems and personalized search. ## Research-Driven Innovation Our commitment to research and development has led to breakthrough technologies like [beeFormer](https://www.recombee.com/blog/introducing-beeformer-a-framework-for-training-foundational-models-for-recommender-systems) \- a transformer architecture optimized specifically for recommendation tasks. We’ve also introduced state-of-the-art solutions in scalable collaborative filtering. Our Research and Machine Learning teams collaborate closely to bring the latest innovations into our product. [Explore Our Research ->](https://www.recombee.com/research) ![AI structure](https://www.recombee.com/img/how-it-works/research-driven-innovations.png) State-of-the-Art AI Next-generation AI technologies * beeFormer architecture. * LLM-powered recommendations and personalized search. * Multimodal deep learning. * Advanced sequence modeling with efficient RNNs. Advanced ML Core Sophisticated machine learning methods * Matrix factorization at scale. * Deep autoencoder architectures. * Reinforcement learning. * Hybrid collaborative filtering. Retrieval-Based & Heuristic Methods Optimized implementations of proven techniques * Exact and approximate sparse and dense vector search. * Advanced popularity and trends time series modeling. * Intelligent rotation, explorations, and reminder models. * Advanced popularity modeling. * Real-time item discovery. * Intelligent reminder systems. ![AI structure](https://www.recombee.com/img/how-it-works/ai.svg) ## Technical Excellence ### Deep Learning Innovation Our advanced deep learning models capture complex user-item interactions using sophisticated neural architectures, allowing for a nuanced understanding of user preferences and content relationships. ### Transformer Technology beeFormer, our custom transformer model, helps solve the cold start problem and improves content relevance and understanding. ### Large Language Models Integration of LLMs enables natural language understanding and generation, powering conversational recommendations and complex content insights. ### Real-Time Hybrid Approaches Our systems dynamically combine multiple recommendation strategies to ensure optimal performance across different scenarios and data conditions. ## Unique Features We’ve engineered our ML models to deliver real-time recommendations and search results across content, products, categories, artists, tags, and more - enhancing discovery in diverse domains. [![](https://www.recombee.com/img/how-it-works/recommendations-search.png)Real-Time RecommendationsExplore ->](https://www.recombee.com/features/recommendations-search#real-time-recommendations) [![](https://www.recombee.com/img/how-it-works/personalized-semantic-search.png)Personalized Semantic SearchExplore ->](https://www.recombee.com/features/recommendations-search#personalized-semantic-search) [![](https://www.recombee.com/img/how-it-works/segmentations.png)SegmentationsExplore ->](https://www.recombee.com/features/recommendations-search#segmentations) ## Why AI from Recombee Our commitment to research excellence and practical innovation sets us apart. By combining cutting-edge AI with proven methodologies, we build recommendation systems that consistently outperform traditional approaches. ### Research Leadership Pioneering new approaches in recommendation systems. ### Scalable Solutions Enterprise-ready implementations that scale with your needs. ### Proven Results Delivering measurable improvements in user engagement. ### Responsibility Safe AI systems trained to help users. ## Read Next [![](https://www.recombee.com/img/how-it-works/data-and-results.png)Data & ResultsLearn about the types of data Recombee processes to tailor your users' experiences.Explore ->](https://www.recombee.com/how-it-works/data-and-results) [![](https://www.recombee.com/img/how-it-works/performance-at-scale.png)Performance at ScaleFind out how we ensure rapid delivery of recommendations, regardless of traffic volume.Explore ->](https://www.recombee.com/how-it-works/performance-at-scale) --- # Data & Results | How It Works > Source: https://www.recombee.com/how-it-works/data-and-results > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. How It Works [Data & Results](https://www.recombee.com/how-it-works/data-and-results) [AI & Machine Learning](https://www.recombee.com/how-it-works/ai-and-machine-learning) [Performance at Scale](https://www.recombee.com/how-it-works/performance-at-scale) Data & ResultsAI & Machine LearningPerformance at Scale Learn # Data & Results Recommendations are based on the provided data. The two most important types of data are the **catalog** & the **interaction data**. * **Catalog** \- information about the content or products you want to recommend (e.g. title, categories, description, images, etc.). * **Interaction Data** \- what a user has viewed, watched, purchased, etc. ![](https://www.recombee.com/img/how-it-works/schema-how-it-works.svg) ## Catalog ### Item Properties Describe the items that you want to recommend. * Categories * Text Description * Images * Labels * Genres * Expiration Date * Destination * Age Restriction * Geo Location * and more ### User Properties Describe your users. * Gender * Language * Age * Subscription * Registration date * and more ## Interactions ### Live Interactions Interactions between Users and Items are the most important data for the recommender system. Various Interaction Types * Views * Purchases * Cart Additions * Ratings (Likes) * Bookmark * View Portions ### Interaction History (Optional) ## Data Processing ### Recombee Processes the Ingested Data Instantly in Real-Time Thanks to our innovative incremental training of the models, new interactions and new content are taken into consideration within milliseconds. ### Machine Learning (ML) To come with the best performing recommendations and search, Recombee uses a **huge portfolio of the ML and AI techniques**. * Collaborative filtering algorithms * Content-based algorithms * Reinforcement learning algorithms * and many more [Explore More About Machine Learning And AI](https://www.recombee.com/how-it-works/ai-and-machine-learning) Scenario ## Configuration ### Recombee offers **unprecedented configurability** of the resulting recommendations and search results. **For each place where you show the recommendations, you can specify:** * The behavior of the recommendation model * What content/products can be recommended * What content/products shall be boosted in the recommendations ... and more, to align the recommendations with **your product vision**. [Explore More About Scenario Settings](https://www.recombee.com/features/full-control-with-scenario-settings) ## Showing Results You can personalize the whole user experience. ### Personalization is often applied to * Homepage rows * Full-text search * Product/Content detail * Watch/Read next * Personal feed * Email campaigns ### Recombee empowers you to implement advanced use cases like * Personalizing the order of rows or sections on the homepage * Recommending specific categories or brands * Suggesting artists or albums based on songs previously listened to Sending data and requesting recommendations is made easy thanks to our [![](https://www.recombee.com/img/how-it-works/api.png)Clear APIExplore](https://www.recombee.com/features/integration) [![](https://www.recombee.com/img/how-it-works/sdks.png)SDKsExplore](https://www.recombee.com/features/integration#sdks) [![](https://www.recombee.com/img/how-it-works/widgets.png)No-Code WidgetsExplore](https://www.recombee.com/features/integration#html-widgets) [Read More About Integration](https://www.recombee.com/features/integration) ![](https://www.recombee.com/img/how-it-works/insights-point.svg) ## Analyzing Results ### Insights provides targeted control over your unique KPIs, enabling informed, strategic decisions. Insights analytics section in the Admin UI offers various **predefined and fully customizable reports** to track how users interact with recommendations and your platform. [Explore Insights](https://www.recombee.com/features/real-time-analytics-insights) ## Read Next [![](https://www.recombee.com/img/how-it-works/ai-and-machine-learning.png)AI & Machine LearningUncover the cutting-edge technologies we use to power your personalized recommendations.Explore ->](https://www.recombee.com/how-it-works/ai-and-machine-learning) [![](https://www.recombee.com/img/how-it-works/performance-at-scale.png)Performance at ScaleFind out how we ensure rapid delivery of recommendations, regardless of traffic volume.Explore ->](https://www.recombee.com/how-it-works/performance-at-scale) --- # Performance at Scale | How It Works > Source: https://www.recombee.com/how-it-works/performance-at-scale > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. How It Works [Data & Results](https://www.recombee.com/how-it-works/data-and-results) [AI & Machine Learning](https://www.recombee.com/how-it-works/ai-and-machine-learning) [Performance at Scale](https://www.recombee.com/how-it-works/performance-at-scale) Data & ResultsAI & Machine LearningPerformance at Scale Learn # Performance at Scale Recombee is built as a real-time, high-performance, distributed system to achieve the following objectives. ![](https://www.recombee.com/img/how-it-works/performance-at-scale-m.png) **Process** vast volumes of recommendation requests for high-traffic websites (up to 30k recommendations per second). **Support real-time model updates** and incremental training to quickly adapt to emerging trends and catalog changes. **Ensure** fault tolerance and maintain **high service availability**. **Provide** robust **security and data protection** at all levels. ## Architecture Recombee operates on several hundred high-performance servers, hosting thousands of applications that interact within a micro-service architecture. This setup includes multiple layers of storage and distributed processing components, all equipped with self-healing and self-recovery capabilities. These features ensure that Recombee maintains high system availability and reliability. ![](https://www.recombee.com/img/how-it-works/architecture.svg) ## Geographical Regions Users depending on our recommendations or search results are spread across the globe. Recombee operates its clusters in several geographic regions. This allows you to **choose the region** that is closest to your users or servers in terms of **latency.** Currently, you can choose from the following regions: * US West Coast * Canadian East Coast * Europe (Germany) * Australia (Sydney) ![](https://www.recombee.com/img/how-it-works/regions.png) ## Technology Stack We build our components using a variety of industry-proven technologies. This ensures reliability, efficiency, and scalability in our systems, meeting the high standards required by our clients across diverse sectors. [![Puppet](https://www.recombee.com/img/techstack/puppet.svg)](https://www.puppet.com/) [![Kubernetes](https://www.recombee.com/img/techstack/kubernetes.svg)](https://kubernetes.io/) [![Docker](https://www.recombee.com/img/techstack/docker.svg)](https://www.docker.com/) [![Aerospike](https://www.recombee.com/img/techstack/aerospike.svg)](https://www.aerospike.com/) [![Grafana](https://www.recombee.com/img/techstack/grafana.svg)](https://grafana.com/) [![Elasticsearch](https://www.recombee.com/img/techstack/elasticsearch.svg)](https://www.elastic.co/) [![GitLab](https://www.recombee.com/img/techstack/gitlab.svg)](https://about.gitlab.com/) [![Sentry](https://www.recombee.com/img/techstack/sentry.svg)](https://sentry.io/) [![Icinga](https://www.recombee.com/img/techstack/icinga.svg)](https://icinga.com/) [![PostgreSQL](https://www.recombee.com/img/techstack/postgresql.svg)](https://www.postgresql.org/) [![Apache Kafka](https://www.recombee.com/img/techstack/apachekafka.svg)](https://kafka.apache.org/) [![Vault](https://www.recombee.com/img/techstack/vault.svg)](https://www.vaultproject.io/) [![TensorFlow](https://www.recombee.com/img/techstack/tensorflow.svg)](https://www.tensorflow.org/) [![Prometheus](https://www.recombee.com/img/techstack/prometheus.svg)](https://prometheus.io/) [![Flux](https://www.recombee.com/img/techstack/flux.svg)](https://fluxcd.io/) [![Cilium](https://www.recombee.com/img/techstack/cilium.svg)](https://cilium.io/) [![Jira](https://www.recombee.com/img/techstack/jira.svg)](https://www.atlassian.com/software/jira) [![NVIDIA CUDA](https://www.recombee.com/img/techstack/nvidiacuda.png)](https://developer.nvidia.com/cuda-zone) [![MinIO](https://www.recombee.com/img/techstack/minio.svg)](https://min.io/) [![Thanos](https://www.recombee.com/img/techstack/thanos.svg)](https://thanos.io/) [![nginx](https://www.recombee.com/img/techstack/nginx.svg)](https://www.nginx.com/) [![ClickHouse](https://www.recombee.com/img/techstack/clickhouse.svg)](https://clickhouse.com/) [![Timescale](https://www.recombee.com/img/techstack/timescale.svg)](https://www.timescale.com/) [![Keras](https://www.recombee.com/img/techstack/keras.png)](http://keras.io) ## Read Next [![](https://www.recombee.com/img/how-it-works/ai-and-machine-learning.png)AI & Machine LearningUncover the cutting-edge technologies we use to power your personalized recommendations.Explore ->](https://www.recombee.com/how-it-works/ai-and-machine-learning) [![](https://www.recombee.com/img/how-it-works/data-and-results.png)Data & ResultsLearn about the types of data Recombee processes to tailor your users' experiences.Explore ->](https://www.recombee.com/how-it-works/data-and-results) --- # Engage Readers with Personalized, Real-Time Recommendations > Source: https://www.recombee.com/domains/articles-news-media > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [Domains](https://www.recombee.com/where-to-use#domains) / Articles, News, Media # Engage Readers with Personalized, Real-Time Recommendations Recombee’s advanced 1:1 content personalization is designed to **boost engagement metrics** and **maximize subscriber lifetime value** while preserving the editorial voice. ![Articles, News, Media](https://www.recombee.com/img/domains/articles-news-media.png) ![The Telegraph](https://www.recombee.com/img/customers-domains/the-telegraph.svg)![9GAG](https://www.recombee.com/img/customers-domains/9gag.svg)![Grow](https://www.recombee.com/img/customers-domains/grow.svg)![Mediavine](https://www.recombee.com/img/customers-domains/mediavine.svg)![FTV Prima](https://www.recombee.com/img/customers-domains/ftv-prima.svg)![The Sporting News](https://www.recombee.com/img/customers-domains/the-sporting-news.svg)![Vltava Labe Media](https://www.recombee.com/img/customers-domains/vltava-labe-media.svg) ## Use Cases ![](https://www.recombee.com/img/use-cases/articles-news-media/articles-for-you.png) ### Articles For You Boost engagement by delivering personalized, 1:1 article recommendations based on user's interests and browsing history. Recipe Soon ![](https://www.recombee.com/img/use-cases/articles-news-media/related-articles.png) ### Related Articles Enable users to discover content that’s contextually relevant, and personalized to their immediate reading experience. [Explore Recipe](https://docs.recombee.com/recipes/news/article/read-next) ![](https://www.recombee.com/img/use-cases/articles-news-media/trending-articles.png) ### Trending Articles Highlight the hottest local articles in each user's region, showcasing the diversity and relevance of your platform. [Explore Recipe](https://docs.recombee.com/recipes/news/homepage/top-stories) ![](https://www.recombee.com/img/use-cases/articles-news-media/search-articles.png) ### Search Articles Make finding articles easy with full-text search capabilities, ensuring desired content is found within a few keystrokes. [Explore Recipe](https://docs.recombee.com/recipes/news/search-articles) ![](https://www.recombee.com/img/use-cases/articles-news-media/weekly-digest-email.png) ### Periodic Newsletters Enhance the discovery experience through personalized emails with articles tailored to the user's unique preferences. [Explore Recipe](https://docs.recombee.com/recipes/news/personalized-emailing) ![](https://www.recombee.com/img/use-cases/articles-news-media/personalized-news-feed.png) ### Personalized News Feed Automate and customize each user's news feed rows, delivering 1:1 personalization for a highly tailored experience. [Explore Recipe](https://docs.recombee.com/recipes/news/homepage/personalized-feed) ![](https://www.recombee.com/img/use-cases/articles-news-media/read-next.png) ### Read Next Keep users engaged in the reading flow by offering infinite real-time scroll recommendations based on their ongoing reading journey. [Explore Recipe](https://docs.recombee.com/recipes/news/article/read-next) ![](https://www.recombee.com/img/use-cases/articles-news-media/cross-site-recommendation.png) ### Cross-Site Recommendation Deliver recommendations seamlessly across all your sites—whether for articles, e-commerce, classifieds, or multimedia content—enabling fluid discovery through customizable, cross-platform suggestions. [Explore Recipe](https://docs.recombee.com/recipes/news/homepage/cross-site-recommendations) ![](https://www.recombee.com/img/use-cases/articles-news-media/top-sections-for-you.png) ### Sections For You Showcase personalized sections for each user, reflecting their interaction history to offer more relevant content. [Explore Recipe](https://docs.recombee.com/recipes/news/homepage/personalized-sections-with-reordering) ### News & Articles Recipes Discover how to personalize key use cases within your news platform. [Explore Integration Recipes](https://docs.recombee.com/recipes/news) "We use Recombee to power our AI personalization & Search at The Telegraph. It immediately proved its value, securing a 35% CTR uplift in an A/B test against a competing solution while simultaneously enhancing our editors' ability to manage and analyse content performance. Beyond the numbers, the collaboration with the Recombee team has been excellent, they helped us push our thinking and we were delighted to jointly present at the RecSys conference and be recognized as an INMA '26 finalist for "Best Use of Generative AI"." _![](https://www.recombee.com/img/customers/the-telegraph.png)_ **Tom Kelleher** Director of Emerging Technology – AI & Personalisation at The Telegraph [Read Case Study](https://www.recombee.com/case-studies/the-telegraph) [Explore Success Stories](https://www.recombee.com/case-studies) ## Why Recombee [Real-Time News Delivery](#deliver-breaking-news-in-real-time) [Boost Readership & Ad Revenue](#increase-reader-engagement-and-ad-revenue) [Acquire & Retain Subscribers](#acquire-retain-subscribers) [Editor-Guided Recommendations](#human-in-the-loop-achieve-better-results-with-editors-in-charge) [Smarter Content Discovery](#enhance-content-discoverability-with-diverse-recommendations) [Handle High Traffic & Major Events](#handle-large-traffic-peaks-and-nationwide-events) ### Deliver Breaking News in Real Time Be the first to present emerging topics in real time, positioning your platform as the go-to source for breaking news. ### Increase Reader Engagement and Ad Revenue Optimize CTR, average session duration, and recirculation rate with tailored content, even for niche topics. ### Acquire & Retain Subscribers Boost, filter, or segment content effortlessly with Recombee's advanced business rules, delivering high-converting recommendations and keeping subscribers engaged through compelling newsletters. ### Human-In-The-Loop: Achieve Better Results with Editors in Charge Blend the scalability of AI with the expertise of human editors using easily configurable rules in Recombee’s interface or through its seamlessly integrable API. ### Enhance Content Discoverability With Diverse Recommendations Combat filter bubbles and echo chambers with varied suggestions powered by scientifically-backed algorithms. ### Handle Large Traffic Peaks and Nationwide Events Utilize Recombee’s scalable solution to handle high-traffic events, ensuring real-time personalized recommendations for all users. ![Recombee Insights](https://www.recombee.com/img/domains/wwd-insights.png) ### Real-Time Analytics & Insights Unlock in-depth data analysis with Recombee’s Insights tool, accessible through the Admin UI. * **Empower** editors and product teams with insights on recommended articles, popular topics, categories, and more. * **Visualize** data, create custom reports, and seamlessly share insights with your team. * Choose from a **library of analytical views** or build custom graphs and reports. * Leverage analytical data to refine rules and ensure the recommender engine aligns with **your product vision**. [Explore Insights](https://www.recombee.com/features/real-time-analytics-insights) ## Enhance Editorial Efficiency, Increase Subscriptions, and Boost Emerging Topics in Real-Time With new content emerging every minute, maintaining a deep connection with individual readers is increasingly challenging. **Recombee's AI swiftly identifies sudden emerging topics** and **prominently features them** on your site, driving traffic and ensuring you remain the go-to platform for trending news. Predictive algorithms analyze content attributes and reader behavior to recommend the most relevant articles. Our models recognize users’ preferences within milliseconds, providing **real-time recommendations**—even to **unknown or first-time visitors.** Recombee offers the application of boosters and filters, allowing for the **automatic organization of content**, such as boosting premium content or editorial picks. The recommendation engine analyzes each content property, including **title, date of publication, author, tags, expiration date, lead paragraphs, and paid/unpaid status**, applying **Natural Language Processing (NLP) in 80 languages**. It reacts to each new click or newly released content immediately, analyzing interactions like **detail views, bookmarks, scroll rates**, and **likes**. ### Explore more on [Content Recommendations](https://www.recombee.com/content-recommendations) [News and Articles Integration Tips](https://docs.recombee.com/integration_tips#integration-tips-content-articles) Discover Even More Features ## Case Studies [![The Telegraph](https://www.recombee.com/img/case-studies/small-covers/case-studies-the-telegraph.png)+35%CTR uplift on article detail pagesMedia Company](https://www.recombee.com/case-studies/the-telegraph) [![9GAG](https://www.recombee.com/img/case-studies/small-covers/case-studies-9gag.png)+37%Post ViewsCross-Platform Entertainment Network](https://www.recombee.com/case-studies/9gag) [![Unfiltered Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-unfiltered-media-group.png)+50%CTRMedia Company](https://www.recombee.com/case-studies/unfiltered-media-group) ## Core Technology ### Adapting to Your Data A robust system that utilizes all available data to generate high-quality recommendations for users, incorporating content-based models and Natural Language Processing (NLP). ### Dynamically Retrained Models Real-time content personalization that adapts to evolving customer preferences while addressing the rapidly changing media landscape. ### Industry-Specific Functionalities for News and Articles Deep Natural Networks that recognize emerging topics and automatically prioritize trending content suggestions. ### AI-powered A/B Testing In-house AutoML AI that maximizes KPIs and continuously optimizes deep learning algorithms. ### Advanced Business Rules Customizable boosters and filters that highlight desired content, with easy-to-adjust rules for further content optimization. ### Real AI Inside Reinforcement learning and advanced algorithms recognize user preferences within milliseconds to deliver real-time recommendations, including the boosting of premium content. [Explore Features](https://www.recombee.com/features) [Explore How Recombee Works](https://www.recombee.com/how-it-works/data-and-results) "We conducted A/B testings of multiple recommendation engines to find the best content personalization solution. Out of all solutions, only **Recombee outperformed our internal read-next recommendations** of news articles. After long-lasting A/B testing, Recombee achieved a **40% higher CTR of suggested articles,** which ultimately led to the deployment of the solution to most of our news sites (iDNES Lidovky, Expres)." **Petr Kelin** Manager at MAFRA, a.s. [Read Case Study](https://www.recombee.com/case-studies/mafra) [Explore Success Stories](https://www.recombee.com/case-studies) --- # Drive Repeat Visits With Personalized Deals > Source: https://www.recombee.com/domains/deal-aggregators > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [Domains](https://www.recombee.com/where-to-use#domains) / Deal Aggregators # Drive Repeat Visits With Personalized Deals Surface the right promotions before they expire based on **user behavior and live inventory** to increase **click-through rates, partner yield, and repeat visits**. ![Deal Aggregators](https://www.recombee.com/img/domains/deal-aggregators.png) ![](https://www.recombee.com/img/customers-domains/tripadvisor.svg)![](https://www.recombee.com/img/customers-domains/slickdeals.svg)![](https://www.recombee.com/img/customers-domains/benefithub.svg)![Showmax](https://www.recombee.com/img/customers-domains/pepper.svg)![Showmax](https://www.recombee.com/img/customers-domains/dealstream.svg)![Showmax](https://www.recombee.com/img/customers-domains/itison.svg) ## Use Cases ![](https://www.recombee.com/img/use-cases/deal-aggregators/fully-personalized-feed.png) ### Fully Personalized Feed Engage users with an infinite personalized feed of deals, blending the user's preferences with current popular deals. ![](https://www.recombee.com/img/use-cases/deal-aggregators/similar-deals.png) ### Similar Deals Offer users compelling alternatives from diverse affiliate partners. ![](https://www.recombee.com/img/use-cases/deal-aggregators/promoted-deals-for-you.png) ### Promoted Deals For You Provide users with personalized selections from premium affiliate partners' deals. ![](https://www.recombee.com/img/use-cases/deal-aggregators/trending-deals-in-your-area.png) ### Trending Deals In Your Area Show the current hottest deals on the entire site or within the user's local area. ![](https://www.recombee.com/img/use-cases/deal-aggregators/faceted-search-and-category-browsing.png) ### Faceted Search & Category Browsing Display deals based on individual search criteria like category, brand, price, or any other specifics from your catalog. ![](https://www.recombee.com/img/use-cases/deal-aggregators/personalized-emailing.png) ### Personalized Emailing Send users a personalized selection of the latest hot deals. ## Why Recombee [Increase Ad Revenue & Affiliate Clicks](#increase-ad-revenue-and-outclicks-to-affiliate-links) [Tackle Cold Start with Real-Time Training](#overcome-cold-start-problem-with-real-time-model-training) [Show Time-Sensitive Offers](#display-emerging-time-sensitive-offers) [Highlight High-Commission Deals](#boost-visibility-of-high-commission-affiliate-offers) [Promote Local Deals](#promote-local-deals) ### Increase Ad Revenue and Outclick’s to Affiliate Links Drive more outclicks and elevate ad revenue by increasing page views with personalized content recommendations. ### Overcome Cold Start Problem with Real-Time Model Training Immediately match new deals to the right users with continuous model training. ### Display Emerging Time-Sensitive Offers Highlight trending items, including limited, short-lived and seasonal deals. ### Boost Visibility of High-Commission Affiliate Offers Prefer deals by the premium and high-commission affiliate partners. ### Promote Local Deals Leverage geo-location data to present users with top deals in their city or area. ![Recombee Insights](https://www.recombee.com/img/domains/wwd-insights.png) ### Real-Time Analytics & Insights Take a deep dive into your data using our advanced real-time analytics tool called Insights within the Recombee Admin UI. * **Analyze** out clicks or purchases per affiliate partner, deal category, area, and much more * **Visualize** your data, create custom reports, and seamlessly share insights with your team * Choose from a **library of analytical views** or create your graphs and reports * Using the analytical data, adjust the rules and let the recommender engine **follow your product vision** [Explore Insights](https://www.recombee.com/features/real-time-analytics-insights) ## Increase Click-Outs and Customer Satisfaction by Delivering a Personalized Experience As a deal aggregator, leveraging machine learning and AI recommendations can significantly enhance your platform's performance and user satisfaction. By implementing Recombee’s AI-driven engine, you can personalize offers based on user preferences and browsing behavior from the very first interaction in real-time, ensuring each customer receives the most relevant deals. This not only increases engagement and conversion rates but also boosts customer loyalty and retention. Unlike competitors, Recombee offers 1:1 personalization for every user, immediate recommendations of new deals (including time-limited and short-lived offers), and instant discovery of trending hot deals. Recombee also provides you with full control over the recommendations, allowing you to align them with your business strategy. Through our intuitive Admin UI, you can prioritize deals from premium and high-commission affiliate partners, highlight soon-to-expire deals, or promote local offers. ### Explore more on [Product Recommendations](https://www.recombee.com/product-recommendations) [Marketplaces Integration Tips](https://docs.recombee.com/integration_tips#product-recommendations) Discover Even More Features ## Case Studies [![Slickdeals](https://www.recombee.com/img/case-studies/small-covers/case-studies-slickdeals.png)+70%CTR to Detail Page ViewsE-commerce + Deal Aggregators](https://www.recombee.com/case-studies/slickdeals) [![Pepper](https://www.recombee.com/img/case-studies/small-covers/case-studies-pepper.png)+21%Click-outs to affiliate links from websiteDeal Aggregators](https://www.recombee.com/case-studies/pepper) [![itison](https://www.recombee.com/img/case-studies/small-covers/case-studies-itison.png)2 000%ROINewsletter Personalization + Deal Aggregators](https://www.recombee.com/case-studies/itison) ## Core Technology ### Adapting to Your Data A robust system that can utilize all data available to generate great recommendations for your users, including collaborative filtering and content-based models. ### Dynamically Retrained Models Real-time content personalization to suit the flourishing customer’s tastes and adaptation of fast-changing user-generated content. ### Specific Functionalities for Deal Aggregators Image processing to analyze items using pictures and NLP to analyze ads’ attributes from the text descriptions. ### AI-powered A/B Testing In-house AutoML AI that maximizes KPIs and continuously optimizes deep learning algorithms. ### Advanced Business Rules Boosters or filters to push forward desired listings and easy to manipulate, adjustable rules for additional optimization of your content. ### Real AI Inside Reinforcement learning and collaborative filtering to recommend personalized ads or listings based on historical on-site behavior. [Explore Features](https://www.recombee.com/features) [Explore How Recombee Works](https://www.recombee.com/how-it-works/data-and-results) "It is a real pleasure collaborating with Recombee. Their problem-solving skills have proven invaluable, helping us overcome various business challenges while allowing us to consistently increase our click-outs and deliver a better user experience. Thanks to their solution we've seen our click-outs increase by up to 21%. They have become a trusted partner I can highly recommend." _![](https://www.recombee.com/img/customers/pepper.png)_ **Heike Guertler** Head of Product at Pepper [Read Case Study](https://www.recombee.com/case-studies/pepper) [Explore Success Stories](https://www.recombee.com/case-studies) --- # Drive Revenue Growth With Personalized Product Discovery > Source: https://www.recombee.com/domains/e-commerce > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [Domains](https://www.recombee.com/where-to-use#domains) / E-commerce # Drive Revenue Growth With Personalized Product Discovery Guide shoppers to products they’re most likely to buy with **real-time 1:1 recommendations** that increase **conversions, average order value, and repeat purchases**. ![E-commerce](https://www.recombee.com/img/domains/e-commerce.png) ![](https://www.recombee.com/img/customers-domains/temple-webster.svg)![](https://www.recombee.com/img/customers-domains/potten-pannen.svg)![Showmax](https://www.recombee.com/img/customers-domains/fithub.svg) ## Use Cases ![](https://www.recombee.com/img/use-cases/e-commerce/fully-personalized-homepage.png) ### Fully Personalized Homepage Automate and tailor all your homepage rows 1:1 for each user. [Explore Recipe](https://docs.recombee.com/recipes/e-commerce#fully-personalized-homepage) ![](https://www.recombee.com/img/use-cases/e-commerce/shopping-cart.png) ### Shopping Cart Allow your customers to purchase products or store a list of the items for the future. [Explore Recipe](https://docs.recombee.com/recipes/e-commerce/cart) ![](https://www.recombee.com/img/use-cases/e-commerce/alternative-products.png) ### Alternative Products Present alternatives to searched items or suggest suitable replacements for discontinued products. [Explore Recipe](https://docs.recombee.com/recipes/e-commerce/product-detail/alternative-products-and-upsell) ![](https://www.recombee.com/img/use-cases/e-commerce/faceted-search.png) ### Faceted Search Show products based on individual search criteria like price, color, size, material, or any other specifics from your catalog. [Explore Recipe](https://docs.recombee.com/recipes/e-commerce/search/faceted-search) ![](https://www.recombee.com/img/use-cases/e-commerce/personalized-emailing.png) ### Personalized Emailing Enhance the shopping experience with tailored complementary product offerings for recent purchases. [Explore Recipe](https://docs.recombee.com/recipes/e-commerce/personalized-emailing) ![](https://www.recombee.com/img/use-cases/e-commerce/popular-and-trending.png) ### Popular & Trending Recommend the most popular and trending products with individual preferences in mind. [Explore Recipe](https://docs.recombee.com/recipes/e-commerce/fully-personalized-homepage/bestsellers) ![](https://www.recombee.com/img/use-cases/e-commerce/upsell.png) ### Upsell Encourage customers to buy a higher-end version of a product. [Explore Recipe](https://docs.recombee.com/recipes/e-commerce/product-detail/alternative-products-and-upsell) ![](https://www.recombee.com/img/use-cases/e-commerce/cross-sell.png) ### Cross-Sell Recommend complementary products while considering individual preferences. [Explore Recipe](https://docs.recombee.com/recipes/e-commerce/product-detail/bought-together-and-accessories) ![](https://www.recombee.com/img/use-cases/e-commerce/top-brands-for-you.png) ### Top Brands for You Show personalized brands to each user based on their interaction history. [Explore Recipe](https://docs.recombee.com/recipes/e-commerce/fully-personalized-homepage/brands-for-you) ![](https://www.recombee.com/img/use-cases/e-commerce/new-arrivals.png) ### New Arrivals Generate tailored product recommendations based on the latest additions to the product catalog and increase the likelihood of a purchase. ![](https://www.recombee.com/img/use-cases/e-commerce/personalized-category-browsing-with-infinite-scroll.png) ### Personalized Category Browsing With Infinite Scroll Let customers browse categories with an infinite scroll of personalized recommendations generated in real-time. ### E-Commerce Recipes Discover how to personalize various use cases within your e-commerce site. [Explore Integration Recipes](https://docs.recombee.com/recipes/e-commerce) ## Why Recombee [Maximize Average Order Value and Sales](#maximize-average-order-value-and-sales) [Optimize Product Discovery](#optimize-product-discovery) [Boost Specific Products](#boost-specific-products) [Drive Sales with Next Basket Prediction](#drive-sales-with-next-basket-prediction) [Utilize Instant User Understanding](#utilize-instant-user-understanding) [Seamlessly Integrate with Widgets and Feeds](#seamlessly-integrate-with-widgets-and-feeds) ### Maximize Average Order Value and Sales Create a personalized experience for each user, driving conversions across the entire funnel. ### Optimize Product Discovery Ensure customers discover the most relevant products through personalized recommendations and search, leveraging your catalog and user behavior data. ### Boost Specific Products Utilize Recombee's business rules to enhance the visibility of specific products based on parameters such as brand, margin, or affiliate partner. ### Drive Sales with Next Basket Prediction Help your users fill their shopping carts based on their previous behavior by employing Recombee’s machine learning models. ### Utilize Instant User Understanding Harness Recombee's real-time models to provide personalized product recommendations to anonymous users right from their first interaction. ### Seamlessly Integrate with Widgets and Feeds Start using Recombee with out-of-the-box tools designed for seamless integration into your online store. ![Recombee Insights](https://www.recombee.com/img/domains/wwd-insights.png) ### Real-Time Analytics & Insights Take a deep dive into your data using our advanced real-time analytics tool called Insights within the Recombee Admin UI. * **Analyze** top-recommended products, conversion rate per individual categories, generated profit thanks to recommendations, and more with our robust analytical tool's visualizations and reports * **Visualize** your data, create custom reports, and seamlessly share insights with your team * Choose from a **library of analytical views** or create your graphs and reports * Using the analytical data, adjust the rules and let the recommender engine **follow your product vision** [Explore Insights](https://www.recombee.com/features/real-time-analytics-insights) ## Convert Visitors to Loyal Customers With a Personalization Service Use geolocation and customer preferences to present what each customer seeks and increase the number of shoppers. Aside from an inflow of positive reviews and an improved user experience, personalization can lead up to a **30% increase of shopping carts,** and personalized email marketing can add up to a **30% increase in CTR.** Put trust in our **product recommendation engine** leveraging your **customer insights** to personalize homepage to **individual tastes,** push forward your **best-seller items** or use our models to cleverly upsell/ cross-sell desired items. Offer the right product at the right time with the use of predictive algorithms and business rules to support your sales strategy and ensure long-term customer loyalty. Maximize the use of your data for various conversions and stay one step ahead of your competition. Recombee’s robust recommendation engine analyzes item properties such as **title, description, availability, price** or other attributes from your **product feed** and interactions like **detail view, adding to a wishlist** or **purchase.** ### Explore more on [Product Recommendations](https://www.recombee.com/product-recommendations) [E-commerce Integration Tips](https://docs.recombee.com/integration_tips#product-recommendations) Discover Even More Features ## Case Studies [![Reliving](https://www.recombee.com/img/case-studies/small-covers/case-studies-reliving.png)+37%Place BidE-commerce + P2P Marketplaces](https://www.recombee.com/case-studies/reliving) [![Cooklist](https://www.recombee.com/img/case-studies/small-covers/case-studies-cooklist.png)+27%CTRE-commerce](https://www.recombee.com/case-studies/cooklist) [![Autohaus Kunzmann](https://www.recombee.com/img/case-studies/small-covers/case-studies-autohaus-kunzmann.png)+14%Conversion RateE-commerce](https://www.recombee.com/case-studies/autohaus-kunzmann) ## Core Technology ### Adapting to Your Data A robust system that can utilize all data available to generate great recommendations for your users, including collaborative filtering and content-based models. ### Dynamically Retrained Models Real-time content personalization to keep up with the number of newly added products and meet the flourishing customer’s tastes. ### Specific Functionalities for E-commerce Image processing to analyze items using pictures and NLP to analyze product attributes of available stock. ### AI-powered A/B Testing In-house AutoML AI that maximizes KPIs and continuously optimizes deep learning algorithms. ### Advanced Business Rules Boosters or filters to push forward desired products and easy to manipulate, adjustable rules for additional optimization of your content. ### Real AI Inside Next basket prediction and on-site recommendation algorithms based on deep learning, reinforcement learning and other methods to optimize conversions. [Explore Features](https://www.recombee.com/features) [Explore How Recombee Works](https://www.recombee.com/how-it-works/data-and-results) "It has been really difficult to find a solution that could integrate many different data sources and be 100% customizable, but with Recombee we get this. Recombee has fundamentally changed how we are serving recommendations and has really helped us grow. In addition to this, they have great customer support and are always ready to help." _![](https://www.recombee.com/img/customers/design-group.png)_ **Nicholas Blicker Larsen** CEO at Design Group [Read Case Study](https://www.recombee.com/case-studies/design-group) [Explore Success Stories](https://www.recombee.com/case-studies) --- # Get Better-Fit Applications With Smarter Job Matching > Source: https://www.recombee.com/domains/jobs-boards-hr-networking > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [Domains](https://www.recombee.com/where-to-use#domains) / Job Boards, HR, Networking # Get Better-Fit Applications With Smarter Job Matching Recommend relevant roles based on each candidate’s **context and real-time behavior** to increase **job discovery, applications, and qualified matches**. ![Job boards, HR, Networking](https://www.recombee.com/img/domains/jobs-boards-hr-networking.png) ![](https://www.recombee.com/img/customers-domains/startupjobs.svg)![](https://www.recombee.com/img/customers-domains/sorbet.svg)![](https://www.recombee.com/img/customers-domains/perkspot.svg)![Poslovi Infostud](https://www.recombee.com/img/customers-domains/poslovi-infostud.svg) ## Use Cases ![](https://www.recombee.com/img/use-cases/jobs-boards-hr-networking/recommended-jobs-for-you.png) ### Recommended Jobs For You Recommend personalized job offers based on history and individual preferences. ![](https://www.recombee.com/img/use-cases/jobs-boards-hr-networking/similar-job-offers.png) ### Similar Job Offers Present alternatives to searched jobs. ![](https://www.recombee.com/img/use-cases/jobs-boards-hr-networking/similar-candidates.png) ### Similar Candidates Showcase candidates with similar qualifications, skills, and experience that may also be a fit for the job. ![](https://www.recombee.com/img/use-cases/jobs-boards-hr-networking/personalized-search.png) ### Personalized Search Match search queries and tailor job offers to individual needs. ![](https://www.recombee.com/img/use-cases/jobs-boards-hr-networking/more-from-this-employer.png) ### More From This Employer Show more job listings from a chosen employer. ![](https://www.recombee.com/img/use-cases/jobs-boards-hr-networking/personalized-emailing.png) ### Personalized Emailing Run campaigns through personalized emails with newly added job offers matching individual criteria. ![](https://www.recombee.com/img/use-cases/jobs-boards-hr-networking/recommended-employers.png) ### Recommended Employers Let candidates discover companies that match their career preferences and offer attractive opportunities. ![Recombee Insights](https://www.recombee.com/img/domains/wwd-insights.png) ### Real-Time Analytics & Insights Take a deep dive into your data using our advanced real-time analytics tool called Insights within the Recombee Admin UI. * **Analyze** what listings are being recommended per sector, the performance of recommendations for individual employers or per area, and much more * **Visualize** your data, create custom reports, and seamlessly share insights with your team * Choose from a **library of analytical views** or create your graphs and reports * Using the analytical data, adjust the rules and let the recommender engine **follow your product vision** [Explore Insights](https://www.recombee.com/features/real-time-analytics-insights) ## Leverage Customer Insights and Become the Go-to Platform for Work Recommendations Applying to jobs has never been a simple task, resulting in thousands of CVs and profiles in circulation every day. Use the combination of our **deep learning recommendations** and hyper-personalization to **push the relevant job opportunities** and applicants **forward.** Our **natural language processing** can recognize texts in any language, enabling our engine to scan and compare applicants within seconds. Make use of our **machine learning** models that read CVs without the need of tagging for smarter match with work recommendations. **Faster** and more **accurate matching** of applicants to employers will guarantee the return of both customers in the future, **benefiting your job listing platform** and giving you a **competitive edge over your rivals.** Recombee’s robust recommendation engine analyzes properties such as **degree level, skillset or availability** and interactions like **viewed** or **recently applied to jobs.** In particular, providing us information such as applicant’s and listing’s geo-locations can be used to increase the precision of matchmaking. ### Explore more on [Specialized Recommendations](https://www.recombee.com/specialized-recommendations) Discover Even More Features ## Case Studies [![Poslovi Infostud](https://www.recombee.com/img/case-studies/small-covers/case-studies-poslovi-infostud.png)+32%Application conversion on job listingsOnline Job Board](https://www.recombee.com/case-studies/poslovi-infostud) ## Core Technology ### Adapting to Your Data A robust system that can utilize all data available to generate great recommendations for your users, including collaborative filtering and content-based models. ### Dynamically Retrained Models Real-time content personalization to keep up with the number of newly added job listings every day. ### Specific Functionalities for Job Boards Machine learning to read CVs without the need of tagging, and NLP recognizing texts in 80+ languages. ### AI-powered A/B Testing In-house AutoML AI that maximizes KPIs and continuously optimizes deep learning algorithms. ### Advanced Business Rules Boosters or filters to push forward desired listings and easy to manipulate, adjustable rules for additional optimization of your content. ### Real AI Inside State of the art machine learning algorithms recommending job vacancies based on historical on-site behavior. [Explore Features](https://www.recombee.com/features) [Explore How Recombee Works](https://www.recombee.com/how-it-works/data-and-results) ## Chosen Customers ![perkspot](https://www.recombee.com/img/logos/perkspot.svg)![sorbet](https://www.recombee.com/img/logos/sorbet.png)![poslovi-infostud](https://www.recombee.com/img/logos/poslovi-infostud.svg) [Explore Success Stories](https://www.recombee.com/case-studies) --- # Keep Listeners Coming Back With Personalized Recommendations > Source: https://www.recombee.com/domains/music-podcasts > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [Domains](https://www.recombee.com/where-to-use#domains) / Music, Podcasts # Keep Listeners Coming Back With Personalized Recommendations Recommend the next song, artist, album, or playlist based on **listener context and real-time behavior** to drive **discovery, engagement, and repeat listening**. ![Music, Podcasts](https://www.recombee.com/img/domains/music-podcasts.png) ![](https://www.recombee.com/img/customers-domains/audiomack.svg)![](https://www.recombee.com/img/customers-domains/sleepiest.svg)![](https://www.recombee.com/img/customers-domains/voxa.svg)![](https://www.recombee.com/img/customers-domains/airbit.svg)![](https://www.recombee.com/img/customers-domains/singa.svg) ## Use Cases ![](https://www.recombee.com/img/use-cases/music-podcasts/fully-personalized-homepage.png) ### Fully Personalized Homepage Automate and tailor all your homepage rows 1:1 for each user. ![](https://www.recombee.com/img/use-cases/music-podcasts/songs-artists-from-similar-genre.png) ### Similar Songs/Artists Present new songs and podcasts based on the content the user is currently enjoying. ![](https://www.recombee.com/img/use-cases/music-podcasts/playlists-made-for-you.png) ### Playlists Made For You Customize playlists by showing the most relevant songs or podcasts from a pre-selected list. ![](https://www.recombee.com/img/use-cases/music-podcasts/new-releases.png) ### New Releases Give listeners an overview of the week's new releases. ![](https://www.recombee.com/img/use-cases/music-podcasts/recommended-artists-for-you.png) ### Recommended Artists For You Inspire your users with work from artists they might enjoy based on their listening history. ![](https://www.recombee.com/img/use-cases/music-podcasts/quick-search.png) ### Quick Search Match search queries and tailor the music and podcast selection to individual preferences. ![](https://www.recombee.com/img/use-cases/music-podcasts/play-next.png) ### Play Next Don't disrupt the stream. Deliver recommendations for more music or podcasts at the end of playback. ![](https://www.recombee.com/img/use-cases/music-podcasts/trending-in-your-country.png) ### Trending in Your Country Showcase the diversity of your platform with the hottest local hits in specific regions. ![](https://www.recombee.com/img/use-cases/music-podcasts/personalized-emailing.png) ### Personalized Emailing Sweeten the streaming experience through personalized emails with music and podcasts tailored to the user's taste. ## Why Recombee [Music & Podcast Suggestions](#recommend-songs-albums-artists-and-podcasts) [Boost Revenue & Playtime](#uplift-revenues-boost-playtime-subscriptions-and-retention-rates) [Personalized Listening](#offer-great-personalized-recommendations-for-music-and-podcasts) [Highlight Local & Niche Picks](#recommend-local-and-niche-content) [Adapt to Mood in Real Time](#provide-real-time-responses-to-mood-changes) [Fine-Tune Recommender Behavior](#control-the-behaviour-of-recommendations) ### Recommend Songs, Albums, Artists and Podcasts Recombee automatically understands the links between songs, albums, and artists. ### Uplift Revenues, Boost Playtime, Subscriptions and Retention Rates Increase number of plays, active users, artist follows, ad revenues and other metrics. ### Offer Great Personalized Recommendations for Music and Podcasts Recommend music that fits the user using advanced collaborative filtering models. ### Recommend Local and Niche Content Allow listeners to discover unique content creators within a large catalog of songs and podcasts, including sensitivity to particular cultural regions. ### Provide Real-Time Responses To Mood Changes Recommend content selections based on different situations and time of the day. ### Control the Behaviour of Recommendations Use Recombee Business rules to set constraints, boosters, filter out content with explicit lyrics or by unverified artists. ![Recombee Insights](https://www.recombee.com/img/domains/wwd-insights.png) ### Real-Time Analytics & Insights Take a deep dive into your data using our advanced real-time analytics tool called Insights within the Recombee Admin UI. * **Analyze** top-recommended content, trending artists or albums, and other aspects of user behavior such as local content consumption * **Visualize** your data, create custom reports, and seamlessly share insights with your team * Choose from a **library of analytical views** or create your graphs and reports * Using the analytical data, adjust the rules and let the recommender engine **follow your product vision** [Explore Insights](https://www.recombee.com/features/real-time-analytics-insights) "Striving to be the ever limitless music sharing and discovery platform, we need to make sure the user experience of our listeners is smooth and sound. And one of the most critical aspects of achieving such a goal is content personalization tailored 1:1 in real-time. That's why we switched to Recombee. Thanks to their recommender engine, our **monthly plays increased by 206%** and **weekly follows by 67%.** Because the recommendations performed so well, we moved them from our Search page to the top of our main Discover tab. They are now **the best-performing module within that tab, accounting for 46% of all plays.**" _![](https://www.recombee.com/img/customers/audiomack-2.png)_ **Christopher Dalla Riva** Senior Product Manager at Audiomack [Read Case Study](https://www.recombee.com/case-studies/audiomack) ## Increase the Number of Subscribers and Time Spent With Personalized Playlists and Podcasts With more aspiring artists and easy access to music and podcasts, it is no easy task to stay competitive. We analyze consumed content, favorite artists, speakers, genres, or descriptions in multiple languages, to help your platform offer podcasts and **music recommendations tailored to personal tastes.** Beyond basic data, Recombee’s recommendation engine works with information about which songs/podcasts were listened to till the end, which halfway or skipped completely. Utilize Recombee to offer recommendations of genres, artists, songs or playlists to keep the **listener entertained and eager to revisit your platform.** Recombee’s robust recommendation engine analyzes item properties such as **title, genre, author, language** or **tags** and interactions like **view, replay, like** or **rating.** Autoplay is one of many features that Recombee offers to the listeners and enhances their time spent on the platform. ### Explore more on [Content Recommendations](https://www.recombee.com/content-recommendations) [Videos & Music Integration Tips](https://docs.recombee.com/integration_tips#integration-tips-content-music-and-podcasts) Discover Even More Features ## Case Studies [![Audiomack](https://www.recombee.com/img/case-studies/small-covers/case-studies-audiomack.png)+206%Monthly plays from recommendationsMusic](https://www.recombee.com/case-studies/audiomack) ## Core Technology ### Adapting to Your Data A robust system that can utilize all data available to generate great recommendations for your users, including collaborative filtering and content-based models. ### Dynamically Retrained Models Real-time content personalization that adapts to the flourishing customer’s tastes and considers the newly added music or podcast content. ### Specific Functionalities for Music Platforms Recommendations taking into account the users’ listening time; which titles were listened to until the end, which were listened to halfway or skipped completely. ### AI-powered A/B Testing In-house AutoML AI that maximizes KPIs and continuously optimizes deep learning algorithms. ### Advanced Business Rules Boosters or filters to push forward desired songs or genres and easy to manipulate, adjustable rules for additional optimization of your content. ### Real AI Inside Reinforcement learning and other algorithms designed to recognize the preferences of individual users and predict desired music or podcast with higher accuracy boosting user engagement. [Explore Features](https://www.recombee.com/features) [Explore How Recombee Works](https://www.recombee.com/how-it-works/data-and-results) "**Working with Recombee** to develop an affordable solution to provide our users with excellent music recommendations has **exceeded our expectations in every way.** They have been able to understand the relationships in the data of our industry and create **effective models to use efficiently.** We also appreciate the **nuance and flexibility** they offer when it comes to deciding the right solution based on quality, cost, complexity, speed, and other factors." _![](https://www.recombee.com/img/customers/audiomack-1.png)_ **Ty Wangsness** Founder/CTO at Audiomack [Read Case Study](https://www.recombee.com/case-studies/audiomack) ## Chosen Customers ![audiomack](https://www.recombee.com/img/logos/audiomack.svg)![stingray](https://www.recombee.com/img/logos/stingray.svg)![airbit](https://www.recombee.com/img/logos/airbit.svg) [Explore Success Stories](https://www.recombee.com/case-studies) --- # Drive Marketplace Growth With Real-Time Recommendations > Source: https://www.recombee.com/domains/p2p-marketplaces > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [Domains](https://www.recombee.com/where-to-use#domains) / P2P Marketplaces # Drive Marketplace Growth With Real‑Time Recommendations Connect buyers with listings they’re most likely to engage with to increase **discovery, conversions, and repeat activity**. ![P2P Marketplaces](https://www.recombee.com/img/domains/marketplaces.png) ![BeatStars](https://www.recombee.com/img/customers-domains/beatstars.svg)![PMG](https://www.recombee.com/img/customers-domains/pmg.svg) +34% **Click-rate** from newsletters +29% **Click-through rate** on suggested adverts +17% **Conversion rate** from visitor to active buyer ## Use Cases ![](https://www.recombee.com/img/use-cases/p2p-marketplaces/fully-personalized-homepage.png) ### Fully Personalized Homepage Automate and tailor all your homepage rows 1:1 for each user. ![](https://www.recombee.com/img/use-cases/p2p-marketplaces/similar-offers.png) ### Similar Offers Showcase users alternative choices for a specific offer. ![](https://www.recombee.com/img/use-cases/p2p-marketplaces/more-from-this-advertiser.png) ### More From This Advertiser Personalize the selection of the advertiser's top offers for individual users. ![](https://www.recombee.com/img/use-cases/p2p-marketplaces/faceted-search-and-category-browsing.png) ### Faceted Search & Category Browsing Show offers based on individual search criteria like category, condition, location, or any other specifics from your catalog. ![](https://www.recombee.com/img/use-cases/p2p-marketplaces/personalized-emailing.png) ### Personalized Emailing Deliver users a personalized showcase of relevant offers based on their browsing and purchase history. ## Why Recombee ### Recommend New Listings Instantly Immediately offer newly created listings to the right users thanks to real-time model training. ### Utilize User-Generated Content Understand unstructured data using cutting-edge image-processing algorithms and natural language models. ### Leverage Live Behavioral Patterns Improve quality of recommendations using real-time collaborative filtering. ### Upload Large Catalog of Listings Without Constraints Rely on horizontally scalable technology with high availability and process hundred millions of items. ### Promote Nearby Offers Within the User’s Location Utilize best-in-class geographic functions and recommend personalized ads within the shopper’s area. ### Make Sure Every Seller is Seen Balance users' attention across different offers, as they can only be bought once. ![Recombee Insights](https://www.recombee.com/img/domains/wwd-insights.png) ### Real-Time Analytics & Insights Take a deep dive into your data using our advanced real-time analytics tool called Insights within the Recombee Admin UI. * **Analyze** ads' success rate per category, area, age of the ads, and much more * **Visualize** your data, create custom reports, and seamlessly share insights with your team * Choose from a **library of analytical views** or create your graphs and reports * Using the analytical data, adjust the rules and let the recommender engine **follow your product vision** [Explore Insights](https://www.recombee.com/features/real-time-analytics-insights) "Our experience with Recombee has been exceptional. Integration was one of the most painless third-party implementations we’ve done, and the flexibility of the system allows us to continuously fine-tune discovery across our platform. Recombee’s ability to handle our catalog scale while delivering high-quality, adaptive recommendations makes it a critical part of our growth strategy." _![](https://www.recombee.com/img/customers/beatstars.png)_ **David Penner** VP Engineering at Beatstars [Explore Success Stories](https://www.recombee.com/case-studies) ## Increase Conversions and Retention by Delivering a Personalized Experience Recombee offers a unique solution that ensures the growth of your P2P marketplace. Buyers benefit from finding items specific to their needs, sellers gain better discoverability of their offers, and you also receive additional revenue from programmatic ads (such as Google AdSense) through higher CTR and boosted user loyalty. Recombee leverages an ensemble of models specifically tuned for P2P marketplaces, featuring deep learning models that process user-generated content such as images and (mostly unstructured) descriptions, as well as collaborative-filtering models tailored to work on typically very sparse interaction data (due to the uniqueness of each offer). This ensemble of models swiftly adapts to the fast-changing catalog typical of classified marketplaces, allowing freshly added ads to be immediately recommended. The recommender engine also accounts for the fact that each offer can typically be purchased only once, maximizing the total number of purchased offers by balancing them among users. Our advanced recommender gives you complete control. Using our user-friendly Admin UI, you can easily decide what gets recommended and where it appears. For example, prefer ads in the user’s area or premium offers. ### Explore more on [Product Recommendations](https://www.recombee.com/product-recommendations) [Marketplaces Integration Tips](https://docs.recombee.com/integration_tips#product-recommendations) Discover Even More Features ## Case Studies [![Segundamano](https://www.recombee.com/img/case-studies/small-covers/case-studies-segundamano.png)3x more conversionsfrom users who engage with recommendationsP2P Marketplaces](https://www.recombee.com/case-studies/segundamano) [![Pet Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-pet-media-group.png)+17%CR from visitor to active buyerP2P Marketplaces](https://www.recombee.com/case-studies/pet-media-group) [![Reliving](https://www.recombee.com/img/case-studies/small-covers/case-studies-reliving.png)+37%Place BidE-commerce + P2P Marketplaces](https://www.recombee.com/case-studies/reliving) ## Core Technology ### Adapting to Your Data A robust system that can utilize all data available to generate great recommendations for your users, including collaborative filtering and content-based models. ### Dynamically Retrained Models Real-time content personalization to suit the flourishing customer’s tastes and adaptation of fast-changing user-generated content. ### Specific Functionalities for Marketplaces Image processing to analyze items using pictures and NLP to analyze ads’ attributes from the text descriptions. ### AI-powered A/B Testing In-house AutoML AI that maximizes KPIs and continuously optimizes deep learning algorithms. ### Advanced Business Rules Boosters or filters to push forward desired listings and easy to manipulate, adjustable rules for additional optimization of your content. ### Real AI Inside Reinforcement learning and collaborative filtering to recommend personalized ads or listings based on historical on-site behavior. [Explore Features](https://www.recombee.com/features) [Explore How Recombee Works](https://www.recombee.com/how-it-works/data-and-results) "Our developers love Recombee documentation as well as quick and valuable technical support. We see Recombee’s “recommendationsToUser“ algorithm as a great option to start offering a personalized experience on our site." _![](https://www.recombee.com/img/customers/segundamano.png)_ **Marco Alvarez** Product Manager at Segundamano [Read Case Study](https://www.recombee.com/case-studies/segundamano) [Explore Success Stories](https://www.recombee.com/case-studies) --- # Boost Property Discovery With Personalized Recommendations > Source: https://www.recombee.com/domains/real-estate > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [Domains](https://www.recombee.com/where-to-use#domains) / Real Estate # Boost Property Discovery With Personalized Recommendations Surface **relevant listings instantly** based on real-time user signals and evolving preferences to reduce search fatigue and drive **higher-quality inquiries**. ![Real Estate](https://www.recombee.com/img/domains/real-estate.png) ![](https://www.recombee.com/img/customers-domains/crexi.svg)![](https://www.recombee.com/img/customers-domains/lighthouse.svg)![](https://www.recombee.com/img/customers-domains/storefront.svg)![Showmax](https://www.recombee.com/img/customers-domains/midland.svg) ## Use Cases ![](https://www.recombee.com/img/use-cases/real-estate/top-listings-for-you.png) ### Top Listings For You Let your customers discover the best properties on the market and stay up-to-date with the latest trends in real estate. ![](https://www.recombee.com/img/use-cases/real-estate/nearby-listings.png) ### Nearby Listings Expand the search with listings near the properties of interest. ![](https://www.recombee.com/img/use-cases/real-estate/more-from-this-broker.png) ### More From This Broker Recommend more listings from preferred brokers. ![](https://www.recombee.com/img/use-cases/real-estate/quick-search.png) ### Quick Search Make sure customers find the best homes and investment opportunities in their desired city or neighborhood. ![](https://www.recombee.com/img/use-cases/real-estate/watchdog-recommendations.png) ### Watchdog Recommendations Run campaigns through personalized emails with newly added listings matching individual criteria. ![](https://www.recombee.com/img/use-cases/real-estate/similar-listings.png) ### Similar Listings Present various alternatives to the searched properties. ## Why Recombee [Employ Best-In-Class Geographic Functions](#employ-best-in-class-geographic-functions) [Upturn The Number of Leads and Closed Deals](#upturn-the-number-of-leads-and-closed-deals) [Process Listings Using Real-Time Model Training](#process-listings-using-real-time-model-training) [Insights and Control](#insights-and-control) [High Availability and Scalability](#high-availability-and-scalability) ### Employ Best-In-Class Geographic Functions Utilize functions such as geo-location based search, recommendations, distances, polygons, areas, and ZIP codes. ### Upturn The Number of Leads and Closed Deals Analyze images, descriptions, and interactions with high sensitivity to important attributes (number of rooms, bedrooms, price range). ### Process Listings Using Real-Time Model Training Immediately predict what a first-time visitor is looking for by processing listings as they appear and disappear. ### Insights and Control Align your business strategy with your premium sellers by monitoring the impressions and leads they receive. ### High Availability and Scalability Rely on horizontally scalable technology with lighting-fast response times for large numbers of concurrently browsing users. ![Recombee Insights](https://www.recombee.com/img/domains/wwd-insights.png) ### Real-Time Analytics & Insights Take a deep dive into your data using our advanced real-time analytics tool called Insights within the Recombee Admin UI. * **Analyze** the performance of promoted / regular deals, user behavior per type of listing or area, and much more * **Visualize** your data, create custom reports, and seamlessly share insights with your team * Choose from a **library of analytical views** or create your graphs and reports * Using the analytical data, adjust the rules and let the recommender engine **follow your product vision** [Explore Insights](https://www.recombee.com/features/real-time-analytics-insights) ## Increase Number of Generated Leads With Real Estate Personalization With diverse customers’ tastes and flourishing offerings in real estate, it may be challenging to find the right fit for each customer. This is why Recombee concentrates on individualization of each journey. Personalization **increases customer satisfaction,** and makes your services more efficient leading to **higher amounts of broker contact and tour requests, successful viewings and properties sold.** Our deep learning recommendations show the most relevant listings to each of your visitors. Recombee doesn’t only analyze the visitor’s behavior and choices but also their geo-location preferences to be the leading recommendation engine for real estate. Our AI also analyzes property features such as year built, images or property descriptions. Recombee’s robust **recommendation engine analyzes** item properties such as **geolocation, number of rooms, floor number and amenities,** and interactions like **viewing the vacancies or likes.** Optionally, additional property information can be translated through natural language processing (NLP) as well as image processing. ### Explore more on [Specialized Recommendations](https://www.recombee.com/specialized-recommendations) [Real Estate Integration Tips](https://docs.recombee.com/integration_tips#real-estate-recommendations) Discover Even More Features ## Case Studies [![Crexi](https://www.recombee.com/img/case-studies/small-covers/case-studies-crexi.png)+40%In buy actions from the recommended propertiesReal Estate](https://www.recombee.com/case-studies/crexi) ## Core Technology ### Adapting to Your Data A robust system that can utilize all data available to generate great recommendations for your users, including collaborative filtering and content-based models. ### Dynamically Retrained Models Real-time content personalization to fit every home-seeker's tastes and adapt to constantly changing content. ### Specific Functionalities for Real Estate Filtering based on polygons in a map incorporating the user's geolocation and earth distance ReQL functions to operate in a set radius. ### AI-powered A/B Testing In-house AutoML AI that maximizes KPIs and continuously optimizes deep learning algorithms. ### Advanced Business Rules Boosters or filters to push forward desired listings and easy to manipulate, adjustable rules for additional optimization of your content. ### Real AI Inside Using Deep Learning to understand similarity of properties from images and attributes. Ensembles of real-estate tailored recommendation algorithms based on historical on-site behavior. [Explore Features](https://www.recombee.com/features) [Explore How Recombee Works](https://www.recombee.com/how-it-works/data-and-results) "Our collaboration with Recombee has supported our platform's capabilities through its intelligent personalization algorithms and sophisticated search results. Our customers now receive highly relevant property recommendations that cater to their specific needs, with one notable email campaign seeing a 178% uplift in CTOR. Their commitment to excellence is evident in the 40% increase in listing engagements on our platform, contributing to our growth in the competitive real estate market. The team at Recombee is responsive, professional, and puts in the effort to ensure that our unique business model is supported." _![](https://www.recombee.com/img/customers/crexi-1.png)_ **Larkin Magner** Director of Product Management at Crexi [Read Case Study](https://www.recombee.com/case-studies/crexi) ## Chosen Customers ![crexi](https://www.recombee.com/img/logos/crexi.svg)![saffron-stays](https://www.recombee.com/img/logos/saffron-stays.png)![zumper](https://www.recombee.com/img/logos/zumper.svg) [Explore Success Stories](https://www.recombee.com/case-studies) --- # Win Every Fan With Real-Time Sports & Live Event Recommendations > Source: https://www.recombee.com/domains/sports-and-live-events > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [Domains](https://www.recombee.com/where-to-use#domains) / Sports & Live Events # Win Every Fan With Real‑Time Sports & Live Event Recommendations Keep fans watching longer and coming back for more. Recombee’s personalization engine delivers instant, **hyper-relevant sports** and **live event content** to every viewer across games, highlights, replays, and news. ![Sports & Live Events](https://www.recombee.com/img/domains/sports-and-live-events.png) ![DAZN](https://www.recombee.com/img/customers-domains/dazn.svg)![The Sporting News](https://www.recombee.com/img/customers-domains/the-sporting-news.svg) ## Use Cases ![](https://www.recombee.com/img/use-cases/sports-and-live-events/fully-personalized-homepage.png) ### Fully Personalized Homepage Automate and tailor homepage rows 1:1 for each fan, from upcoming matches and live streams to trending replays, hot debates, and rails dedicated to their favorite leagues and teams. ![](https://www.recombee.com/img/use-cases/sports-and-live-events/upcoming-matches-events.png) ### Upcoming Matches & Events Help users plan their sports week - personalized schedules of upcoming games, streams, and events. ![](https://www.recombee.com/img/use-cases/sports-and-live-events/live-game-alerts-and-recommendations.png) ### Live Game Alerts & Recommendations Surface real-time “what to watch next” and ongoing games tailored to each viewer’s favorite leagues and teams, keeping every fan in the action as it happens. ![](https://www.recombee.com/img/use-cases/sports-and-live-events/personalized-highlight-reels-and-replays.png) ### Personalized Highlight Reels & Replays Generate personal highlight packages: goals, touchdowns, or knockout moments matched to each fan’s watch history and preferences. ![](https://www.recombee.com/img/use-cases/sports-and-live-events/powerful-sports-search.png) ### Powerful Sports Search Deliver instant, semantic search results for complex queries like “goals this weekend,” “touchdowns by \[player\],” “UFC finishes 1st round.” ![](https://www.recombee.com/img/use-cases/sports-and-live-events/trending-in-your-country.png) ### Trending in Your Country Show what’s heating up locally - from national league finals to viral post-game interviews. ![](https://www.recombee.com/img/use-cases/sports-and-live-events/fan-tailored-notifications-and-newsletters.png) ### Fan-tailored Notifications & Newsletters Guide fans back to the platform with personalized recommendations highlighting upcoming games, trending content, and updates related to their favorite teams. ![](https://www.recombee.com/img/use-cases/sports-and-live-events/related-content-and-deep-dives.png) ### Related Content & Deep Dives After the final whistle, keep them engaged with related interviews, stats breakdowns, or tactical analyses. ## Why Recombee [Real-Time Play-by-Play Personalization](#real-time-play-by-play-personalization) [Seamless Integration with Live & On-Demand](#seamless-integration-with-live-on-demand) [Support for Diverse Sports Business Models](#support-for-diverse-sports-business-models) [Smarter Editorial Control](#smarter-editorial-control) [Handle Massive Matchday Traffic & Global Tournaments](#handle-massive-matchday-traffic-global-tournaments) ### Real-Time Play-by-Play Personalization Our machine learning engine processes live interactions as they happen, instantly reshaping recommendations to reflect current momentum and user excitement. ### Seamless Integration with Live & On-Demand Connect live streams, VOD libraries, and sports archives into one continuous experience - no more silos. ### Support for Diverse Sports Business Models Whether you run subscriptions, ad-based content, PPV, or free streaming, Recombee adapts to your monetization strategy while boosting retention and watch time. ### Smarter Editorial Control Blend AI-driven recommendations with editorial priorities. Highlight premium matches, push sponsored events, or spotlight local leagues - while staying 1:1 personalized. ### Handle Massive Matchday Traffic & Global Tournaments Utilize Recombee’s scalable solution to power real-time personalization even during major live events like finals, derbies, or worldwide tournaments. ![Recombee Insights](https://www.recombee.com/img/domains/wwd-insights.png) ### Real-Time Analytics & Insights Take a deep dive into fan engagement data using our advanced analytics tool, Recombee Insights - built right into the Admin UI. * **Understand** how fans interact with live and on-demand content. * **Visualize** fan behavior and performance metrics in real time. * **Compare** engagement across leagues, matches, and highlight reels, from viewer retention and replay frequency to trending teams and events. * **Create** custom dashboards and reports to reveal what keeps audiences watching longer and where drop-offs occur. * **Choose** from a **library of analytical views,** or create your graphs and reports. * **Transform** insights into action. Use the data to fine-tune personalization rules, highlight premium matches, or boost underperforming content - ensuring the recommender engine always follows your platform strategy. [Explore Insights](https://www.recombee.com/features/real-time-analytics-insights) ## Stand Out Like the Big Leagues Recombee powers next-level sports personalization with 100+ proprietary ML models that continuously learn from each play, click, and watch pattern. Boost live engagement, replay views, and fan loyalty - all while keeping editorial and brand control. ### Explore more on [Content Recommendations](https://www.recombee.com/content-recommendations) [Video Integrations Tips](https://docs.recombee.com/integration_tips#integration-tips-content-videos) Discover Even More Features "At DAZN, being the Global Home of Sports means delivering the right matches, highlights, and moments to audiences in 200+ markets - bringing fans even closer to the live game. That’s why we’ve teamed up with Recombee to personalize experiences at scale. Their tech enables us to connect each viewer on any device with the right game or clip in real time through flexible solution built for growth. This partnership sets the pace for a smarter, more connected global sports experience." _![](https://www.recombee.com/img/customers/dazn.png)_ **Christoph Haas** EVP Product & Platform Engineering at DAZN [Explore Success Stories](https://www.recombee.com/case-studies) --- # Increase Bookings With Relevant Travel Recommendations > Source: https://www.recombee.com/domains/travel-trips > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [Domains](https://www.recombee.com/where-to-use#domains) / Travel, Trips # Increase Bookings With Relevant Travel Recommendations Turn lookers into bookers with recommendations that adapt to **traveler preferences, context, and real-time behavior** to drive **bookings and repeat stays**. ![Travel, Trips](https://www.recombee.com/img/domains/travel-trips.png) ![](https://www.recombee.com/img/customers-domains/tripadvisor.svg)![](https://www.recombee.com/img/customers-domains/cruisecritic.svg)![](https://www.recombee.com/img/customers-domains/itison.svg)![](https://www.recombee.com/img/customers-domains/mba.svg) ## Use Cases ![](https://www.recombee.com/img/use-cases/travel-trips/recommended-for-you.png) ### Recommended For You Customize travel tips by showing the most relevant destinations from a pre-selected list. ![](https://www.recombee.com/img/use-cases/travel-trips/geolocation-recommendations.png) ### Geolocation Recommendations Recommend the best country and language for your users based on their search preferences and location. ![](https://www.recombee.com/img/use-cases/travel-trips/promoted-offers-for-you.png) ### Promoted Offers For You Inspire your users with promoted offers tailored to their desires. ![](https://www.recombee.com/img/use-cases/travel-trips/faceted-search.png) ### Faceted Search Enable an efficient and flexible search option for finding the most relevant offers. ![](https://www.recombee.com/img/use-cases/travel-trips/personalized-emailing.png) ### Personalized Emailing Recommend the most relevant offers with individual preferences in mind. ![](https://www.recombee.com/img/use-cases/travel-trips/nearby-listings.png) ### Nearby Listings Expand the search with listings near the destinations of interest. ![Recombee Insights](https://www.recombee.com/img/domains/wwd-insights.png) ### Real-Time Analytics & Insights Take a deep dive into your data using our advanced real-time analytics tool called Insights within the Recombee Admin UI. * **Understand** how **users interact** with the recommendations and your product in general * Choose from a **library of analytical views** or create your graphs and reports * Using the analytical data, adjust the rules and let the recommender engine **follow your product vision** [Explore Insights](https://www.recombee.com/features/real-time-analytics-insights) ## Turn “Lookers-Into-Bookers” by Offering Hyper-Personalization to Each Visitor **Tailored services** and **personal approach** increases the number of loyal customers by delivering the best travel tips and strengthens your customer base. With the engines’ ability to recognize travel preferences and with filters added by your customers, we can offer customized travel **recommendations to each of your travelers.** **Increase your conversion rates** and **customer satisfaction** with AI. It is an uneasy task to gather data since many users travel rarely. One of the advantages of Recombee’s solution is the usage of **collaborative filtering** models that from the first click, within milliseconds, recognize your user’s preferences, and provide real-time recommendations - even to unknown or first-times visitors. Recombee’s robust recommendation engine analyzes item properties such as **category, place, availability** or **price,** and interactions like **trip detail view, bookmarks,** or **purchases.** Keep your customers engaged and loyal with **personalized homepage, emailing, internal search** or **push notifications,** adapting to any language or style. ### Explore more on [Specialized Recommendations](https://www.recombee.com/specialized-recommendations) Discover Even More Features ## Core Technology ### Adapting to Your Data A robust system that can utilize all data available to generate great recommendations for your users, including collaborative filtering and content-based models. ### Dynamically Retrained Models Real-time content personalization to meet the flourishing customer’s tastes and adaptation to changing traveling options. ### Specific Functionalities for Travel & Trips NLP recognizing texts in 80+ languages to analyze trip attributes for an international clientele. ### AI-powered A/B Testing In-house AutoML AI that maximizes KPIs and continuously optimizes deep learning algorithms. ### Advanced Business Rules Boosters or filters to push forward desired travel agencies and easy to manipulate, adjustable rules for additional optimization of your services. ### Real AI Inside Modern collaborative filtering and other methods recommending desired trip options based on historical on-site behavior. [Explore Features](https://www.recombee.com/features) [Explore How Recombee Works](https://www.recombee.com/how-it-works/data-and-results) ## Chosen Customers ![tripadvisor](https://www.recombee.com/img/logos/tripadvisor.svg)![mba](https://www.recombee.com/img/logos/mba.svg) [Explore Success Stories](https://www.recombee.com/case-studies) --- # Maximize Engagement With Real-Time Video Recommendations > Source: https://www.recombee.com/domains/video > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [Domains](https://www.recombee.com/where-to-use#domains) / Video # Maximize Engagement With Real‑Time Video Recommendations Surface the right content at the right moment with recommendations that **adapt in milliseconds**. Increase **watch time, completion, subscriptions, and retention** across every device. ![Video](https://www.recombee.com/img/domains/video.png) ![DAZN](https://www.recombee.com/img/customers-domains/dazn.svg)![PBS](https://www.recombee.com/img/customers-domains/pbs.svg)![FTV Prima](https://www.recombee.com/img/customers-domains/ftv-prima.svg)![Modern TV](https://www.recombee.com/img/customers-domains/modern-tv.svg)![Pathe Thuis](https://www.recombee.com/img/customers-domains/pathe-thuis.svg)![TVA+](https://www.recombee.com/img/customers-domains/tva-plus.svg)![Illico+](https://www.recombee.com/img/customers-domains/illico-plus.svg)![Gaia](https://www.recombee.com/img/customers-domains/gaia.svg) ## Use Cases ![](https://www.recombee.com/img/use-cases/video/fully-personalized-homepage.png) ### Fully Personalized Homepage Automate and tailor all your homepage rows 1:1 for each user. [Explore Recipe](https://docs.recombee.com/recipes/video#fully-personalized-homepage) ![](https://www.recombee.com/img/use-cases/video/short-videos-feed.png) ### Short Videos Feed Keep viewers engaged with an endless short-form video feed that adapts instantly to what they watch, skip, like, or dislike. [Explore Recipe](https://docs.recombee.com/recipes/video/feed/swiping-feed) ![](https://www.recombee.com/img/use-cases/video/editorial-picks-for-you.png) ### Editors' Picks For You Personalize editors' picks by showing the most relevant titles from a pre-selected list. [Explore Recipe](https://docs.recombee.com/recipes/video/fully-personalized-homepage/editors-picks-for-you) ![](https://www.recombee.com/img/use-cases/video/search-movies-series.png) ### Search Movies/Series Make sure the desired titles get found within a few keystrokes with a full-text search for movies and TV shows. [Explore Recipe](https://docs.recombee.com/recipes/video/search-movies-and-series) ![](https://www.recombee.com/img/use-cases/video/personalized-emailing.png) ### Personalized Emailing Sweeten the user's experience through personalized emails with movies and series tailored to the viewer's taste. [Explore Recipe](https://docs.recombee.com/recipes/video/personalized-emailing) ![](https://www.recombee.com/img/use-cases/video/because-you-watched.png) ### Because You Watched Show highly relevant titles based on what the user recently watched. Recombee’s API natively supports returning the watched title and its related content. [Explore Recipe](https://docs.recombee.com/recipes/video/fully-personalized-homepage/because-you-watched) ![](https://www.recombee.com/img/use-cases/video/watch-next.png) ### Watch Next Motivate users to stay in the binge-watching zone using end-of-playback recommendations. [Explore Recipe](https://docs.recombee.com/recipes/video/asset-detail-and-player/watch-next) ![](https://www.recombee.com/img/use-cases/video/top-genres-for-you.png) ### Top Genres for You Show personalized genres to each user based on their watching history. ![](https://www.recombee.com/img/use-cases/video/related-movies-series.png) ### Related Movies/Series Satisfy your users' desire for more content based on titles they already enjoyed. [Explore Recipe](https://docs.recombee.com/recipes/video/asset-detail-and-player/more-like-this) ![](https://www.recombee.com/img/use-cases/video/continue-watching.png) ### Continue Watching Serve the next episode of a recently watched series or resume recently unfinished movies. [Explore Recipe](https://docs.recombee.com/recipes/video/fully-personalized-homepage/continue-watching) ![](https://www.recombee.com/img/use-cases/video/new-releases.png) ### New Releases Surprise users with personalized picks for the freshest releases. ![](https://www.recombee.com/img/use-cases/video/last-chance.png) ### Last Chance Encourage watching of expiring titles before they're gone for good. ![](https://www.recombee.com/img/use-cases/video/trending-in-your-country.png) ### Trending In Your Country Showcase the diversity of your catalog with the hottest local content in your users' region. ### Video Recipes Discover how to personalize various use cases within your video platform. [Explore Integration Recipes](https://docs.recombee.com/recipes/video) "Recombee helped us make content discovery far more relevant for every viewer. Following a seamless implementation, we saw a 23% increase in click-through rate and a 39% increase in video views reaching at least 25% completion. The platform’s configuration flexibility and the team’s hands-on support made it easy to tailor recommendations to our specific needs." _![](https://www.recombee.com/img/customers/gaia.png)_ **Michal Lebowitsch** SVP International & Content Operations at Gaia [Explore Success Stories](https://www.recombee.com/case-studies) ## Why Recombee [Netflix-Style Homepage](#set-up-your-homepage-with-netflix-like-rows) [Series, Season & Episode Hierarchy](#recommendations-of-series-seasons-and-episodes) [Supports SVOD, AVOD & TVOD](#support-for-svod-avod-and-tvod-business-models) [Personalized OTT Viewing](#deliver-personalized-experiences-on-all-ott-platforms) [Connect Linear TV, Live Streams & VOD](#connect-linear-tv-live-streams-and-vod-libraries) [Real-Time Video Suggestions](#recommend-video-content-in-real-time) ### Set-up Your Homepage With Netflix-Like Rows Personalize all rows on your homepage using rich mixture of algorithms and advanced configuration options. ### Recommendations of Series, Seasons and Episodes Utilize video-specific events, such as “progress of watching” or “watching milestones” and algorithms supporting content hierarchy and sequentiality. ### Support for SVOD, AVOD and TVOD Business Models Improve subscription and retention rates, ad and transaction revenues, watch time, number of plays, and other metrics. ### Deliver Personalized Experiences on All OTT Platforms Provide recommendations to a full scale of supported devices ranging from mobile, smart TV, desktop, HbbTV, and others. ### Connect Linear TV, Live Streams, and VOD Libraries Use data from multiple sources and connect your devices with cross-platform recommendations. ### Recommend Video Content in Real-Time Process stream of live events using machine learning models to learn immediate user preferences and recommend content which is hot at the very moment. ![Recombee Insights](https://www.recombee.com/img/domains/wwd-insights.png) ### Real-Time Analytics & Insights Take a deep dive into your data using our advanced real-time analytics tool called Insights within the Recombee Admin UI. * **Understand** your top-recommended assets, and compare views of free and premium content from recommendations, or genres popularity * **Visualize** your data, create custom reports, and seamlessly share insights with your team * Choose from a **library of analytical views** or create your graphs and reports * Using the analytical data, adjust the rules and let the recommender engine **follow your product vision** [Explore Insights](https://www.recombee.com/features/real-time-analytics-insights) ## Offer YouTube and Netflix-Like Recommendations Based on analysis of video attributes and user behavior suggest the next best title on any platform - be it **VOD (video on demand), SVOD** or **AVOD.** Invest in your user experience with **video recommendations** and don’t let the result go unrewarded - **add to your watch time, increase views on videos** and gather positive reviews of your platform! With the growing number of video streaming platforms, it is crucial to create unique experiences for each of your users to prevent churn. Our recommender engine **recognizes the preferences of users** in **real-time.** We work with information about which titles were watched till the end, which halfway or skipped completely. Recombee offers application of boosters and filters, that allow for **automatic organization of content,** e.g. boosting premium videos or chosen genres. Utilize Recombee to offer **“watch next”** scenarios, **homepage, full-text search** or **infinite scroll personalization.** Recombee’s robust recommendation engine analyzes video properties such as **title, genre, description, year, language, available since/ expiration date** or **cast** and user interactions like **detail views, view portions, ratings,** or **purchases.** User attributes such as **age, language, location,** or **type** of subscription are also incorporated. ### Explore more on [Content Recommendations](https://www.recombee.com/content-recommendations) [Video Integrations Tips](https://docs.recombee.com/integration_tips#integration-tips-content-videos) Discover Even More Features ## Case Studies [![Showmax](https://www.recombee.com/img/case-studies/small-covers/case-studies-showmax.png)70Countries and multiple languagesSVOD Service](https://www.recombee.com/case-studies/showmax) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima.png)+34%Video views on VOD platformMedia Company](https://www.recombee.com/case-studies/ftv-prima) ## Core Technology ### Adapting to Your Data Content-based algorithms using NLP to analyze all video attributes for different platforms, VOD, SVOD, or TVOD (or others). ### Dynamically Retrained Models Real-time content personalization to adapt to the flourishing customer’s tastes and consider the fast-changing video content. ### Specific Functionalities for Video Platforms Recommendations taking into account the users’ watch time; which titles were watched till the end, which halfway or skipped completely. ### AI-powered A/B Testing In-house AutoML AI that maximizes KPIs and continuously optimizes deep learning algorithms. ### Advanced Business Rules Boosters or filters to push forward desired videos and easy to manipulate, adjustable rules for additional optimization of your content. ### Real AI Inside Recommender system designed to recognize the preferences of individual users and predict relevant videos faster and with higher accuracy. [Explore Features](https://www.recombee.com/features) [Explore How Recombee Works](https://www.recombee.com/how-it-works/data-and-results) "At DAZN, being the Global Home of Sports means delivering the right matches, highlights, and moments to audiences in 200+ markets - bringing fans even closer to the live game. That’s why we’ve teamed up with Recombee to personalize experiences at scale. Their tech enables us to connect each viewer on any device with the right game or clip in real time through flexible solution built for growth. This partnership sets the pace for a smarter, more connected global sports experience." _![](https://www.recombee.com/img/customers/dazn.png)_ **Christoph Haas** EVP Product & Platform Engineering at DAZN [Explore Success Stories](https://www.recombee.com/case-studies) --- # Unlocking 15% More Actor Runs for the World’s Largest Marketplace of Tools for AI > Source: https://www.recombee.com/case-studies/apify > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # [Success Stories](https://www.recombee.com/case-studies) # Unlocking 15% More Actor Runs for the World’s Largest Marketplace of Tools for AI Apify ![](https://www.recombee.com/img/case-studies/apify.png) Apify is building the world's largest marketplace of **tools for AI, known as Actors**, powering the next generation of **AI agents and autonomous AI applications**. Developers and businesses use Actors to discover, combine, and automate **ready-to-run cloud applications** for web scraping, browser automation, data extraction, and **agentic coding workflows**. With **more than 50,000 Actors** available and thousands of new ones added every month, helping users quickly find the right Actor has become a critical part of the platform experience. As the marketplace expanded, Apify needed search and recommendation capabilities that could scale with its **rapidly growing catalog** while delivering personalized discovery from the very first interaction. +15% Actor Runs (existing users) Personalized Search +6% Subscriptions (new users) Personalized Search +9% CTR on First Four Search Results (existing users) Personalized Search +12% Overall CTR Actors For You ### Situation Apify's marketplace was growing rapidly, expanding from thousands to **tens of thousands of Actors**. Existing discovery relied primarily on **keyword search**, making it increasingly difficult to surface the most relevant Actors as the catalog expanded. At the same time: * Search infrastructure **limited personalization** * Dynamic ranking at scale became **increasingly difficult to maintain manually** * Recommendation experiences were missing entirely * The team needed a **solution that continuously optimized recommendations without manual intervention**, while still allowing **full control over business rules and personalization strategies** ![](https://www.recombee.com/img/case-studies/apify-situation.png) ### Objectives * Deliver **personalized search results** at marketplace scale * Introduce **recommendations across key discovery journeys** * Increase Actor usage and subscriptions * **Reduce engineering effort** required to optimize ranking * Support rapid marketplace growth without sacrificing relevance ### Solution **Personalized Discovery at Scale** Recombee combines behavioral signals, real-time interactions, and contextual relevance to personalize search results and recommendations across the marketplace. As users explore more Actors, personalized suggestions continuously improve, making discovery more relevant with every interaction. **Search and Recommendations in One Platform** Apify uses a single platform to power both personalized search and recommendations. This enables a consistent discovery experience across search, homepage recommendations, similar Actors, and other touchpoints while simplifying implementation, configuration, and ongoing optimization. **Flexible Configuration Without Heavy Engineering** Using Recombee's Scenario Settings, Apify can independently configure ranking logic, filters, boosters, and recommendation strategies for different search and discovery experiences. This gives the team the flexibility to experiment, optimize, and adapt individual user journeys as the marketplace evolves without extensive engineering effort. **A Collaborative Partnership** Beyond the technology, Apify valued Recombee's hands-on collaboration throughout implementation. Direct access to the engineering and support teams enabled rapid iteration, quick problem solving, and continuous optimization as new discovery challenges emerged. ### Benefits & Results Personalised Search #### New Users **+3%** Actor Runs **+6%** Subscriptions **+7.5%** CTR on First Four Results **+4.5%** Overall CTR #### Existing Users **+15%** Actor Runs **+9%** CTR on First Four Results **+5%** Overall CTR Similar Actors #### New Users **+8%** Actor Runs #### Existing Users **+6%** Actor Runs Actors For You **+12%** Overall CTR \* Results were measured separately for **new users** (with limited behavioral data) and **existing users** (with established interaction history). The primary KPI, **Actor Runs**, measures how often users execute an Actor, making it a key indicator of successful discovery and engagement. ### Recombee’s Solution in Action ![apify](https://www.recombee.com/img/case-studies/scenarios/apify-1.png) #### Search Personalization Personalized Search Instead of returning identical search results for every user, Recombee personalizes ranking using **behavioral signals, user preferences, and real-time interactions**. As users continue interacting with the marketplace, search results become increasingly relevant while still respecting textual relevance and business rules. **Key Benefits** * **Personalized homepage** discovery * **Faster exploration** of relevant Actors * **Better engagement** with recommended content * **Higher** click-through rates ![apify](https://www.recombee.com/img/case-studies/scenarios/apify-2.png) #### Similar Actors You Might Also Like When users open an Actor detail page, Recombee recommends relevant alternatives and complementary Actors based on **behavioral similarity and content understanding**. This helps users quickly discover **better-fitting solutions** while **increasing exploration** across the marketplace. **Key Benefits** * **Improve** Actor discovery * **Surface** relevant alternatives * **Increase** Actor usage * **Reduce** dead ends in user journeys ![apify](https://www.recombee.com/img/case-studies/scenarios/apify-3.png) #### Dashboard Personalization Actors For You The personalized homepage recommends Actors based on **user interests, previous interactions, and evolving usage patterns**. **Recommendations adapt continuously** as users explore new Actors, creating a tailored marketplace experience for every visitor. **Key Benefits** * **Personalized homepage** discovery * **Faster exploration** of relevant Actors * **Better engagement** with recommended content * **Higher** click-through rates “As our marketplace grew to tens of thousands of Actors, delivering relevant discovery became increasingly challenging. With many Actors matching the same search queries, we needed more than keyword-based search. We needed smart personalization that could scale. Recombee helped us roll out personalized search and recommendations across key discovery journeys, and we quickly saw measurable improvements. Among existing users, Actor Runs increased by 15% and click-through rate on the top four search results increased by 9%, along with a 6% uplift in subscriptions from new users. Beyond the technology, what really stood out was the collaboration. The Recombee team genuinely cared about helping us solve difficult problems, worked closely with us throughout implementation, and was always willing to explore solutions together.” ![Jan Kuzelik](https://www.recombee.com/img/case-studies/testimonials/apify.png) **Jan Kuzelik** Product Manager ![Apify](https://www.recombee.com/img/logos/apify.svg) ### About Apify Apify is building the world's largest marketplace of **tools for AI, known as Actors**, powering the next generation of **AI agents and autonomous AI applications**. Developers and businesses use Actors to discover, combine, and automate **ready-to-run cloud applications** for web scraping, browser automation, data extraction, and**agentic coding workflows**. [Visit Apify](https://apify.com) **Next Case Studies** [![The Telegraph](https://www.recombee.com/img/case-studies/small-covers/case-studies-the-telegraph.png)+35%CTR uplift on article detail pagesMedia Company](https://www.recombee.com/case-studies/the-telegraph) [![Diagnal](https://www.recombee.com/img/case-studies/small-covers/case-studies-diagnal.png)+35%Playbacks Across Video PlatformsVideo Streaming](https://www.recombee.com/case-studies/diagnal) [![9GAG](https://www.recombee.com/img/case-studies/small-covers/case-studies-9gag.png)+37%Post ViewsCross-Platform Entertainment Network](https://www.recombee.com/case-studies/9gag) [![Poslovi Infostud](https://www.recombee.com/img/case-studies/small-covers/case-studies-poslovi-infostud.png)+32%Application conversion on job listingsOnline Job Board](https://www.recombee.com/case-studies/poslovi-infostud) [![Perlego](https://www.recombee.com/img/case-studies/small-covers/case-studies-perlego.png)+44%CTRDigital Learning](https://www.recombee.com/case-studies/perlego) [![Slickdeals](https://www.recombee.com/img/case-studies/small-covers/case-studies-slickdeals.png)+70%CTR to Detail Page ViewsE-commerce + Deal Aggregators](https://www.recombee.com/case-studies/slickdeals) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima.png)+34%Video views on VOD platformMedia Company](https://www.recombee.com/case-studies/ftv-prima) [![Audiomack](https://www.recombee.com/img/case-studies/small-covers/case-studies-audiomack.png)+206%Monthly plays from recommendationsMusic](https://www.recombee.com/case-studies/audiomack) [![Pet Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-pet-media-group.png)+17%CR from visitor to active buyerP2P Marketplaces](https://www.recombee.com/case-studies/pet-media-group) [![Pepper](https://www.recombee.com/img/case-studies/small-covers/case-studies-pepper.png)+21%Click-outs to affiliate links from websiteDeal Aggregators](https://www.recombee.com/case-studies/pepper) [![Crexi](https://www.recombee.com/img/case-studies/small-covers/case-studies-crexi.png)+40%In buy actions from the recommended propertiesReal Estate](https://www.recombee.com/case-studies/crexi) [![Triola](https://www.recombee.com/img/case-studies/small-covers/case-studies-triola.png)10%Total orders from recommendationsE-commerce](https://www.recombee.com/case-studies/triola) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima-content.png)+10%RecirculationMedia Company](https://www.recombee.com/case-studies/ftv-prima-content) [![Unfiltered Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-unfiltered-media-group.png)+50%CTRMedia Company](https://www.recombee.com/case-studies/unfiltered-media-group) [![Reliving](https://www.recombee.com/img/case-studies/small-covers/case-studies-reliving.png)+37%Place BidE-commerce + P2P Marketplaces](https://www.recombee.com/case-studies/reliving) [![Autohaus Kunzmann](https://www.recombee.com/img/case-studies/small-covers/case-studies-autohaus-kunzmann.png)+14%Conversion RateE-commerce](https://www.recombee.com/case-studies/autohaus-kunzmann) [![Showmax](https://www.recombee.com/img/case-studies/small-covers/case-studies-showmax.png)70Countries and multiple languagesSVOD Service](https://www.recombee.com/case-studies/showmax) [![Mafra](https://www.recombee.com/img/case-studies/small-covers/case-studies-mafra.png)40% higherCTR of suggested articlesMedia Company](https://www.recombee.com/case-studies/mafra) [![itison](https://www.recombee.com/img/case-studies/small-covers/case-studies-itison.png)2 000%ROINewsletter Personalization + Deal Aggregators](https://www.recombee.com/case-studies/itison) [![Cooklist](https://www.recombee.com/img/case-studies/small-covers/case-studies-cooklist.png)+27%CTRE-commerce](https://www.recombee.com/case-studies/cooklist) [![Economia](https://www.recombee.com/img/case-studies/small-covers/case-studies-economia.png)+64%Click Through RateMedia Company](https://www.recombee.com/case-studies/economia) [![Segundamano](https://www.recombee.com/img/case-studies/small-covers/case-studies-segundamano.png)3x more conversionsfrom users who engage with recommendationsP2P Marketplaces](https://www.recombee.com/case-studies/segundamano) [![Design Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-design-group.png)+52%Number of PurchasesE-commerce](https://www.recombee.com/case-studies/design-group) --- # Personalized Streaming Experience Tailored 1:1 in Real-Time > Source: https://www.recombee.com/case-studies/audiomack > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [Success Stories](https://www.recombee.com/case-studies) # One-On-One Content Recommendations for the Global Music Sharing and Discovery Platform Audiomack [Music](https://www.recombee.com/domains/music-podcasts) [Content Recommendations](https://www.recombee.com/content-recommendations) ![](https://www.recombee.com/img/case-studies/audiomack.png) As an artist-first global music-sharing platform, Audiomack is here to move music forward. The **on-demand global music discovery platform operating in more than 200 countries** enables creators to share unlimited songs, albums, mixtapes, playlists, and podcasts and **allows listeners to stream that content for free.** Thanks to Recombee's AI content recommendations, Audiomack provides its **20 million monthly active users** and 5 million daily active users with a **personalized streaming experience** tailored one-on-one in real-time while significantly increasing the number of monthly follows and plays. 206% Increase in Monthly Plays from recommendations 67% Increase in Weekly Account Follows from recommendations ### Situation * Millions of listeners and artists from different parts of the world with different taste * Millions of permanently accessible songs and other content in the constantly growing catalog * In-house system for recommending globally popular content ### Requirements * State-of-the-art recommender system enriching user’s experience through personalized music discovery * High model sensitivity and reactivity to the user’s latest interactions * Unique recommendations based on the user’s geolocation ### Solution Development of **custom algorithms and models** that perfectly suit Audiomack’s product vision and expectations of how the recommendations should behave. Real-time data processing for near **real-time deep learning models.** Delivery of recommendations **in 50ms on average.** **Geolocation-based recommendations to reflect specifics of 100+ cultural regions.** **Leveraging hierarchical relationships between songs and artists through Item Segmentations.** Diverse and complex ensemble of models constantly optimized by artificial intelligence. Collaborative Filtering Content-Based * Cold-start recommendation for onboarding users based on provided demographical data and explicit preferences. Reinforcement Learning (Contextual Bandits) ![audiomack](https://www.recombee.com/img/case-studies/scenarios/audiomack-1.png) ### Item Segments Item segmentation is an **abstraction on top of the catalog of items** that allows you to **group items into segments based on their properties** which can be then recommended to your users. Audiomack was our very first customer who had a chance to properly test this revolutionary feature and measured **exciting performance uplift** in the artist following. There were **two use cases** in the app to start with: **recommending artists** based on your music taste and **similar artists** based on the ones you are already following. Rolling out this functionality was a **huge success** and helped increase registered **users’ engagement and satisfaction** with the platform. Not only did it account for **10% of the total “follows”,** but if a user clicks follow, there is a 9% chance they immediately follow another artist. If they follow one of the recommendations, there's a **75% chance they follow at least one more.** ![audiomack](https://www.recombee.com/img/case-studies/scenarios/audiomack-2.png) ### Cold Start User Onboarding One of the most known obstacles recommender systems need to **overcome** at the beginning is the **“cold start problem”.** When there is a **newly added item or visitor** who came to the platform **for the first time,** there is essentially **no historical data** to work with so the models have to be a little creative in finding relevant items to recommend. At Recombee, there are **many ways we are reducing this problem** whereby coming up with a new approach of **advanced cold start user onboarding.** **When registering on a platform, there is a set of questions being asked for explicit user preferences about e.g. their geolocation or genre taste.** This allows our **unique ensemble of models to provide relevant offerings from the very beginning,** making the recommendations more and more precise in real time after every new interaction and helping the **newly onboarded users enjoy and bond with the platform.** ### Benefits & Results * **206% increase** in “Monthly Plays” from recommendations * **67% increase** in “Weekly Follows” from recommended accounts * **Best performing module** within the Discovery tab, accounting for **46% of all plays** “Striving to be the ever limitless music sharing and discovery platform, we need to make sure the user experience of our listeners is smooth and sound. And one of the most critical aspects of achieving such a goal is content personalization tailored 1:1 in real-time. That's why we switched to Recombee. Thanks to their recommender engine, our **monthly plays increased by 206%** and **weekly follows by 67%.** Because the recommendations performed so well, we moved them from our Search page to the top of our main Discover tab. They are now **the best-performing module within that tab, accounting for 46% of all plays.”** ![Christopher Dalla Riva](https://www.recombee.com/img/case-studies/testimonials/audiomack-2.png) **Christopher Dalla Riva** Senior Product Manager at Audiomack ### Scenarios ![audiomack](https://www.recombee.com/img/case-studies/scenarios/audiomack-3.png) #### “Recommended For You” Initially tested on the search page, the performance was so good that it was swiftly placed on a more visible place in the platform. Now, when you open the Audiomack mobile app, the first thing you will see on the home screen is the Discover section with multiple rows of personalized songs powered fully by Recombee. Those songs are carefully picked by our models to provide you with the most relevant offering. ![audiomack](https://www.recombee.com/img/case-studies/scenarios/audiomack-4.png) #### “Similar Songs” State-of-the-art ensemble of models used across many different places on the platform to extend the users listening time and provide relevant similar songs. * General Song Recommendations * Queue End Recommendations * Radio Recommendations The mentioned use cases are highly reactive to the user’s most recent plays and can be flexibly adjusted to e.g. show specific numbers of songs from the same artist/same album. ![audiomack](https://www.recombee.com/img/case-studies/scenarios/audiomack-5.png) #### “Accounts For You” A unique row of recommended artists based on your personal taste provided on the home screen of the Audiomack mobile app. Rolling out this feature significantly increased the total number of accounts following and thus enabled the user to discover creators and artists they would have otherwise had a hard time finding. There is also a great chance that if the user starts following one of the recommended accounts, he will immediately follow another one hence enriching its content portfolio. ![audiomack](https://www.recombee.com/img/case-studies/scenarios/audiomack-6.png) #### “Fans Also Like” This unique scenario works very well together with the previously mentioned one helping you find the artists whose work resonates with you the most. You will get a list of similar accounts right after you start following a new one, giving you the opportunity to relevantly broaden your musical palette. **“Working with Recombee** to develop an affordable solution to provide our users with excellent music recommendations has **exceeded our expectations in every way.** They have been able to understand the relationships in the data of our industry and create **effective models to use efficiently.** We also appreciate the **nuance and flexibility** they offer when it comes to deciding the right solution based on quality, cost, complexity, speed, and other factors.” ![Ty Wangsness](https://www.recombee.com/img/case-studies/testimonials/audiomack-1.png) **Ty Wangsness** Founder/CTO at Audiomack ![Audiomack](https://www.recombee.com/img/logos/audiomack.svg) ### About Audiomack Audiomack allows creators to share their content with millions of **highly engaged listeners.** Millions of fans use the platform daily to discover buzzing new songs and the hottest trending music anywhere. Being one the **most innovative companies in music,** Audiomack also enables fans to communicate with their favorite artists and support them directly through the platform. Simply put, Audiomack is an **open creative space for artists** who don't like limitations. [Visit Audiomack](https://www.audiomack.com) **Next Case Studies** [![The Telegraph](https://www.recombee.com/img/case-studies/small-covers/case-studies-the-telegraph.png)+35%CTR uplift on article detail pagesMedia Company](https://www.recombee.com/case-studies/the-telegraph) [![Apify](https://www.recombee.com/img/case-studies/small-covers/case-studies-apify.png)+6%SubscriptionsAI Marketplace](https://www.recombee.com/case-studies/apify) [![Diagnal](https://www.recombee.com/img/case-studies/small-covers/case-studies-diagnal.png)+35%Playbacks Across Video PlatformsVideo Streaming](https://www.recombee.com/case-studies/diagnal) [![9GAG](https://www.recombee.com/img/case-studies/small-covers/case-studies-9gag.png)+37%Post ViewsCross-Platform Entertainment Network](https://www.recombee.com/case-studies/9gag) [![Poslovi Infostud](https://www.recombee.com/img/case-studies/small-covers/case-studies-poslovi-infostud.png)+32%Application conversion on job listingsOnline Job Board](https://www.recombee.com/case-studies/poslovi-infostud) [![Perlego](https://www.recombee.com/img/case-studies/small-covers/case-studies-perlego.png)+44%CTRDigital Learning](https://www.recombee.com/case-studies/perlego) [![Slickdeals](https://www.recombee.com/img/case-studies/small-covers/case-studies-slickdeals.png)+70%CTR to Detail Page ViewsE-commerce + Deal Aggregators](https://www.recombee.com/case-studies/slickdeals) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima.png)+34%Video views on VOD platformMedia Company](https://www.recombee.com/case-studies/ftv-prima) [![Pet Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-pet-media-group.png)+17%CR from visitor to active buyerP2P Marketplaces](https://www.recombee.com/case-studies/pet-media-group) [![Pepper](https://www.recombee.com/img/case-studies/small-covers/case-studies-pepper.png)+21%Click-outs to affiliate links from websiteDeal Aggregators](https://www.recombee.com/case-studies/pepper) [![Crexi](https://www.recombee.com/img/case-studies/small-covers/case-studies-crexi.png)+40%In buy actions from the recommended propertiesReal Estate](https://www.recombee.com/case-studies/crexi) [![Triola](https://www.recombee.com/img/case-studies/small-covers/case-studies-triola.png)10%Total orders from recommendationsE-commerce](https://www.recombee.com/case-studies/triola) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima-content.png)+10%RecirculationMedia Company](https://www.recombee.com/case-studies/ftv-prima-content) [![Unfiltered Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-unfiltered-media-group.png)+50%CTRMedia Company](https://www.recombee.com/case-studies/unfiltered-media-group) [![Reliving](https://www.recombee.com/img/case-studies/small-covers/case-studies-reliving.png)+37%Place BidE-commerce + P2P Marketplaces](https://www.recombee.com/case-studies/reliving) [![Autohaus Kunzmann](https://www.recombee.com/img/case-studies/small-covers/case-studies-autohaus-kunzmann.png)+14%Conversion RateE-commerce](https://www.recombee.com/case-studies/autohaus-kunzmann) [![Showmax](https://www.recombee.com/img/case-studies/small-covers/case-studies-showmax.png)70Countries and multiple languagesSVOD Service](https://www.recombee.com/case-studies/showmax) [![Mafra](https://www.recombee.com/img/case-studies/small-covers/case-studies-mafra.png)40% higherCTR of suggested articlesMedia Company](https://www.recombee.com/case-studies/mafra) [![itison](https://www.recombee.com/img/case-studies/small-covers/case-studies-itison.png)2 000%ROINewsletter Personalization + Deal Aggregators](https://www.recombee.com/case-studies/itison) [![Cooklist](https://www.recombee.com/img/case-studies/small-covers/case-studies-cooklist.png)+27%CTRE-commerce](https://www.recombee.com/case-studies/cooklist) [![Economia](https://www.recombee.com/img/case-studies/small-covers/case-studies-economia.png)+64%Click Through RateMedia Company](https://www.recombee.com/case-studies/economia) [![Segundamano](https://www.recombee.com/img/case-studies/small-covers/case-studies-segundamano.png)3x more conversionsfrom users who engage with recommendationsP2P Marketplaces](https://www.recombee.com/case-studies/segundamano) [![Design Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-design-group.png)+52%Number of PurchasesE-commerce](https://www.recombee.com/case-studies/design-group) --- # Increasing Click-Outs by 21% for Pepper | Case Study > Source: https://www.recombee.com/case-studies/pepper > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [Success Stories](https://www.recombee.com/case-studies) # Increasing Click-Outs by 21% for Pepper - The World's Largest Shopping Community Pepper [Deal Aggregators](https://www.recombee.com/domains/deal-aggregators) [Product Recommendations](https://www.recombee.com/product-recommendations) ![](https://www.recombee.com/img/case-studies/pepper.png) **Pepper is the world's largest social shopping community with more than 25 million shoppers** sharing deals and tips. To provide its shoppers with the most relevant and diverse deals, Pepper has focused on continuously improving its product recommendations. The integration of Recombee's AI-driven product recommendations successfully established a personalized user experience in real-time across multiple platforms (iOS, Android, and web) in different countries. This enhancement resulted in a **significant increase in click-outs and user satisfaction,** outperforming the existing in-house recommendation engine. +21% **Click-outs** to affiliate links from website +6% **Click-outs** to affiliate links from mobile apps ### Situation * Advanced in-house recommendation solution * Over 25 million shoppers and 500 million page views per month * 10 countries and a rapidly changing inventory * Mobile application available for iOS and Android ### Objectives * Increase click-outs to affiliate links while keeping shoppers happy with recommendations * Easily manage and configure recommendations in each country from one place * Real-time personalized home page feed for every user * A solution that can replace the current Home Infinite Feed under high traffic and minimal response time * Provide shoppers with the most relevant products while also ensuring they have the opportunity to discover additional items within the platform * The ability to promote the offers of specific affiliate partners ### Solution Personalized recommendation section **“For you”** on the home page. Smooth shopping experience with **infinite scrolling recommendations.** **Cross-device recommendations** (iOS, Android, website). A complex and diverse ensemble of incrementally trained recommendation models that **help shoppers discover new and relevant offers:** * Collaborative filtering models * Popularity-based models * Reinforcement learning and contextual bandit models ![pepper](https://www.recombee.com/img/case-studies/scenarios/pepper-1.png) ### Personalized Infinite Scroll With a fast-changing inventory and user-generated content, Pepper used Recombee's **Infinite Scroll to create an endless feed of offers,** which is especially important for a mobile application. As users reach the bottom of the page, new personalized offers are automatically loaded in real-time, **eliminating the need to click through to different pages or even churn from the app.** ### Benefits & Results * **+21%** in click-outs to affiliate links from the website. * **+6%** in click-outs to affiliate links from mobile apps. * **Significantly improved user experience based on the feedback from users.** * **Real-time personalization** for each individual user. * **One place for easy management and configuration** of recommendations across all Pepper platforms. ### Scenarios ![pepper](https://www.recombee.com/img/case-studies/scenarios/pepper-2.png) #### For you As a marketplace promoting 3rd party products, Pepper is heavily focused on providing its users with relevant offers via a personalized home page section **tailored to their individual preferences based on what the user purchased or clicked on previously.** In this scenario, a combination of **collaborative filtering models with popularity-based models and contextual bandit models** is used to show the users the hottest and most relevant offers every time they open the app or visit the website. “It is a real pleasure collaborating with Recombee. Their problem-solving skills have proven invaluable, helping us overcome various business challenges while allowing us to consistently increase our click-outs and deliver a better user experience. Thanks to their solution we've seen our click-outs increase by up to 21%. They have become a trusted partner I can highly recommend.“ ![Heike Guertler](https://www.recombee.com/img/case-studies/testimonials/pepper.png) **Heike Guertler** Head of Product at Pepper ![Pepper](https://www.recombee.com/img/logos/pepper.svg) ### About Pepper Pepper is the world's largest social commerce site with the largest shopping community of over **25 million shoppers per month, 500 million page views and 12,000 purchase decisions per minute.** The company operates in 10 countries across the Americas and Europe. It consists of several market leading platforms including Hotukdeals in the UK, Dealabs in France, Mydealz in Germany, Promodescuentos in Mexico and others. By empowering its buyers to post, vote and engage, **Pepper is transforming shopping into an interactive experience** and shaping the future of online marketplaces. [Visit Pepper](https://www.pepper.com/) **Next Case Studies** [![The Telegraph](https://www.recombee.com/img/case-studies/small-covers/case-studies-the-telegraph.png)+35%CTR uplift on article detail pagesMedia Company](https://www.recombee.com/case-studies/the-telegraph) [![Apify](https://www.recombee.com/img/case-studies/small-covers/case-studies-apify.png)+6%SubscriptionsAI Marketplace](https://www.recombee.com/case-studies/apify) [![Diagnal](https://www.recombee.com/img/case-studies/small-covers/case-studies-diagnal.png)+35%Playbacks Across Video PlatformsVideo Streaming](https://www.recombee.com/case-studies/diagnal) [![9GAG](https://www.recombee.com/img/case-studies/small-covers/case-studies-9gag.png)+37%Post ViewsCross-Platform Entertainment Network](https://www.recombee.com/case-studies/9gag) [![Poslovi Infostud](https://www.recombee.com/img/case-studies/small-covers/case-studies-poslovi-infostud.png)+32%Application conversion on job listingsOnline Job Board](https://www.recombee.com/case-studies/poslovi-infostud) [![Perlego](https://www.recombee.com/img/case-studies/small-covers/case-studies-perlego.png)+44%CTRDigital Learning](https://www.recombee.com/case-studies/perlego) [![Slickdeals](https://www.recombee.com/img/case-studies/small-covers/case-studies-slickdeals.png)+70%CTR to Detail Page ViewsE-commerce + Deal Aggregators](https://www.recombee.com/case-studies/slickdeals) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima.png)+34%Video views on VOD platformMedia Company](https://www.recombee.com/case-studies/ftv-prima) [![Audiomack](https://www.recombee.com/img/case-studies/small-covers/case-studies-audiomack.png)+206%Monthly plays from recommendationsMusic](https://www.recombee.com/case-studies/audiomack) [![Pet Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-pet-media-group.png)+17%CR from visitor to active buyerP2P Marketplaces](https://www.recombee.com/case-studies/pet-media-group) [![Crexi](https://www.recombee.com/img/case-studies/small-covers/case-studies-crexi.png)+40%In buy actions from the recommended propertiesReal Estate](https://www.recombee.com/case-studies/crexi) [![Triola](https://www.recombee.com/img/case-studies/small-covers/case-studies-triola.png)10%Total orders from recommendationsE-commerce](https://www.recombee.com/case-studies/triola) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima-content.png)+10%RecirculationMedia Company](https://www.recombee.com/case-studies/ftv-prima-content) [![Unfiltered Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-unfiltered-media-group.png)+50%CTRMedia Company](https://www.recombee.com/case-studies/unfiltered-media-group) [![Reliving](https://www.recombee.com/img/case-studies/small-covers/case-studies-reliving.png)+37%Place BidE-commerce + P2P Marketplaces](https://www.recombee.com/case-studies/reliving) [![Autohaus Kunzmann](https://www.recombee.com/img/case-studies/small-covers/case-studies-autohaus-kunzmann.png)+14%Conversion RateE-commerce](https://www.recombee.com/case-studies/autohaus-kunzmann) [![Showmax](https://www.recombee.com/img/case-studies/small-covers/case-studies-showmax.png)70Countries and multiple languagesSVOD Service](https://www.recombee.com/case-studies/showmax) [![Mafra](https://www.recombee.com/img/case-studies/small-covers/case-studies-mafra.png)40% higherCTR of suggested articlesMedia Company](https://www.recombee.com/case-studies/mafra) [![itison](https://www.recombee.com/img/case-studies/small-covers/case-studies-itison.png)2 000%ROINewsletter Personalization + Deal Aggregators](https://www.recombee.com/case-studies/itison) [![Cooklist](https://www.recombee.com/img/case-studies/small-covers/case-studies-cooklist.png)+27%CTRE-commerce](https://www.recombee.com/case-studies/cooklist) [![Economia](https://www.recombee.com/img/case-studies/small-covers/case-studies-economia.png)+64%Click Through RateMedia Company](https://www.recombee.com/case-studies/economia) [![Segundamano](https://www.recombee.com/img/case-studies/small-covers/case-studies-segundamano.png)3x more conversionsfrom users who engage with recommendationsP2P Marketplaces](https://www.recombee.com/case-studies/segundamano) [![Design Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-design-group.png)+52%Number of PurchasesE-commerce](https://www.recombee.com/case-studies/design-group) --- # Driving Growth for the World’s Leading Pet Marketplaces | Case Study > Source: https://www.recombee.com/case-studies/pet-media-group > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [Success Stories](https://www.recombee.com/case-studies) # Driving Growth for the World’s Leading Pet Marketplaces Pet Media Group [P2P Marketplaces](https://www.recombee.com/domains/p2p-marketplaces) [Product Recommendations](https://www.recombee.com/product-recommendations) ![](https://www.recombee.com/img/case-studies/pet-media-group.png) **Pet Media Group** is a dynamic and innovative digital media company dedicated to **helping thousands of animals** every day **find their loving forever home** and guide their human caregivers through the process. On the mission to partner with and acquire the **market-leading pet and horse marketplaces across Europe**, Pet Media Group adds value to consumers, businesses, and animals through technology, marketing, operations, and capital. Recognizing the **significance of delivering highly relevant and personalized content** to their audience to drive their discovery and re-engagement, **the company decided to implement Recombee.** +34% **Click-rate** from newsletters +29% **Click-through rate** on suggested adverts +17% **Conversion rate** from visitor to active buyer ### Situation * Simple in-house solution for showing relevant adverts * Multiple “dead-ends” on the platform where customers frequently ended their session * Presence in markets with less supply and more demand (need to maximize the customers’ probability of finding the right pet) ### Requirements * Flexibility in supporting multiple recommendation use cases across all the brands * Simplicity of implementation & configurability for own specific KPIs * Better discovery of adverts & customer re-engagement (more recurring & loyal customers) ### Solution **It only takes a third of the time** to go live compared to other alternatives on the market Thanks to the existing product feed, **working proof of concept was a matter of minutes.** A **highly customizable feed reader** supporting various formats allows our customers to upload the whole catalog in a few clicks. A **dedicated support team** guiding through the implementation in case of any uncertainties. No more "Sorry, 0 results found" pages Recombee covers all those situations to make the user **discover other relevant content, rather than churning.** AI-powered personalized recommendations based on multiple parameters including the pet taxonomy. Fully Recombee-powered email recommendations Using emails as a channel has gone from being something that only drives marginal value to a **core part of PMG’s overall growth strategy.** **Highly scalable infrastructure** enabling sending of millions of monthly emails at ease. Fine-tuned rotation mechanism to increase the recipients' **discovery and content relevance.** ### Benefits & Results Website Personalization * **29% increase** in click-through rate on suggested adverts * **17% increase** in conversion rate from visitor to active buyer * **Increased number of listings** that each buyer interacts with * Expansion of the **session duration** Emailing * **\-36%** of unsubscribes * **+18%** in open-rate * **+34%** in click-rate ### Scenarios ![pet-media-group](https://www.recombee.com/img/case-studies/scenarios/pet-media-group-1.png) #### Homepage Personalization One of the first personalization touchpoints for first-time users on **every brand homepage** under the Pet Media Group portfolio. A mixture of **collaborative filtering models** combined with **popularity-based models** to show relevant content from the very first session which is **more tailored** as the user interacts with the platform. ![pet-media-group](https://www.recombee.com/img/case-studies/scenarios/pet-media-group-2.png) #### Advert Detail Personalization Customers often come to the platform with non-specific notions expecting that **relevant adverts will be shown** to them and help them through the decision-making process. This unique **collaborative-filtering** scenario backed by **content-based models** on the advert detail helps undecided customers find the pet they love based on their similarities and behavior characteristics (e.g. playfulness, fearlessness, etc.). ![pet-media-group](https://www.recombee.com/img/case-studies/scenarios/pet-media-group-3.png) #### Search Results Personalization A crucial scenario to keep the visitors engaged and help them **discover possible alternatives** when they search for something specific and have no results for it. One of the **most tailored sets of models** based on animal taxonomy to find the right candidates to show and **prevent visitors from churning** to other platforms. ![pet-media-group](https://www.recombee.com/img/case-studies/scenarios/pet-media-group-4.png) #### Newsletter Personalization Contacting registered users via our personalized emails became a **key component of growth across all the markets** thanks to its relevance and effectiveness. **State-of-the-art models** together with **distinctive parametrization** of how often should be recommended content shown again based on the success of other emails make it easy to **discover new content** of interest. “Our experience with Recombee has been exceptional, improving our marketplace's efficiency and user engagement. Integration was quick, taking less than two hours, accelerating our time to market. The low total cost of ownership and the intuitive UI for seamless system adjustments allowed our team to innovate at an unprecedented pace. Recombee's impact on our metrics across various touchpoints - on-site, in-app, and through email and push notifications - has been phenomenal: a 29% increase in click-through rate, a 17% boost in on-site conversion rate, and an 18% increase in email open rates, 34% increase in click-through rate and 36% decrease in unsubscribes, significantly elevating the relevance of our communications. The value of Recombee's robust recommendation system is essential for any e-commerce player aiming for a competitive edge without the massive investment of developing this expertise internally.“ ![Eyass Shakrah](https://www.recombee.com/img/case-studies/testimonials/pmg.png) **Eyass Shakrah** Co-Founder at Pet Media Group ![Pet Media Group](https://www.recombee.com/img/case-studies/pet-media-group-logos.png) ### About Pet Media Group Through innovation, technology, and access to capital, Pet Media Group has a **unique position to lead the modernization of pet marketplaces** all over Europe to help **over 1.5 million animals** every year find their new loving home. With a diverse portfolio of websites, mobile apps, and social media platforms, **Pet Media Group offers a comprehensive range of content,** including articles, videos, product reviews, and interactive experiences centered around pets and horses. Recombee powers **personalization on all brands** under the Pet Media Group. [Visit Pet Media Group](https://www.petmediagroup.com/) **Next Case Studies** [![The Telegraph](https://www.recombee.com/img/case-studies/small-covers/case-studies-the-telegraph.png)+35%CTR uplift on article detail pagesMedia Company](https://www.recombee.com/case-studies/the-telegraph) [![Apify](https://www.recombee.com/img/case-studies/small-covers/case-studies-apify.png)+6%SubscriptionsAI Marketplace](https://www.recombee.com/case-studies/apify) [![Diagnal](https://www.recombee.com/img/case-studies/small-covers/case-studies-diagnal.png)+35%Playbacks Across Video PlatformsVideo Streaming](https://www.recombee.com/case-studies/diagnal) [![9GAG](https://www.recombee.com/img/case-studies/small-covers/case-studies-9gag.png)+37%Post ViewsCross-Platform Entertainment Network](https://www.recombee.com/case-studies/9gag) [![Poslovi Infostud](https://www.recombee.com/img/case-studies/small-covers/case-studies-poslovi-infostud.png)+32%Application conversion on job listingsOnline Job Board](https://www.recombee.com/case-studies/poslovi-infostud) [![Perlego](https://www.recombee.com/img/case-studies/small-covers/case-studies-perlego.png)+44%CTRDigital Learning](https://www.recombee.com/case-studies/perlego) [![Slickdeals](https://www.recombee.com/img/case-studies/small-covers/case-studies-slickdeals.png)+70%CTR to Detail Page ViewsE-commerce + Deal Aggregators](https://www.recombee.com/case-studies/slickdeals) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima.png)+34%Video views on VOD platformMedia Company](https://www.recombee.com/case-studies/ftv-prima) [![Audiomack](https://www.recombee.com/img/case-studies/small-covers/case-studies-audiomack.png)+206%Monthly plays from recommendationsMusic](https://www.recombee.com/case-studies/audiomack) [![Pepper](https://www.recombee.com/img/case-studies/small-covers/case-studies-pepper.png)+21%Click-outs to affiliate links from websiteDeal Aggregators](https://www.recombee.com/case-studies/pepper) [![Crexi](https://www.recombee.com/img/case-studies/small-covers/case-studies-crexi.png)+40%In buy actions from the recommended propertiesReal Estate](https://www.recombee.com/case-studies/crexi) [![Triola](https://www.recombee.com/img/case-studies/small-covers/case-studies-triola.png)10%Total orders from recommendationsE-commerce](https://www.recombee.com/case-studies/triola) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima-content.png)+10%RecirculationMedia Company](https://www.recombee.com/case-studies/ftv-prima-content) [![Unfiltered Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-unfiltered-media-group.png)+50%CTRMedia Company](https://www.recombee.com/case-studies/unfiltered-media-group) [![Reliving](https://www.recombee.com/img/case-studies/small-covers/case-studies-reliving.png)+37%Place BidE-commerce + P2P Marketplaces](https://www.recombee.com/case-studies/reliving) [![Autohaus Kunzmann](https://www.recombee.com/img/case-studies/small-covers/case-studies-autohaus-kunzmann.png)+14%Conversion RateE-commerce](https://www.recombee.com/case-studies/autohaus-kunzmann) [![Showmax](https://www.recombee.com/img/case-studies/small-covers/case-studies-showmax.png)70Countries and multiple languagesSVOD Service](https://www.recombee.com/case-studies/showmax) [![Mafra](https://www.recombee.com/img/case-studies/small-covers/case-studies-mafra.png)40% higherCTR of suggested articlesMedia Company](https://www.recombee.com/case-studies/mafra) [![itison](https://www.recombee.com/img/case-studies/small-covers/case-studies-itison.png)2 000%ROINewsletter Personalization + Deal Aggregators](https://www.recombee.com/case-studies/itison) [![Cooklist](https://www.recombee.com/img/case-studies/small-covers/case-studies-cooklist.png)+27%CTRE-commerce](https://www.recombee.com/case-studies/cooklist) [![Economia](https://www.recombee.com/img/case-studies/small-covers/case-studies-economia.png)+64%Click Through RateMedia Company](https://www.recombee.com/case-studies/economia) [![Segundamano](https://www.recombee.com/img/case-studies/small-covers/case-studies-segundamano.png)3x more conversionsfrom users who engage with recommendationsP2P Marketplaces](https://www.recombee.com/case-studies/segundamano) [![Design Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-design-group.png)+52%Number of PurchasesE-commerce](https://www.recombee.com/case-studies/design-group) --- # Content Recommendations Drive Showmax Engagement Rate > Source: https://www.recombee.com/case-studies/showmax > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [Success Stories](https://www.recombee.com/case-studies) # Satisfaction Delivered Through Personalized Content Recommendations Showmax [Video](https://www.recombee.com/domains/video) [Content Recommendations](https://www.recombee.com/content-recommendations) ![](https://www.recombee.com/img/case-studies/showmax.png) Showmax has chosen Recombee as a long term strategic partner for personalization service. Together, we are developing fast growing global SVoD service that understands the preferences of individual users. 70 Countries and multiple languages ### Situation * Millions of Recommendation possibilities. * Tens of thousand of videos to recommend. * Static recommendations managed by editors. ### Requirements * Recognize preferences of individual users. * Respond in real-time under large traffic. * Real-time response in large traffic. * Multiple countries and languages. ### Solution * Automatic personalized recommendations of Movies and series. * Model for every single user with real-time updates. ### Benefits & Results * **Increased** user engagement. * **Higher satisfaction** with Showmax products. * **Decreased** churn rates. * **Intense** collaboration on new features. * **Constant innovations and state** of the art algorithms. * **Ability to customize** recommendations for business needs. * **Ability to handle** country specific content and limitations. ### Scenarios Personalize Video [Homepage](#homepage) [Browsing Page](#browsing) [Detail Page](#detail) #### Fully Personalized Homepage Automated personalized rows with movies and series using a rich mixture of algorithms and advanced configuration options. ![showmax](https://www.recombee.com/img/case-studies/scenarios/showmax-1.png) #### Recommended For You Leveraging advanced machine learning algorithms, the system provides an automatically curated list of content tailored to individual tastes, analyzing various data points such as genre and release date. ![showmax](https://www.recombee.com/img/case-studies/scenarios/showmax-2.png) #### Most Popular Showing the most popular content across the whole platform while still being personalized to individual users. ![showmax](https://www.recombee.com/img/case-studies/scenarios/showmax-3.png) #### Series Picked Just for You The personalized row of Series that are tailored to meet your unique taste and spark your interest. ![showmax](https://www.recombee.com/img/case-studies/scenarios/showmax-4.png) #### Showmax Collections Recommending collections of precisely hand-picked movies based on the behavior on the platform and content similarities. #### Browse Pages Personalization Transforming the browsing experience into a curated film festival with smart recommendations that resonate with users' unique cinematic taste. ![showmax](https://www.recombee.com/img/case-studies/scenarios/showmax-5.png) #### Movies for You Utilizing sophisticated machine learning algorithms, the system conducts a multi-dimensional analysis of data points to generate highly accurate recommendations so the users find their next movie binge in a matter of seconds. ![showmax](https://www.recombee.com/img/case-studies/scenarios/showmax-6.png) #### Movies from Same Genre AI-driven recommendations that bring viewers closer to their next favorite film, making every movie night a memorable experience. #### Video Detail Personalization AI-driven recommendations that bring viewers closer to their next favorite film, making every movie night a memorable experience. ![showmax](https://www.recombee.com/img/case-studies/scenarios/showmax-7.png) #### People also Watched Presenting relevant similar movies based on the video the user is currently watching and the behavior of other users on the platform. ![showmax](https://www.recombee.com/img/case-studies/scenarios/showmax-8.png) #### Watch Next Motivating users to stay in the binge-watching zone using end-of-playback recommendations while maximizing enjoyment and satisfaction with the platform. “Recombee is capable of scaling the service and **keeps pace with our rapid growth. Constant innovation and proactive** development of new features makes our collaboration smooth and pleasant.“ **Meindert van der Meulen** Head of Strategy and Business Intelligence at Showmax ![Showmax](https://www.recombee.com/img/logos/showmax.svg) ### About Showmax Showmax is a part of MultiChoice Group. For a single monthly fee, get all-you-can-eat access to a huge online catalogue of TV shows, movies, kids’ shows and documentaries. Start and stop when you want. No ads. Cancel anytime – there’s no contract. Stream Showmax using apps for smart TVs, smartphones, tablets, computers and media players. Showmax also works with Chromecast and AirPlay. Manage data consumption using the bandwidth capping feature. No internet? No problem – download up to 25 shows to smartphones and tablets to watch later offline. Showmax was born in 2015 and is part of the Naspers group. Showmax is currently available in more than 70 countries. For a free, no-risk trial, visit www.showmax.com. [Visit Showmax](https://www.showmax.com/) **Next Case Studies** [![The Telegraph](https://www.recombee.com/img/case-studies/small-covers/case-studies-the-telegraph.png)+35%CTR uplift on article detail pagesMedia Company](https://www.recombee.com/case-studies/the-telegraph) [![Apify](https://www.recombee.com/img/case-studies/small-covers/case-studies-apify.png)+6%SubscriptionsAI Marketplace](https://www.recombee.com/case-studies/apify) [![Diagnal](https://www.recombee.com/img/case-studies/small-covers/case-studies-diagnal.png)+35%Playbacks Across Video PlatformsVideo Streaming](https://www.recombee.com/case-studies/diagnal) [![9GAG](https://www.recombee.com/img/case-studies/small-covers/case-studies-9gag.png)+37%Post ViewsCross-Platform Entertainment Network](https://www.recombee.com/case-studies/9gag) [![Poslovi Infostud](https://www.recombee.com/img/case-studies/small-covers/case-studies-poslovi-infostud.png)+32%Application conversion on job listingsOnline Job Board](https://www.recombee.com/case-studies/poslovi-infostud) [![Perlego](https://www.recombee.com/img/case-studies/small-covers/case-studies-perlego.png)+44%CTRDigital Learning](https://www.recombee.com/case-studies/perlego) [![Slickdeals](https://www.recombee.com/img/case-studies/small-covers/case-studies-slickdeals.png)+70%CTR to Detail Page ViewsE-commerce + Deal Aggregators](https://www.recombee.com/case-studies/slickdeals) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima.png)+34%Video views on VOD platformMedia Company](https://www.recombee.com/case-studies/ftv-prima) [![Audiomack](https://www.recombee.com/img/case-studies/small-covers/case-studies-audiomack.png)+206%Monthly plays from recommendationsMusic](https://www.recombee.com/case-studies/audiomack) [![Pet Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-pet-media-group.png)+17%CR from visitor to active buyerP2P Marketplaces](https://www.recombee.com/case-studies/pet-media-group) [![Pepper](https://www.recombee.com/img/case-studies/small-covers/case-studies-pepper.png)+21%Click-outs to affiliate links from websiteDeal Aggregators](https://www.recombee.com/case-studies/pepper) [![Crexi](https://www.recombee.com/img/case-studies/small-covers/case-studies-crexi.png)+40%In buy actions from the recommended propertiesReal Estate](https://www.recombee.com/case-studies/crexi) [![Triola](https://www.recombee.com/img/case-studies/small-covers/case-studies-triola.png)10%Total orders from recommendationsE-commerce](https://www.recombee.com/case-studies/triola) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima-content.png)+10%RecirculationMedia Company](https://www.recombee.com/case-studies/ftv-prima-content) [![Unfiltered Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-unfiltered-media-group.png)+50%CTRMedia Company](https://www.recombee.com/case-studies/unfiltered-media-group) [![Reliving](https://www.recombee.com/img/case-studies/small-covers/case-studies-reliving.png)+37%Place BidE-commerce + P2P Marketplaces](https://www.recombee.com/case-studies/reliving) [![Autohaus Kunzmann](https://www.recombee.com/img/case-studies/small-covers/case-studies-autohaus-kunzmann.png)+14%Conversion RateE-commerce](https://www.recombee.com/case-studies/autohaus-kunzmann) [![Mafra](https://www.recombee.com/img/case-studies/small-covers/case-studies-mafra.png)40% higherCTR of suggested articlesMedia Company](https://www.recombee.com/case-studies/mafra) [![itison](https://www.recombee.com/img/case-studies/small-covers/case-studies-itison.png)2 000%ROINewsletter Personalization + Deal Aggregators](https://www.recombee.com/case-studies/itison) [![Cooklist](https://www.recombee.com/img/case-studies/small-covers/case-studies-cooklist.png)+27%CTRE-commerce](https://www.recombee.com/case-studies/cooklist) [![Economia](https://www.recombee.com/img/case-studies/small-covers/case-studies-economia.png)+64%Click Through RateMedia Company](https://www.recombee.com/case-studies/economia) [![Segundamano](https://www.recombee.com/img/case-studies/small-covers/case-studies-segundamano.png)3x more conversionsfrom users who engage with recommendationsP2P Marketplaces](https://www.recombee.com/case-studies/segundamano) [![Design Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-design-group.png)+52%Number of PurchasesE-commerce](https://www.recombee.com/case-studies/design-group) --- # Slickdeals Boosts CTR With Personalized Recommendations > Source: https://www.recombee.com/case-studies/slickdeals > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [Success Stories](https://www.recombee.com/case-studies) # AI Product Recommendations With Highly Volatile Inventory and User-Generated Content Slickdeals [E-commerce](https://www.recombee.com/domains/e-commerce), [Deal Aggregators](https://www.recombee.com/domains/deal-aggregators) [Product Recommendations](https://www.recombee.com/product-recommendations) ![](https://www.recombee.com/img/case-studies/slickdeals.png) Slickdeals is the largest social platform for shopping, dedicated to helping savvy shoppers score deals on the products they love with insight from millions of real people. With a high number of available products and the continually evolving content, smart product recommendations were a must. Recombee handled this specific use case with huge success and helped Slickdeals to provide the best-picked deals that are remarkable for their worth, quality, and timeliness. 70% Increase in CTR to Detail Page Views 30% Increase in CTR to Affiliate Links ### Situation * Highly volatile and fast-changing inventory with user-generated content. * 25 million users. * Slickdeals team of experts curates the frontpage. ### Requirements * Recommendations of time-sensitive deals and coupons. * Immediate response to recent deals. * Recommendations based on customers’ votes and feedback. * Real-time personalization for every individual user. ### Solution * **“Just For You”** section on the homepage. * **Custom made business rules** for old and expired deals. * **A complex ensemble of** incrementally-trained ML models: * **Collaborative filtering models** accelerated through sparse locality-sensitive hashing. * **Natural language processing** using deep recurrent neural networks. * **Image processing** using deep convolutional neural networks. * Fully leveraging Recombee technology for **real-time data processing:** * Most models designed for **incremental training and live updates.** * **Queued processing** of new incoming data (with constant reprioritization) for near-real-time deep-learning models. * **Discovery of hot deals through** reinforcement learning and contextual bandit models. ### Benefits & Results * **70% increase** in CTR to a product detail page. * **30% increase** in CTR to affiliate links. * Improvement of user experience and engagement. * Real-time personalization for every individual user. ### Scenarios ![slickdeals](https://www.recombee.com/img/case-studies/scenarios/slickdeals-1.png) #### Just For You Main personalization strategy behind this scenario is to provide unique experience for every individual user who visits Slickdeals homepage by offering the newest, diverse and most relevant deals. “Recombee was able to handle our very specific use case around providing recommendations with a highly volatile inventory of user- generated content. Placing recommendations on our homepage was a huge success — 70%+ higher product detail page views and 30%+ higher clickthroughs. The Recombee team is a great partner in helping solve our unique use cases, and we look forward to continue working with them.” **Daniel Uhm** Product Manager at Slickdeals ![Slickdeals](https://www.recombee.com/img/logos/slickdeals.svg) ### About Slickdeals Slickdeals is one of the top 100 most visited sites in the U.S. with 25 million users in its social shopping network. Their members are connecting with each other to share the most up-to-date deals and coupons and Slickdeals helped them to save over $6.8 billion by offering a messaging platform, as well as shopping resources like its free Android or iOS applications and browser extensions for Chrome and Edge. [Visit Slickdeals](https://slickdeals.net/) **Next Case Studies** [![The Telegraph](https://www.recombee.com/img/case-studies/small-covers/case-studies-the-telegraph.png)+35%CTR uplift on article detail pagesMedia Company](https://www.recombee.com/case-studies/the-telegraph) [![Apify](https://www.recombee.com/img/case-studies/small-covers/case-studies-apify.png)+6%SubscriptionsAI Marketplace](https://www.recombee.com/case-studies/apify) [![Diagnal](https://www.recombee.com/img/case-studies/small-covers/case-studies-diagnal.png)+35%Playbacks Across Video PlatformsVideo Streaming](https://www.recombee.com/case-studies/diagnal) [![9GAG](https://www.recombee.com/img/case-studies/small-covers/case-studies-9gag.png)+37%Post ViewsCross-Platform Entertainment Network](https://www.recombee.com/case-studies/9gag) [![Poslovi Infostud](https://www.recombee.com/img/case-studies/small-covers/case-studies-poslovi-infostud.png)+32%Application conversion on job listingsOnline Job Board](https://www.recombee.com/case-studies/poslovi-infostud) [![Perlego](https://www.recombee.com/img/case-studies/small-covers/case-studies-perlego.png)+44%CTRDigital Learning](https://www.recombee.com/case-studies/perlego) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima.png)+34%Video views on VOD platformMedia Company](https://www.recombee.com/case-studies/ftv-prima) [![Audiomack](https://www.recombee.com/img/case-studies/small-covers/case-studies-audiomack.png)+206%Monthly plays from recommendationsMusic](https://www.recombee.com/case-studies/audiomack) [![Pet Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-pet-media-group.png)+17%CR from visitor to active buyerP2P Marketplaces](https://www.recombee.com/case-studies/pet-media-group) [![Pepper](https://www.recombee.com/img/case-studies/small-covers/case-studies-pepper.png)+21%Click-outs to affiliate links from websiteDeal Aggregators](https://www.recombee.com/case-studies/pepper) [![Crexi](https://www.recombee.com/img/case-studies/small-covers/case-studies-crexi.png)+40%In buy actions from the recommended propertiesReal Estate](https://www.recombee.com/case-studies/crexi) [![Triola](https://www.recombee.com/img/case-studies/small-covers/case-studies-triola.png)10%Total orders from recommendationsE-commerce](https://www.recombee.com/case-studies/triola) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima-content.png)+10%RecirculationMedia Company](https://www.recombee.com/case-studies/ftv-prima-content) [![Unfiltered Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-unfiltered-media-group.png)+50%CTRMedia Company](https://www.recombee.com/case-studies/unfiltered-media-group) [![Reliving](https://www.recombee.com/img/case-studies/small-covers/case-studies-reliving.png)+37%Place BidE-commerce + P2P Marketplaces](https://www.recombee.com/case-studies/reliving) [![Autohaus Kunzmann](https://www.recombee.com/img/case-studies/small-covers/case-studies-autohaus-kunzmann.png)+14%Conversion RateE-commerce](https://www.recombee.com/case-studies/autohaus-kunzmann) [![Showmax](https://www.recombee.com/img/case-studies/small-covers/case-studies-showmax.png)70Countries and multiple languagesSVOD Service](https://www.recombee.com/case-studies/showmax) [![Mafra](https://www.recombee.com/img/case-studies/small-covers/case-studies-mafra.png)40% higherCTR of suggested articlesMedia Company](https://www.recombee.com/case-studies/mafra) [![itison](https://www.recombee.com/img/case-studies/small-covers/case-studies-itison.png)2 000%ROINewsletter Personalization + Deal Aggregators](https://www.recombee.com/case-studies/itison) [![Cooklist](https://www.recombee.com/img/case-studies/small-covers/case-studies-cooklist.png)+27%CTRE-commerce](https://www.recombee.com/case-studies/cooklist) [![Economia](https://www.recombee.com/img/case-studies/small-covers/case-studies-economia.png)+64%Click Through RateMedia Company](https://www.recombee.com/case-studies/economia) [![Segundamano](https://www.recombee.com/img/case-studies/small-covers/case-studies-segundamano.png)3x more conversionsfrom users who engage with recommendationsP2P Marketplaces](https://www.recombee.com/case-studies/segundamano) [![Design Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-design-group.png)+52%Number of PurchasesE-commerce](https://www.recombee.com/case-studies/design-group) --- # Increasing Engagement by 35% for The Telegraph, The UK’s Leading News Publisher > Source: https://www.recombee.com/case-studies/the-telegraph > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # [Success Stories](https://www.recombee.com/case-studies) # Increasing Engagement by 35% for The Telegraph, The UK’s Leading News Publisher The Telegraph [Articles, News, Media](https://www.recombee.com/domains/articles-news-media) [Content Recommendations](https://www.recombee.com/content-recommendations) ![](https://www.recombee.com/img/case-studies/the-telegraph.png) **The Telegraph** is one of the UK’s most established news publishers, reaching millions of readers across its website and mobile app. With a large, fast-moving content catalogue and a highly diverse audience, helping readers find relevant stories quickly and naturally is an ongoing priority. To further strengthen content discovery and reader engagement, The Telegraph partnered with **Recombee** to support real-time, personalized recommendations, designed to work alongside strong editorial standards rather than replace them. +35% **CTR uplift** versus a competitive solution in A/B testing on article detail pages +8% **Average sessions** per user within the personalized Life section ### Situation * A large and continuously evolving content catalogue spanning news, politics, business, lifestyle, culture, and more * Millions of readers with highly diverse interests and reading behaviors across web and mobile platforms * Increasing expectations for relevance, speed, and personalization at scale * A clear need to balance advanced personalization with strong editorial standards and content quality ### Objectives * Deepen reader engagement and content discovery across digital platforms * Optimize click-through rates and onward journeys from article pages * Deliver real-time personalization across web and mobile app experiences * Apply AI-driven recommendations in a way that aligns with editorial priorities and oversight * Gain deeper insight into content performance and reader behavior to support data-informed decisions ### Solution Recombee supports personalized content recommendations across The Telegraph’s digital platforms. Personalization influences which content is shown, in what order, and where, adapting in real time as readers interact with the site or app. Personalization is applied in a way that complements editorial decisions rather than overriding them. **Article Page Recommendations** Dynamic recommendation boxes surface related, same-category, and top-performing articles, encouraging continued reading beyond the initial visit. ##### Key Benefits * Stronger onward journeys from article pages * Increased exposure to relevant content across the catalogue **Life Section Personalization** The Life section adapts for each reader through personalized article selection, dynamic topic ordering, and a continuously updating feed. Editors define curated article pools, while Reecombee personalized: * The order of editor-approved content * The ordering of lifestyle sections (Film & TV, Cars, Fashion, etc.) * The articles surfaced within each section * The dynamic feed experience This ensures personalization operates within clear editorial boundaries. ##### Key Benefits * More relevant lifestyle content for each reader * Faster access to preferred topics * Stronger onward journeys within the section * Preserved editorial control ### Benefits & Results * **+35% CTR uplift** versus a competitive solution in A/B testing on article detail pages * **+8% avg. sessions** per user within the personalized Life section * **Increased content exploration** across article pages and sections * **More relevant and personalized reader journeys** across web and mobile app ### Recombee’s Solution in Action ![the-telegraph](https://www.recombee.com/img/case-studies/scenarios/the-telegraph-1.png) ### Real-Time Article Recommendations At the end of articles on both web and mobile, readers see dynamic recommendation boxes featuring: * Related articles * Content from the same category * Top stories from across The Telegraph These placements help readers naturally continue their journey after finishing an article. ![the-telegraph](https://www.recombee.com/img/case-studies/scenarios/the-telegraph-2.png) ### Personalized Life Section In the mobile app, the **Life section** is fully personalized for each reader. This includes: * Personalized Editors’ Picks * Dynamic Section Ordering * Continuous Personalized Feed Instead of a fixed layout, the section adapts to what each reader is most likely to engage with. ![the-telegraph](https://www.recombee.com/img/case-studies/scenarios/the-telegraph-3.png) #### Personalized Editors’ Picks Editors curate a pool of recommended articles. Recombee then personalizes the order in which those editor-approved articles appear for each reader. This keeps editorial selection intact while allowing the reading experience to feel personal. ![the-telegraph](https://www.recombee.com/img/case-studies/scenarios/the-telegraph-4.png) #### Dynamic Section Ordering Lifestyle sections such as Film & TV, Cars, and Fashion are ordered differently for each reader. Both the order of the sections and the content within them are personalized, allowing the most relevant topics to surface first. ![the-telegraph](https://www.recombee.com/img/case-studies/scenarios/the-telegraph-5.png) #### Continuous Personalized Feed Readers also see a continuously updating feed of articles tailored to their interests. The feed is designed to: * Avoid repeating content already shown elsewhere * Encourage broader discovery * Keep the experience fresh and varied ![the-telegraph](https://www.recombee.com/img/case-studies/scenarios/the-telegraph-6.png) ### Expert Support Recombee works closely with The Telegraph’s teams to refine personalization over time. Hands-on support from ML and data science specialists helps test new approaches, adjust recommendation strategies, and continuously improve performance as reader behavior and editorial needs evolve. ![the-telegraph](https://www.recombee.com/img/case-studies/scenarios/the-telegraph-7.png) ### Segment-Aware Analytics Recombee's GenAI analytics pipeline gives The Telegraph's editorial and analytics teams **real-time visibility** into fast-moving newsroom data, turning raw content signals into **clear, actionable intelligence**. Using **LLM embeddings and sparse autoencoders**, articles are automatically clustered into **emerging topic segments** and named in plain language, with no need for analysts to interpret raw model outputs. Dashboards **refresh every 15 minutes**, surfacing which themes are rising, which are lasting, and what is driving high-value engagement. The result is a shift from article-level metrics to **segment-level insight**, helping editors make faster decisions on promotion, commissioning, and content focus. Published and presented at the ACM RecSys Conference (September 2025). [Read the paper](https://ceur-ws.org/Vol-4056/short1.pdf). “We use Recombee to power our AI personalization & Search at The Telegraph. It immediately proved its value, securing a 35% CTR uplift in an A/B test against a competing solution while simultaneously enhancing our editors' ability to manage and analyse content performance. Beyond the numbers, the collaboration with the Recombee team has been excellent, they helped us push our thinking and we were delighted to jointly present at the RecSys conference and be recognized as an INMA '26 finalist for "Best Use of Generative AI".” ![Tom Kelleher](https://www.recombee.com/img/case-studies/testimonials/the-telegraph.png) **Tom Kelleher** Director of Emerging Technology, AI and Personalisation ![The Telegraph](https://www.recombee.com/img/logos/the-telegraph.svg) ### About The Telegraph The Telegraph is one of the United Kingdom’s leading news publishers, delivering content across news, politics, business, lifestyle, and culture through its website and mobile app. With millions of readers, personalization plays an important role in shaping modern digital experiences. [Visit The Telegraph](https://www.telegraph.co.uk) **Next Case Studies** [![Apify](https://www.recombee.com/img/case-studies/small-covers/case-studies-apify.png)+6%SubscriptionsAI Marketplace](https://www.recombee.com/case-studies/apify) [![Diagnal](https://www.recombee.com/img/case-studies/small-covers/case-studies-diagnal.png)+35%Playbacks Across Video PlatformsVideo Streaming](https://www.recombee.com/case-studies/diagnal) [![9GAG](https://www.recombee.com/img/case-studies/small-covers/case-studies-9gag.png)+37%Post ViewsCross-Platform Entertainment Network](https://www.recombee.com/case-studies/9gag) [![Poslovi Infostud](https://www.recombee.com/img/case-studies/small-covers/case-studies-poslovi-infostud.png)+32%Application conversion on job listingsOnline Job Board](https://www.recombee.com/case-studies/poslovi-infostud) [![Perlego](https://www.recombee.com/img/case-studies/small-covers/case-studies-perlego.png)+44%CTRDigital Learning](https://www.recombee.com/case-studies/perlego) [![Slickdeals](https://www.recombee.com/img/case-studies/small-covers/case-studies-slickdeals.png)+70%CTR to Detail Page ViewsE-commerce + Deal Aggregators](https://www.recombee.com/case-studies/slickdeals) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima.png)+34%Video views on VOD platformMedia Company](https://www.recombee.com/case-studies/ftv-prima) [![Audiomack](https://www.recombee.com/img/case-studies/small-covers/case-studies-audiomack.png)+206%Monthly plays from recommendationsMusic](https://www.recombee.com/case-studies/audiomack) [![Pet Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-pet-media-group.png)+17%CR from visitor to active buyerP2P Marketplaces](https://www.recombee.com/case-studies/pet-media-group) [![Pepper](https://www.recombee.com/img/case-studies/small-covers/case-studies-pepper.png)+21%Click-outs to affiliate links from websiteDeal Aggregators](https://www.recombee.com/case-studies/pepper) [![Crexi](https://www.recombee.com/img/case-studies/small-covers/case-studies-crexi.png)+40%In buy actions from the recommended propertiesReal Estate](https://www.recombee.com/case-studies/crexi) [![Triola](https://www.recombee.com/img/case-studies/small-covers/case-studies-triola.png)10%Total orders from recommendationsE-commerce](https://www.recombee.com/case-studies/triola) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima-content.png)+10%RecirculationMedia Company](https://www.recombee.com/case-studies/ftv-prima-content) [![Unfiltered Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-unfiltered-media-group.png)+50%CTRMedia Company](https://www.recombee.com/case-studies/unfiltered-media-group) [![Reliving](https://www.recombee.com/img/case-studies/small-covers/case-studies-reliving.png)+37%Place BidE-commerce + P2P Marketplaces](https://www.recombee.com/case-studies/reliving) [![Autohaus Kunzmann](https://www.recombee.com/img/case-studies/small-covers/case-studies-autohaus-kunzmann.png)+14%Conversion RateE-commerce](https://www.recombee.com/case-studies/autohaus-kunzmann) [![Showmax](https://www.recombee.com/img/case-studies/small-covers/case-studies-showmax.png)70Countries and multiple languagesSVOD Service](https://www.recombee.com/case-studies/showmax) [![Mafra](https://www.recombee.com/img/case-studies/small-covers/case-studies-mafra.png)40% higherCTR of suggested articlesMedia Company](https://www.recombee.com/case-studies/mafra) [![itison](https://www.recombee.com/img/case-studies/small-covers/case-studies-itison.png)2 000%ROINewsletter Personalization + Deal Aggregators](https://www.recombee.com/case-studies/itison) [![Cooklist](https://www.recombee.com/img/case-studies/small-covers/case-studies-cooklist.png)+27%CTRE-commerce](https://www.recombee.com/case-studies/cooklist) [![Economia](https://www.recombee.com/img/case-studies/small-covers/case-studies-economia.png)+64%Click Through RateMedia Company](https://www.recombee.com/case-studies/economia) [![Segundamano](https://www.recombee.com/img/case-studies/small-covers/case-studies-segundamano.png)3x more conversionsfrom users who engage with recommendationsP2P Marketplaces](https://www.recombee.com/case-studies/segundamano) [![Design Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-design-group.png)+52%Number of PurchasesE-commerce](https://www.recombee.com/case-studies/design-group) --- # Real-Time User Generated Content Personalization > Source: https://www.recombee.com/case-studies/9gag > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # [Success Stories](https://www.recombee.com/case-studies) # Real-time User Generated Content Personalization for a Global Entertainment Giant 9GAG [Cross-Platform Entertainment Network](https://www.recombee.com/domains/articles-news-media) [Content Recommendations](https://www.recombee.com/content-recommendations) ![](https://www.recombee.com/img/case-studies/9gag.png) **9GAG is a global cross-platform entertainment network with 200+ million audiences worldwide.** Ranked #1 in cross-platform video creation in the US, 9GAG distributes 8.4 billion video views and 3.5 billion page views monthly through all platforms. Recombee's smart content recommendations helped 9GAG to fully personalize their homepage infinite scroll to increase multiple KPIs, including post views, the overall number of interactions, and total session duration of all active registered users. Recombee's recommender engine brought a significant increase also in user engagement as well as overall satisfaction. 37% Increase in post views 22% More of overall interactions 7.5% Higher users’ session duration 2.8% Increase in the number of visits ### Situation * Hundreds of millions of monthly post views and other interactions * Millions of heavy users visiting the site very frequently, always looking for something new * Huge catalog of continuously growing user-generated content * In-house recommender system for personalized feeds ### Requirements * A solution capable of replacing the current Home infinite feed under high traffic and minimal response time * Instant recommendation of newly added posts * Comprehensive set of business requirements to ensure compatibility with the current feed * Incremental model learning with constantly incoming interaction, text, and image data * Ensuring that every user sees a brand new feed whenever they return to the page ### Solution Complex and diverse ensemble of incrementally-trained recommendation models. Collaborative Filtering Content-Based * Utilizing latest deep-learning approaches * NLP of titles and meme OCRs * Image processing of both static memes and sampled video frames Reinforcement Learning (Contextual Bandits) Development of a **custom recommendation logic** that exactly matches 9GAG’s product vision and expectations. Optimization for **multiple complex user engagement KPIs** including the number of interactions, total visits, total session duration, retention rate, and other metrics. Delivery of recommendations in 50ms on average. ![9gag](https://www.recombee.com/img/case-studies/scenarios/9gag-1.png) ### Challenges of the User-Generated Content Users are **continuously adding new posts** to 9GAG's platform, so to achieve high-quality personalization, the recommendation system must be able to process them instantly. Newly added posts must **gain immediate traction** by being recommended to the right users through either similarity of visual/textual content or first interactions received. Traditional recommender systems that utilize a standalone "model training phase" cannot handle such situation well, because when the batch training of traditional recsys model is finished, the model is already obsolete. Recombee's unique technology of **online and incremental model training** is designed to ensure smooth personalization for huge and ever-growing catalogs. In combination with reinforcement learning, Recombee can also **identify and promote "hidden gems",** which is critical for any social network. ![9gag](https://www.recombee.com/img/case-studies/scenarios/9gag-2.png) ### Infinite Scroll Personalization To let its users flow through the extensive amount of continuously growing content with no boundaries, 9GAG deployed Recombee's Infinite Scroll. The feature allows users to infinitely scroll through auto-generated recommendations without switching to new tabs, providing the infinite feed for any screen width and autoloading the personalized recommendations in real-time on a mere scroll. * Continuous delivery of personalized content as users scroll down * Bottomless browsing experience with automated feed * Great user experience, especially for touchscreens on e.g., mobile devices ### Benefits & Results #### Home Feed Interactions **+22%** More Overall **+37%** Post Views **+15%** Post Saves **+5%** Post Shares **+14%** Post Upvotes **\-4%** Post Downvotes #### Home Feed Sessions **+3%** Number of Sessions **+7.5%** Total Session Duration of all Users **+4%** Average Session Duration #### Visits **+2.8%** Number of Total Visits **+1.5%** Number of Unique Visitors #### User Engagement * More Clicks, Comments, Upvotes * Less Downvotes * Longer Sessions #### User Retention * More Visiting Users * Increased Frequency of Visits * Higher Total Time Spent ### Scenarios ![9gag](https://www.recombee.com/img/case-studies/scenarios/9gag-3.png) #### “Hot Feed” (“Home Feed”) Recombee was tested in an **extensive A/B** test directly on the user's **Home feed,** which is the main place where users consume most content. The A/B test ran for over **50 days** and looked at a range of metrics, including those measuring the **long-term impact of Recombee** on the Home feed. Recombee has been convincingly proven to optimize not only immediate engagement (CTR, upvotes, session duration) but also **long-term user retention and frequency of visits.** Based on the test results, Recombee now powers **100% of the Home feeds** of all active registered users. “Daily traffic to the 9GAG site amounts to **hundreds of millions of users worldwide.** To provide a **unique user experience** for each and every one of them, we needed to implement a recommendation system that would allow for the most advanced, real-time personalization of their experience. So we chose Recombee. Thanks to their solution, we managed to increase, among other KPIs, **post views by 37%, overall interactions by 22%, and users’ session duration by 7.5%.** With Recombee, we've taken engagement with user-generated content to the next level.” **Kristie Chen** Product Head at 9GAG ![9GAG](https://www.recombee.com/img/logos/w/9gag.svg) ### About 9GAG The Largest Meme Community 9GAG started in 2008 as a site for hosting and distributing funny pictures and videos. Back then, social networks like Facebook or Twitter were on the rise, and people would use them to share the user-generated content they found on 9GAG. 9GAG introduced the revolutionary infinite scroll in 2012. Today, the global entertainment giant has more than 40 million registered users worldwide, making it the largest meme community on the internet. [Visit 9GAG](https://www.9gag.com) **Next Case Studies** [![The Telegraph](https://www.recombee.com/img/case-studies/small-covers/case-studies-the-telegraph.png)+35%CTR uplift on article detail pagesMedia Company](https://www.recombee.com/case-studies/the-telegraph) [![Apify](https://www.recombee.com/img/case-studies/small-covers/case-studies-apify.png)+6%SubscriptionsAI Marketplace](https://www.recombee.com/case-studies/apify) [![Diagnal](https://www.recombee.com/img/case-studies/small-covers/case-studies-diagnal.png)+35%Playbacks Across Video PlatformsVideo Streaming](https://www.recombee.com/case-studies/diagnal) [![Poslovi Infostud](https://www.recombee.com/img/case-studies/small-covers/case-studies-poslovi-infostud.png)+32%Application conversion on job listingsOnline Job Board](https://www.recombee.com/case-studies/poslovi-infostud) [![Perlego](https://www.recombee.com/img/case-studies/small-covers/case-studies-perlego.png)+44%CTRDigital Learning](https://www.recombee.com/case-studies/perlego) [![Slickdeals](https://www.recombee.com/img/case-studies/small-covers/case-studies-slickdeals.png)+70%CTR to Detail Page ViewsE-commerce + Deal Aggregators](https://www.recombee.com/case-studies/slickdeals) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima.png)+34%Video views on VOD platformMedia Company](https://www.recombee.com/case-studies/ftv-prima) [![Audiomack](https://www.recombee.com/img/case-studies/small-covers/case-studies-audiomack.png)+206%Monthly plays from recommendationsMusic](https://www.recombee.com/case-studies/audiomack) [![Pet Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-pet-media-group.png)+17%CR from visitor to active buyerP2P Marketplaces](https://www.recombee.com/case-studies/pet-media-group) [![Pepper](https://www.recombee.com/img/case-studies/small-covers/case-studies-pepper.png)+21%Click-outs to affiliate links from websiteDeal Aggregators](https://www.recombee.com/case-studies/pepper) [![Crexi](https://www.recombee.com/img/case-studies/small-covers/case-studies-crexi.png)+40%In buy actions from the recommended propertiesReal Estate](https://www.recombee.com/case-studies/crexi) [![Triola](https://www.recombee.com/img/case-studies/small-covers/case-studies-triola.png)10%Total orders from recommendationsE-commerce](https://www.recombee.com/case-studies/triola) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima-content.png)+10%RecirculationMedia Company](https://www.recombee.com/case-studies/ftv-prima-content) [![Unfiltered Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-unfiltered-media-group.png)+50%CTRMedia Company](https://www.recombee.com/case-studies/unfiltered-media-group) [![Reliving](https://www.recombee.com/img/case-studies/small-covers/case-studies-reliving.png)+37%Place BidE-commerce + P2P Marketplaces](https://www.recombee.com/case-studies/reliving) [![Autohaus Kunzmann](https://www.recombee.com/img/case-studies/small-covers/case-studies-autohaus-kunzmann.png)+14%Conversion RateE-commerce](https://www.recombee.com/case-studies/autohaus-kunzmann) [![Showmax](https://www.recombee.com/img/case-studies/small-covers/case-studies-showmax.png)70Countries and multiple languagesSVOD Service](https://www.recombee.com/case-studies/showmax) [![Mafra](https://www.recombee.com/img/case-studies/small-covers/case-studies-mafra.png)40% higherCTR of suggested articlesMedia Company](https://www.recombee.com/case-studies/mafra) [![itison](https://www.recombee.com/img/case-studies/small-covers/case-studies-itison.png)2 000%ROINewsletter Personalization + Deal Aggregators](https://www.recombee.com/case-studies/itison) [![Cooklist](https://www.recombee.com/img/case-studies/small-covers/case-studies-cooklist.png)+27%CTRE-commerce](https://www.recombee.com/case-studies/cooklist) [![Economia](https://www.recombee.com/img/case-studies/small-covers/case-studies-economia.png)+64%Click Through RateMedia Company](https://www.recombee.com/case-studies/economia) [![Segundamano](https://www.recombee.com/img/case-studies/small-covers/case-studies-segundamano.png)3x more conversionsfrom users who engage with recommendationsP2P Marketplaces](https://www.recombee.com/case-studies/segundamano) [![Design Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-design-group.png)+52%Number of PurchasesE-commerce](https://www.recombee.com/case-studies/design-group) --- # Replacing Outdated Recommendation Algorithm With Easy to Integrate Solution Across 5 Media Brands > Source: https://www.recombee.com/case-studies/unfiltered-media-group > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [Success Stories](https://www.recombee.com/case-studies) # Replacing Outdated Recommendation Algorithm With Easy to Integrate Solution Across 5 Media Brands Unfiltered Media Group [Articles, News, Media](https://www.recombee.com/domains/articles-news-media) [Product](https://www.recombee.com/product-recommendations) and [Content Recommendations](https://www.recombee.com/content-recommendations) ![](https://www.recombee.com/img/case-studies/unfiltered-media-group.png) The next-generation media company, Unfiltered Media Group, LLC, seeks ways in growth markets to create connections with enthusiastic audiences using both print and digital magazines, books, videos, online courses, apps, festivals, and more, with which they have rich experience. After simple integration, testing, and deployment which took just a couple of hours, Recombee was able to increase click-throughs in time across their 5 different media brands by 50% and improve the user experience and engagement of millions of monthly readers. Recombee is currently deployed on five of their websites and in the 17 different user experience points (scenarios). 50% Increase in CTR ### Situation * Use of general recommendation algorithm. * Millions of monthly readers. * 5 different media brands. ### Requirements * Recommender engine able to provide both product and content recommendations. * Personalization of periodical emails sent to the customers. * Customized homepage with top-notch recommendations. * A fully personalized experience for every individual reader. ### Solution * Use of different models for both content and product recommendations. * Unique recommendation scenarios on each brand’s homepage. * Personalization of periodically sent emails. * Recommendation of trending articles. ### Benefits & Results * **50% increase** in click-throughs. * **Improvement** of user satisfaction. * **Increase** of average time spent. ### Scenarios ![unfiltered-media-group](https://www.recombee.com/img/case-studies/scenarios/unfiltered-media-group-1.png) #### For You This Items to User scenario is using logic type called recombee:default to personalize the offer of the articles on the homepage of the Brewing Industry Guide magazine. Application of business rules to filter out specific items and boost recently published items provides a special experience to every single customer. ![unfiltered-media-group](https://www.recombee.com/img/case-studies/scenarios/unfiltered-media-group-2.png) #### Business Articles For You The main goal of this Items to User scenario using recombee:default logic is to deliver personalized content for every individual visitor of the Craft Beer & Brewing homepage. ![unfiltered-media-group](https://www.recombee.com/img/case-studies/scenarios/unfiltered-media-group-3.png) #### Beer Reviews recombee:default logic used for Items to Item recommendations situated on the right side of the Craft Beer & Brewing homepage. ![unfiltered-media-group](https://www.recombee.com/img/case-studies/scenarios/unfiltered-media-group-4.png) #### For You Items to User homepage scenario used for cross-posting between three different media brands - Spin Off, PieceWork, and Handwoven. This ultimately led to spreading the traffic between all of those sites and an increase in user engagement. ![unfiltered-media-group](https://www.recombee.com/img/case-studies/scenarios/unfiltered-media-group-5.png) #### More For You Every time PieceWork’s potential customer clicks on the desired product and gets himself to the detailed view page, the Items to User default scenario is used at the bottom of the site to recommend other relevant products. ![unfiltered-media-group](https://www.recombee.com/img/case-studies/scenarios/unfiltered-media-group-6.png) #### Other Articles for You Craft Beer & Brewing is periodically sending newsletters to their subscribers with exclusive content using Items to User recommendations and recombee:default logic. At the bottom of those emails, you can find other interesting articles specifically tailored for every individual recipient. ![unfiltered-media-group](https://www.recombee.com/img/case-studies/scenarios/unfiltered-media-group-7.png) #### Trending Articles This scenario is used at the bottom of every AMP article (mobile). Using the recombee:popular logic, the main goal of those Items to User recommendations is to offer currently popular or widely discussed content. “Prior to Recombee, we used a general recommendation algorithm based on popularity and date published. Since moving our recommendation system to Recombee, we’ve seen a 50% increase in click-through across our 5 media brands (millions of readers per month). Recombee was easy to integrate, test, and deploy within just a couple of hours.“ **Haydn Strauss** Chief Operations Officer at Unfiltered Media Group ![Unfiltered Media Group](https://www.recombee.com/img/logos/unfiltered-media-group.png) ### About Unfiltered Media Group The Unfiltered Media Group is channel-agnostic. Through print, digital, and social media they reach demographically diverse buyers by developing media brands and products that resonate with today's consumers on every relevant platform. Recombee is the go-to recommender engine for [Craft Beer & Brewing Magazine](https://beerandbrewing.com/), [Brewing Industry Guide](https://brewingindustryguide.com/), and Long Thread Media ([Handwoven](https://handwovenmagazine.com/), [PieceWork](https://pieceworkmagazine.com/), and [Spin Off](https://spinoffmagazine.com/)). [Visit Unfiltered Media Group](https://www.unflt.com/) **Next Case Studies** [![The Telegraph](https://www.recombee.com/img/case-studies/small-covers/case-studies-the-telegraph.png)+35%CTR uplift on article detail pagesMedia Company](https://www.recombee.com/case-studies/the-telegraph) [![Apify](https://www.recombee.com/img/case-studies/small-covers/case-studies-apify.png)+6%SubscriptionsAI Marketplace](https://www.recombee.com/case-studies/apify) [![Diagnal](https://www.recombee.com/img/case-studies/small-covers/case-studies-diagnal.png)+35%Playbacks Across Video PlatformsVideo Streaming](https://www.recombee.com/case-studies/diagnal) [![9GAG](https://www.recombee.com/img/case-studies/small-covers/case-studies-9gag.png)+37%Post ViewsCross-Platform Entertainment Network](https://www.recombee.com/case-studies/9gag) [![Poslovi Infostud](https://www.recombee.com/img/case-studies/small-covers/case-studies-poslovi-infostud.png)+32%Application conversion on job listingsOnline Job Board](https://www.recombee.com/case-studies/poslovi-infostud) [![Perlego](https://www.recombee.com/img/case-studies/small-covers/case-studies-perlego.png)+44%CTRDigital Learning](https://www.recombee.com/case-studies/perlego) [![Slickdeals](https://www.recombee.com/img/case-studies/small-covers/case-studies-slickdeals.png)+70%CTR to Detail Page ViewsE-commerce + Deal Aggregators](https://www.recombee.com/case-studies/slickdeals) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima.png)+34%Video views on VOD platformMedia Company](https://www.recombee.com/case-studies/ftv-prima) [![Audiomack](https://www.recombee.com/img/case-studies/small-covers/case-studies-audiomack.png)+206%Monthly plays from recommendationsMusic](https://www.recombee.com/case-studies/audiomack) [![Pet Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-pet-media-group.png)+17%CR from visitor to active buyerP2P Marketplaces](https://www.recombee.com/case-studies/pet-media-group) [![Pepper](https://www.recombee.com/img/case-studies/small-covers/case-studies-pepper.png)+21%Click-outs to affiliate links from websiteDeal Aggregators](https://www.recombee.com/case-studies/pepper) [![Crexi](https://www.recombee.com/img/case-studies/small-covers/case-studies-crexi.png)+40%In buy actions from the recommended propertiesReal Estate](https://www.recombee.com/case-studies/crexi) [![Triola](https://www.recombee.com/img/case-studies/small-covers/case-studies-triola.png)10%Total orders from recommendationsE-commerce](https://www.recombee.com/case-studies/triola) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima-content.png)+10%RecirculationMedia Company](https://www.recombee.com/case-studies/ftv-prima-content) [![Reliving](https://www.recombee.com/img/case-studies/small-covers/case-studies-reliving.png)+37%Place BidE-commerce + P2P Marketplaces](https://www.recombee.com/case-studies/reliving) [![Autohaus Kunzmann](https://www.recombee.com/img/case-studies/small-covers/case-studies-autohaus-kunzmann.png)+14%Conversion RateE-commerce](https://www.recombee.com/case-studies/autohaus-kunzmann) [![Showmax](https://www.recombee.com/img/case-studies/small-covers/case-studies-showmax.png)70Countries and multiple languagesSVOD Service](https://www.recombee.com/case-studies/showmax) [![Mafra](https://www.recombee.com/img/case-studies/small-covers/case-studies-mafra.png)40% higherCTR of suggested articlesMedia Company](https://www.recombee.com/case-studies/mafra) [![itison](https://www.recombee.com/img/case-studies/small-covers/case-studies-itison.png)2 000%ROINewsletter Personalization + Deal Aggregators](https://www.recombee.com/case-studies/itison) [![Cooklist](https://www.recombee.com/img/case-studies/small-covers/case-studies-cooklist.png)+27%CTRE-commerce](https://www.recombee.com/case-studies/cooklist) [![Economia](https://www.recombee.com/img/case-studies/small-covers/case-studies-economia.png)+64%Click Through RateMedia Company](https://www.recombee.com/case-studies/economia) [![Segundamano](https://www.recombee.com/img/case-studies/small-covers/case-studies-segundamano.png)3x more conversionsfrom users who engage with recommendationsP2P Marketplaces](https://www.recombee.com/case-studies/segundamano) [![Design Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-design-group.png)+52%Number of PurchasesE-commerce](https://www.recombee.com/case-studies/design-group) --- # Personalization Examples and Case Studies > Source: https://www.recombee.com/case-studies > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. Success Stories # Explore How Businesses Use Recombee to Drive KPIs Recombee’s solution is highly versatile and applicable to numerous domains. We tailor our algorithms to fit clients from e-commerce, media, job portals, mobile apps and many others. ![Works everywhere](https://www.recombee.com/img/case-studies.png) [![The Telegraph](https://www.recombee.com/img/case-studies/small-covers/case-studies-the-telegraph.png)+35%CTR uplift on article detail pagesMedia Company](https://www.recombee.com/case-studies/the-telegraph) [![Apify](https://www.recombee.com/img/case-studies/small-covers/case-studies-apify.png)+6%SubscriptionsAI Marketplace](https://www.recombee.com/case-studies/apify) [![Diagnal](https://www.recombee.com/img/case-studies/small-covers/case-studies-diagnal.png)+35%Playbacks Across Video PlatformsVideo Streaming](https://www.recombee.com/case-studies/diagnal) [![9GAG](https://www.recombee.com/img/case-studies/small-covers/case-studies-9gag.png)+37%Post ViewsCross-Platform Entertainment Network](https://www.recombee.com/case-studies/9gag) [![Poslovi Infostud](https://www.recombee.com/img/case-studies/small-covers/case-studies-poslovi-infostud.png)+32%Application conversion on job listingsOnline Job Board](https://www.recombee.com/case-studies/poslovi-infostud) [![Perlego](https://www.recombee.com/img/case-studies/small-covers/case-studies-perlego.png)+44%CTRDigital Learning](https://www.recombee.com/case-studies/perlego) [![Slickdeals](https://www.recombee.com/img/case-studies/small-covers/case-studies-slickdeals.png)+70%CTR to Detail Page ViewsE-commerce + Deal Aggregators](https://www.recombee.com/case-studies/slickdeals) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima.png)+34%Video views on VOD platformMedia Company](https://www.recombee.com/case-studies/ftv-prima) [![Audiomack](https://www.recombee.com/img/case-studies/small-covers/case-studies-audiomack.png)+206%Monthly plays from recommendationsMusic](https://www.recombee.com/case-studies/audiomack) [![Pet Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-pet-media-group.png)+17%CR from visitor to active buyerP2P Marketplaces](https://www.recombee.com/case-studies/pet-media-group) [![Pepper](https://www.recombee.com/img/case-studies/small-covers/case-studies-pepper.png)+21%Click-outs to affiliate links from websiteDeal Aggregators](https://www.recombee.com/case-studies/pepper) [![Crexi](https://www.recombee.com/img/case-studies/small-covers/case-studies-crexi.png)+40%In buy actions from the recommended propertiesReal Estate](https://www.recombee.com/case-studies/crexi) [![Triola](https://www.recombee.com/img/case-studies/small-covers/case-studies-triola.png)10%Total orders from recommendationsE-commerce](https://www.recombee.com/case-studies/triola) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima-content.png)+10%RecirculationMedia Company](https://www.recombee.com/case-studies/ftv-prima-content) [![Unfiltered Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-unfiltered-media-group.png)+50%CTRMedia Company](https://www.recombee.com/case-studies/unfiltered-media-group) [![Reliving](https://www.recombee.com/img/case-studies/small-covers/case-studies-reliving.png)+37%Place BidE-commerce + P2P Marketplaces](https://www.recombee.com/case-studies/reliving) [![Autohaus Kunzmann](https://www.recombee.com/img/case-studies/small-covers/case-studies-autohaus-kunzmann.png)+14%Conversion RateE-commerce](https://www.recombee.com/case-studies/autohaus-kunzmann) [![Showmax](https://www.recombee.com/img/case-studies/small-covers/case-studies-showmax.png)70Countries and multiple languagesSVOD Service](https://www.recombee.com/case-studies/showmax) [![Mafra](https://www.recombee.com/img/case-studies/small-covers/case-studies-mafra.png)40% higherCTR of suggested articlesMedia Company](https://www.recombee.com/case-studies/mafra) [![itison](https://www.recombee.com/img/case-studies/small-covers/case-studies-itison.png)2 000%ROINewsletter Personalization + Deal Aggregators](https://www.recombee.com/case-studies/itison) [![Cooklist](https://www.recombee.com/img/case-studies/small-covers/case-studies-cooklist.png)+27%CTRE-commerce](https://www.recombee.com/case-studies/cooklist) [![Economia](https://www.recombee.com/img/case-studies/small-covers/case-studies-economia.png)+64%Click Through RateMedia Company](https://www.recombee.com/case-studies/economia) [![Segundamano](https://www.recombee.com/img/case-studies/small-covers/case-studies-segundamano.png)3x more conversionsfrom users who engage with recommendationsP2P Marketplaces](https://www.recombee.com/case-studies/segundamano) [![Design Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-design-group.png)+52%Number of PurchasesE-commerce](https://www.recombee.com/case-studies/design-group) ## Our Customers Say "At DAZN, being the Global Home of Sports means delivering the right matches, highlights, and moments to audiences in 200+ markets - bringing fans even closer to the live game. That’s why we’ve teamed up with Recombee to personalize experiences at scale. Their tech enables us to connect each viewer on any device with the right game or clip in real time through flexible solution built for growth. This partnership sets the pace for a smarter, more connected global sports experience." ![Christoph Haas](https://www.recombee.com/img/customers/dazn.png) Christoph HaasEVP Product & Platform Engineering at DAZN "We use Recombee to power our AI personalization & Search at The Telegraph. It immediately proved its value, securing a 35% CTR uplift in an A/B test against a competing solution while simultaneously enhancing our editors' ability to manage and analyse content performance. Beyond the numbers, the collaboration with the Recombee team has been excellent, they helped us push our thinking and we were delighted to jointly present at the RecSys conference and be recognized as an INMA '26 finalist for "Best Use of Generative AI"." ![Tom Kelleher](https://www.recombee.com/img/customers/the-telegraph.png) Tom KelleherDirector of Emerging Technology – AI & Personalisation at The Telegraph "Our experience with Recombee has been exceptional. Integration was one of the most painless third-party implementations we’ve done, and the flexibility of the system allows us to continuously fine-tune discovery across our platform. Recombee’s ability to handle our catalog scale while delivering high-quality, adaptive recommendations makes it a critical part of our growth strategy." ![David Penner](https://www.recombee.com/img/customers/beatstars.png) David PennerVP Engineering at Beatstars "Daily traffic to the 9GAG site amounts to **hundreds of millions of users worldwide.** To provide a **unique user experience** for each and every one of them, we needed to implement a recommendation system that would allow for the most advanced, real-time personalization of their experience. So we chose Recombee. Thanks to their solution, we managed to increase, among other KPIs, **post views by 37%, overall interactions by 22%, and users' session duration by 7.5%**. With Recombee, we've taken engagement with user-generated content to the next level." ![Kristie Chen](https://www.recombee.com/img/customers/9gag.png) Kristie ChenProduct Head at 9GAG "**Working with Recombee** to develop an affordable solution to provide our users with excellent music recommendations has **exceeded our expectations in every way.** They have been able to understand the relationships in the data of our industry and create **effective models to use efficiently.** We also appreciate the **nuance and flexibility** they offer when it comes to deciding the right solution based on quality, cost, complexity, speed, and other factors." ![Ty Wangsness](https://www.recombee.com/img/customers/audiomack-1.png) Ty WangsnessFounder/CTO at Audiomack "Our collaboration with Recombee has supported our platform's capabilities through its intelligent personalization algorithms and sophisticated search results. Our customers now receive highly relevant property recommendations that cater to their specific needs, with one notable email campaign seeing a 178% uplift in CTOR. Their commitment to excellence is evident in the 40% increase in listing engagements on our platform, contributing to our growth in the competitive real estate market. The team at Recombee is responsive, professional, and puts in the effort to ensure that our unique business model is supported." ![Larkin Magner](https://www.recombee.com/img/customers/crexi-1.png) Larkin MagnerDirector of Product Management at Crexi "Striving to be the ever limitless music sharing and discovery platform, we need to make sure the user experience of our listeners is smooth and sound. And one of the most critical aspects of achieving such a goal is content personalization tailored 1:1 in real-time. That's why we switched to Recombee. Thanks to their recommender engine, our **monthly plays increased by 206%** and **weekly follows by 67%**. Because the recommendations performed so well, we moved them from our Search page to the top of our main Discover tab. They are now **the best-performing module within that tab, accounting for 46% of all plays**." ![Christopher Dalla Riva](https://www.recombee.com/img/customers/audiomack-2.png) Christopher Dalla RivaSenior Product Manager at Audiomack "Instead of the lengthy and costly process of building an in-house personalization solution, we seamlessly implemented Recombee's AI-powered recommender engine to improve our services. Applying their tailored scenarios and boosters, we have registered steady improvements ever since, with a **6% increase in 'Add to Cart' and a 37% increase in 'Place Bid'**. Thanks to their easy and intuitive integration, Recombee was an obvious choice from the range of personalization solutions." ![Vincent van Leeuwen](https://www.recombee.com/img/customers/reliving.png) Vincent van LeeuwenCo-Founder & CTO/CPO at Reliving "Recombee was able to handle our very specific use case around providing recommendations with a highly volatile inventory of user-generated content. Placing recommendations on our homepage was a huge success: **70%+ higher product detail page views and 30%+ higher clickthroughs**. The Recombee team is a great partner in helping solve our unique use cases, and we look forward to continue working with them." ![Daniel Uhm](https://www.recombee.com/img/customers/slickdeals.png) Daniel UhmProduct Manager at Slickdeals "Recombee is capable of scaling the service and **keeps pace with our rapid growth. Constant innovation and proactive** development of new features makes our collaboration smooth and pleasant." ![Meindert van der Meulen](https://www.recombee.com/img/customers/showmax.png) Meindert van der MeulenHead of Strategy at Showmax "Recombee is an amazing recommendation engine which we use for personalizing different parts on our website, including homepage, product detail page, and search. With their solution, we managed to increase our conversion rate by 14% and shopping cart volume by 8%. A great partnership and looking forward to improving our customer journey even more." ![Dennis Ostner](https://www.recombee.com/img/customers/kunzmann.png) Dennis OstnerHead of E-commerce at Robert Kunzmann GmbH & Co. "We conducted A/B testings of multiple recommendation engines to find the best content personalization solution. Out of all solutions, only **Recombee outperformed our internal read-next recommendations** of news articles. After long-lasting A/B testing, Recombee achieved a **40% higher CTR of suggested articles,** which ultimately led to the deployment of the solution to most of our news sites (iDNES, Lidovky, Expres)." ![Petr Kelin](https://www.recombee.com/img/customers/mafra.png) Petr KelinManager at MAFRA, a.s. "Recombee allows us to modify how we want the recommendations to behave across our platforms in very specific use cases. The implementation of Recombee on the VOD platform prima+ helped us to increase video views by 34% and ad views by 73%, resulting in a significant rise in advertising revenue. Another major success has been the growth of recirculation, which is our top priority on online magazine platforms. We are currently discussing extending their recommendations also to our emails. Highly valued partnership!" ![Jan Lajka](https://www.recombee.com/img/customers/ftv-prima.png) Jan LajkaChief Data Officer at FTV Prima "Thanks to Recombee's recommendation service and Geneea's NLP, we were able to personalize news and articles for visitors of our major portals (aktualne.cz, volny.cz, atlas.cz), **increasing the number of pageviews by 64 percent**. We expand recommendations to other scenarios such as video recommendations or personalized galleries." ![Vojtech Kostelecky](https://www.recombee.com/img/customers/economia.png) Vojtech KosteleckyProduct Manager at Economia "As a huge **e-commerce site with millions of users** we were looking for a stable and reliable partner that would back up our robust item catalogue and high traffic. The integration was simple and quick with **incredible support assistance** from Recombee when we needed it. Recombee, with its **impressive real-time product recommendations**, provides us with a perfect personalization solution to accomplish our KPIs and **improve our customer satisfaction**. Recombee solution currently **drives an impressive 24% of Konga's revenues.**" ![Andrew Mori](https://www.recombee.com/img/customers/kongacom.png) Andrew MoriTech. Director at Konga.com "Prior to Recombee, we used a general recommendation algorithm based on popularity and date published. Since moving our recommendation system to Recombee, we've seen a **50% increase in click-through across our 5 media brands (millions of readers per month).** Recombee was easy to integrate, test, and deploy within just a couple of hours." ![Haydn Strauss](https://www.recombee.com/img/customers/unfiltered-media-group.png) Haydn StraussCOO at Unfiltered Media Group "It has been really difficult to find a solution that could integrate many different data sources and be 100% customizable, but with Recombee we get this. Recombee has fundamentally changed how we are serving recommendations and has really helped us grow. In addition to this, they have great customer support and are always ready to help." ![Nicholas Blicker Larsen](https://www.recombee.com/img/customers/design-group.png) Nicholas Blicker LarsenCEO at Design Group "Recombee is easy to use, the support is fantastic and costs are affordable. We changed our previous recommendation engine with good improvements without efforts." ![Giovanni Bartoli](https://www.recombee.com/img/customers/casa-cenina.png) Giovanni BartoliCEO at Casa Cenina "Recombee's recommendation solution is incredible. Our e-commerce platform can recommend products with much more intelligence for our customers. The support is also fantastic!" ![Lucas Colett](https://www.recombee.com/img/customers/bubb-store.png) Lucas ColettCEO at BUBB.Store "Surprisingly easy to integrate yet powerful." ![Mikael Setterberg](https://www.recombee.com/img/customers/keyflow.png) Mikael SetterbergCTO at Keyflow "If you are looking for a flexible, well documented and powerful recommendation engine, Recombee is definitely the best option in the market. Based on our tests, Recombee provided up to 19% lift in recommendation revenue and increased the conversion rate by 12%" ![Rodrigo Bendia](https://www.recombee.com/img/customers/chicorei.png) Rodrigo BendiaCTO at ChicoRei "Recombee does a fantastic job of personalizing recipe recommendations and **optimizing search results** for our users in the Cooklist app. Their documentation is good and the support team is super helpful. Highly recommend their service to anyone looking to personalize their service." ![Daniel Vitiello](https://www.recombee.com/img/customers/cooklist.png) Daniel VitielloCEO at Cooklist "We use Recombee for the personalization of the user experience on our website, focusing mainly on boosting shopping cart upsell via product offerings. Even though we've only scratched the surface of Recombee's potential with only 14% of all detailed views, it's **already driving 30% of all purchases and revenue**." ![Bartłomiej Gajda](https://www.recombee.com/img/customers/savicki.png) Bartłomiej GajdaBoard Member, IT operations at Savicki.pl "As a founder of a new age media platform, increasing customer engagement is our number one goal and Recombee helped us in the same with their fantastic recommendation engine. NewsBytes has been working with Recombee for more than 6 months now and we are extremely happy with Recombee's product." ![Sumedh Chaudhry](https://www.recombee.com/img/customers/newsbytes.png) Sumedh ChaudhryCEO at NewsBytes "Recombee's service is an essential part of the user experience on my site that sells pop culture merchandise. With many different categories and attributes of the products, the AI recommendations are the best way to offer my customers the best products for them at the right time. After the integration of the service, there was a **17% increase in page views as well as in sales where we saw a conversion rate of more than 95%**. I am very satisfied with their product and service. " ![Borislav Ivanov](https://www.recombee.com/img/customers/geekybg.png) Borislav IvanovCo-owner at Geeky.bg "We had seen significantly better conversions of users to paying customers on Wishbook app, a part of which was due to the similar product recommendations & relevant products for the user being served on the app by Recombee." ![Arvind Saraf](https://www.recombee.com/img/customers/wishbook-infoservices.png) Arvind SarafFounder, CEO & CTO at Wishbook Infoservices "Recombee recommendation service is **surprisingly powerful!** We were able to personalize events to match our customer interests. Recombee not only has a great support team but is also easy to use, customizable, and well documented." ![Pathompon Jirawanidchakorn (Boat)](https://www.recombee.com/img/customers/eventpop.png) Pathompon Jirawanidchakorn (Boat)CTO at Eventpop "Recombee will bring your sales platform to life. It's easy to integrate, customizable and provides timely recommendations. All this combined with a great support team." ![Christian Medina](https://www.recombee.com/img/customers/osigu.png) Christian MedinaData Scientist at Osigu ### Read more on [![](https://www.recombee.com/img/capterra.svg)](https://www.capterra.com/p/163489/Recombee/#reviews) [![](https://www.recombee.com/img/g2.svg)](https://www.g2.com/products/recombee/reviews#reviews) ## Recombee Is Used All Around the World ![The Telegraph](https://www.recombee.com/img/logos/w/the-telegraph.svg)![DAZN](https://www.recombee.com/img/logos/w/dazn.svg)![9GAG](https://www.recombee.com/img/logos/w/9gag.svg)![Cascade PBS](https://www.recombee.com/img/logos/w/cascade-pbs.svg)![Tripadvisor](https://www.recombee.com/img/logos/w/tripadvisor.svg)![Audiomack](https://www.recombee.com/img/logos/w/audiomack.svg)![FTV Prima](https://www.recombee.com/img/logos/w/ftv-prima.svg)![Grow.me](https://www.recombee.com/img/logos/w/grow.svg)![Mediavine](https://www.recombee.com/img/logos/w/mediavine.svg)![Pepper](https://www.recombee.com/img/logos/w/pepper.svg)![Clip It](https://www.recombee.com/img/logos/w/clip-it.svg)![Crexi](https://www.recombee.com/img/logos/w/crexi.svg)![Temple & Webster](https://www.recombee.com/img/logos/w/temple-webster.svg)![Pet Media Group](https://www.recombee.com/img/logos/w/pmg.svg)![Pelando](https://www.recombee.com/img/logos/w/pelando.svg)![Autohaus Kunzmann](https://www.recombee.com/img/logos/w/kunzmann.svg)![Midland Realty](https://www.recombee.com/img/logos/w/midland.svg)![Cruise Critic](https://www.recombee.com/img/logos/w/cruisecritic.svg)![BenefitHub](https://www.recombee.com/img/logos/w/benefithub.svg) and 10,000+ other sites and apps. --- # Driving +35% Increase in Playbacks Across Video Platforms > Source: https://www.recombee.com/case-studies/diagnal > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # [Success Stories](https://www.recombee.com/case-studies) # Driving +35% Increase in Playbacks Across Video Platforms Diagnal [Video](https://www.recombee.com/domains/video) [Content Recommendations](https://www.recombee.com/content-recommendations) ![](https://www.recombee.com/img/case-studies/diagnal.png) To enhance personalization across their OTT client base, **DIAGNAL integrated Recombee’s recommendation engine into their end‑to‑end OTT solution**. The integration powers tailored experiences through DIAGNAL’s Enhance Content Management System and Enlight App Manager solution, helping customers deliver relevant, real-time content across web, mobile, Smart TVs, and set-top boxes. This partnership enabled measurable improvements in user engagement and subscription growth across multiple clients, all without compromising performance, scalability, or speed to market. +35% User Playbacks +9% Paying Subscribers ### Situation DIAGNAL regularly works with leading OTT clients to deliver fast, scalable platforms with rich personalization. For new launches and existing roadmaps, DIAGNAL needed a recommendation engine that could be: * Easily integrated across their modular OTT stack * Fast and responsive at scale * Flexible enough to handle varied use cases * Cost-effective without limiting functionality ### Requirements * Enable AI-driven content personalization in real-time. * Seamlessly integrate into DIAGNAL’s modular architecture. * Meet a fast deployment timeline across several markets and device types. * Increase user engagement and drive subscriber conversion. * Support a wide range of recommendation scenarios across regions and content types. ### Solution Recombee’s recommendation engine was fully integrated into DIAGNAL’s **Enhance Content Management System** and **Enlight App Manager**, to provide fast and flexible personalization across different client platforms. Real-Time Personalization Content rails update instantly based on user behavior, preferences, and viewing patterns. Flexible Use Cases Recommendations adapted to various needs; episodic content, newly trending titles, and different regional catalogs. Operator-Friendly Dashboard Easy-to-use interface enabled efficient tracking of recommendation performance across markets and clients. Scalable Setup The integration was built to support rapid reuse across other DIAGNAL OTT projects . ### DIAGNAL + Recombee Solution for Video Clients DIAGNAL offers a streamlined integration of Recombee’s recommendation engine across its OTT platform. Client apps benefit from a simplified architecture where user interactions and metadata are seamlessly passed from DIAGNAL to Recombee, enabling real-time, AI-driven recommendations with minimal development overhead. ![Recombee + DIAGNAL Solution](https://www.recombee.com/img/case-studies/schema/recombee-diagnal.svg) ### Benefits & Results Results across selected DIAGNAL OTT clients * **+35%** in user playbacks * **+9%** in paying subscribers Easier performance tracking via Recombee’s operator-friendly dashboard. Ongoing support from Recombee’s team for integration and custom use cases. ### Scenarios ![diagnal](https://www.recombee.com/img/case-studies/scenarios/diagnal-1.png) #### Recommended For You Content Recommendations On the homepage, DIAGNAL deploys Recombee-powered “Recommended For You” rails across client platforms. These use a hybrid of collaborative filtering and popularity-based models to generate personalized content suggestions from the first session. The row updates instantly after a user’s first interaction, such as clicking a title, and continues to refine based on ongoing activity. ![diagnal](https://www.recombee.com/img/case-studies/scenarios/diagnal-2.png) #### More Like This Personalized Discovery Through Similar Content To drive continued exploration and surface relevant content, DIAGNAL-powered apps use Recombee to deliver intelligent “More Like This” recommendations. These suggestions appear on title detail pages and present users with similar shows or movies based on their current selection and viewing history. The result is increased content engagement and improved user satisfaction, while keeping the viewing experience personalized and intuitive. ![diagnal](https://www.recombee.com/img/case-studies/scenarios/diagnal-3.png) #### Watch Next Seamless Continuation Post-Playback To maintain user momentum and support continuous viewing sessions, DIAGNAL-powered apps leverage end-of-playback recommendations powered by Recombee. These suggestions appear automatically after a title finishes playing and are designed to align with the viewer’s taste and recent behavior. This not only supports user retention and longer watch times, but also enhances satisfaction by offering content that feels naturally aligned with what the viewer just finished watching. ![diagnal](https://www.recombee.com/img/case-studies/scenarios/diagnal-4.png) #### Personalized Search Intelligent Title Discovery When users search for specific films or shows, DIAGNAL-powered platforms rely on Recombee’s fine-tuned search engine to deliver fast, accurate results. The engine uses real-time popularity signals alongside personalized logic to boost the visibility of the most relevant titles. This helps users find what they’re looking for quickly, while also surfacing unexpected yet fitting suggestions, keeping the overall discovery experience fluid and satisfying. “The Recombee team has been incredibly helpful and supportive in providing Recommendation solutions for the advanced requirements from our customers. They were hands-on throughout the integration and production stages, ensuring our client was satisfied and the integration was healthy.” ![Priyank Mathur](https://www.recombee.com/img/case-studies/testimonials/diagnal.png) **Priyank Mathur** Project Manager & Business Analyst ![Diagnal](https://www.recombee.com/img/logos/diagnal.svg) ### About Diagnal DIAGNAL provides modular, multi-device OTT solutions to global media companies. Their platform allows for rapid deployment, flexible configuration, and deep integration with personalization and analytics tools such as Recombee. [Visit Diagnal](https://www.diagnal.com) **Next Case Studies** [![The Telegraph](https://www.recombee.com/img/case-studies/small-covers/case-studies-the-telegraph.png)+35%CTR uplift on article detail pagesMedia Company](https://www.recombee.com/case-studies/the-telegraph) [![Apify](https://www.recombee.com/img/case-studies/small-covers/case-studies-apify.png)+6%SubscriptionsAI Marketplace](https://www.recombee.com/case-studies/apify) [![9GAG](https://www.recombee.com/img/case-studies/small-covers/case-studies-9gag.png)+37%Post ViewsCross-Platform Entertainment Network](https://www.recombee.com/case-studies/9gag) [![Poslovi Infostud](https://www.recombee.com/img/case-studies/small-covers/case-studies-poslovi-infostud.png)+32%Application conversion on job listingsOnline Job Board](https://www.recombee.com/case-studies/poslovi-infostud) [![Perlego](https://www.recombee.com/img/case-studies/small-covers/case-studies-perlego.png)+44%CTRDigital Learning](https://www.recombee.com/case-studies/perlego) [![Slickdeals](https://www.recombee.com/img/case-studies/small-covers/case-studies-slickdeals.png)+70%CTR to Detail Page ViewsE-commerce + Deal Aggregators](https://www.recombee.com/case-studies/slickdeals) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima.png)+34%Video views on VOD platformMedia Company](https://www.recombee.com/case-studies/ftv-prima) [![Audiomack](https://www.recombee.com/img/case-studies/small-covers/case-studies-audiomack.png)+206%Monthly plays from recommendationsMusic](https://www.recombee.com/case-studies/audiomack) [![Pet Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-pet-media-group.png)+17%CR from visitor to active buyerP2P Marketplaces](https://www.recombee.com/case-studies/pet-media-group) [![Pepper](https://www.recombee.com/img/case-studies/small-covers/case-studies-pepper.png)+21%Click-outs to affiliate links from websiteDeal Aggregators](https://www.recombee.com/case-studies/pepper) [![Crexi](https://www.recombee.com/img/case-studies/small-covers/case-studies-crexi.png)+40%In buy actions from the recommended propertiesReal Estate](https://www.recombee.com/case-studies/crexi) [![Triola](https://www.recombee.com/img/case-studies/small-covers/case-studies-triola.png)10%Total orders from recommendationsE-commerce](https://www.recombee.com/case-studies/triola) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima-content.png)+10%RecirculationMedia Company](https://www.recombee.com/case-studies/ftv-prima-content) [![Unfiltered Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-unfiltered-media-group.png)+50%CTRMedia Company](https://www.recombee.com/case-studies/unfiltered-media-group) [![Reliving](https://www.recombee.com/img/case-studies/small-covers/case-studies-reliving.png)+37%Place BidE-commerce + P2P Marketplaces](https://www.recombee.com/case-studies/reliving) [![Autohaus Kunzmann](https://www.recombee.com/img/case-studies/small-covers/case-studies-autohaus-kunzmann.png)+14%Conversion RateE-commerce](https://www.recombee.com/case-studies/autohaus-kunzmann) [![Showmax](https://www.recombee.com/img/case-studies/small-covers/case-studies-showmax.png)70Countries and multiple languagesSVOD Service](https://www.recombee.com/case-studies/showmax) [![Mafra](https://www.recombee.com/img/case-studies/small-covers/case-studies-mafra.png)40% higherCTR of suggested articlesMedia Company](https://www.recombee.com/case-studies/mafra) [![itison](https://www.recombee.com/img/case-studies/small-covers/case-studies-itison.png)2 000%ROINewsletter Personalization + Deal Aggregators](https://www.recombee.com/case-studies/itison) [![Cooklist](https://www.recombee.com/img/case-studies/small-covers/case-studies-cooklist.png)+27%CTRE-commerce](https://www.recombee.com/case-studies/cooklist) [![Economia](https://www.recombee.com/img/case-studies/small-covers/case-studies-economia.png)+64%Click Through RateMedia Company](https://www.recombee.com/case-studies/economia) [![Segundamano](https://www.recombee.com/img/case-studies/small-covers/case-studies-segundamano.png)3x more conversionsfrom users who engage with recommendationsP2P Marketplaces](https://www.recombee.com/case-studies/segundamano) [![Design Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-design-group.png)+52%Number of PurchasesE-commerce](https://www.recombee.com/case-studies/design-group) --- # Kunzmann Increases Shopping Cart Volume With Personalization > Source: https://www.recombee.com/case-studies/autohaus-kunzmann > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [Success Stories](https://www.recombee.com/case-studies) # Personalization of Customer Experience Yields Dividends for Autohaus Kunzmann Autohaus Kunzmann [E-commerce](https://www.recombee.com/domains/e-commerce) [Product Recommendations](https://www.recombee.com/product-recommendations) ![](https://www.recombee.com/img/case-studies/autohaus-kunzmann.png) In cooperation with our integration partner Complex, Recombee provides product recommendations to Autohaus Kunzmann - one of the most popular automotive car dealer e-commerce platforms in Germany. The renowned platform has several thousand first places in SERPs as the go-to shop for tuning, interior, original parts, tires, and wheels for renowned car brands. Application of Recombee’s advanced solution based on Complex’s valuable market know-how led to a significant increase in shopping cart volume, click-through rate, and conversion rate. Our fully AI-driven recommendations also provided truly better metrics such as time after search, search depth, and drop in bounce rate. 14% Increase in conversion rate 8% Increase in shopping cart volume 8% Increase in click-through rate ### Situation * A large number of interactions. * Customers distracted from their actual interest by other products listed in the online shop and thus leaving prematurely. * CMS pages consist of a selection of many modules that can be configured and placed freely. ### Requirements * Personalization based on user behavior insights and product attributes. * Recommendation of complementary products (cross-sell). * Set the relevance of the respective products and their display order. * Real-time response in large traffic. * Solve the cold start problem. ### Solution * More than **20 scenarios with fully AI-driven recommendations** on different parts over the website. * **Add-to-cart principle** product recommendations to encourage further purchases. * **Low-price and price-independent** items based on the business rules set by Complex. * Recommending available **products from the same category** (e.g. AMG or Mercedes Benz), that fit user behavior. * **Personalized full-text search results** based on user’s interaction data and metadata. ### Benefits & Results * **14% increase** in conversion rate after implementing Recombee * **8% increase** in shopping cart volume * **8% increase** in click-through rate * Improvement and extension of customer’s shopping session. * Better values in time after search, search depth and drop in bounce rate. * Activating personalization within 0.4 seconds, **avoiding downtimes** ### Scenarios ![autohaus-kunzmann](https://www.recombee.com/img/case-studies/scenarios/autohaus-kunzmann-1.png) #### Search Personalization Combination of search engine and recommender system to narrow individual searches to specific items to save customer’s time. By using our search:personalized logic, leveraging our business rules to filter specific items, and setting up over 120 search synonyms in our Admin UI, Kunzmann managed to improve user experience across different sections of their website. ![autohaus-kunzmann](https://www.recombee.com/img/case-studies/scenarios/autohaus-kunzmann-2.png) #### Recommendations For You When selecting a specific category of products (in this case Rim&Wheels), you will get a list of recommended products, which are provided to you with the recombee:default logic. The goal of this logic (unique ensemble of models) is to offer the most relevant products even when there’s not enough data in the given context. The ensemble is being constantly improved by our own artificial intelligence to adapt to the incoming data. ![autohaus-kunzmann](https://www.recombee.com/img/case-studies/scenarios/autohaus-kunzmann-3.png) #### Other Customers Also Bought The scenario used on the product detail page for recommending alternative products to a currently viewed one by utilizing our ecommerce:similar-products logic. In this case, as in most of the e-commerce use cases, the ensemble of models is used in combination with business rules (filtering cheaper products) to offer similar products which are more expensive (up-sell). ![autohaus-kunzmann](https://www.recombee.com/img/case-studies/scenarios/autohaus-kunzmann-4.png) #### Lifestyle To Match Your Star The scenario used on the article detail page for recommending complementary products leveraging our ecommerce:cross-sell logic. In this case, as in most of the e-commerce use cases, the ensemble of models is used for offering products, that are compatible with the currently viewed product (cross-sell). ![autohaus-kunzmann](https://www.recombee.com/img/case-studies/scenarios/autohaus-kunzmann-5.png) #### You Might Be Also Interested In After putting a specific product into the shopping cart during the checkout, there would be products recommended on the confirmation page with recombee:default logic. Automatically AI optimized ensemble of both content-based and collaborative filtering models backed by popularity-based models to offer products that might increase the shopping cart value. [_![Pascal Pischel](https://www.recombee.com/img/customers/complex.png)_Pascal PischelBusiness Development at Complex GmbH & Co. KG](#customer-complex) [_![Dennis Ostner](https://www.recombee.com/img/customers/kunzmann.png)_Dennis OstnerHead of E-commerce at Robert Kunzmann GmbH & Co.](#customer-kunzmann) "At Complex, we choose our tools very carefully - invest into intensive analysis, test phases and target specific KPIs. With Recombee, we found a compatible match that positively surprised us - easy integration, the API is comparatively simple and easy to understand, great personal support, and any flexibility in the front end that you could wish for." _![Pascal Pischel](https://www.recombee.com/img/customers/complex.png)_Pascal PischelBusiness Development at Complex GmbH & Co. KG "Recombee is an amazing recommendation engine which we use for personalizing different parts on our website, including homepage, product detail page, and search. With their solution, we managed to increase our conversion rate by 14% and shopping cart volume by 8%. A great partnership and looking forward to improving our customer journey even more." _![Dennis Ostner](https://www.recombee.com/img/customers/kunzmann.png)_Dennis OstnerHead of E-commerce at Robert Kunzmann GmbH & Co. ![Complex](https://www.recombee.com/img/logos/complex.png) ### About Complex With 35 years of rich experience in software development, Complex is an internationally renowned strategic business partner based in Aschaffenburg, Germany. The team specializes in cross-sector e-business strategies, future-oriented trends, and developments. Complex assists their customers in numerous areas starting with analyzing their own processes all the way up to the implementation of highly individual ERP and CRM systems which are produced and maintained in their house. [Visit complex-it.de](https://www.complex-it.de) ![Kunzmann](https://www.recombee.com/img/logos/w/kunzmann.svg) ### About Autohaus Kunzmann Established in 1935, the Kunzmann dealership transformed into one of the strongest e- commerce platforms focused on automotive and became a far-reaching employer in the region, recruiting over 1000 employees and 180 trainees every year. Their offer consists of a wide range of products such as tuning, exteriors/interiors, replacement parts & wear parts, tires & wheels, and accessories from reputed brands such as Mercedes-Benz, Brabus, Lorinser, Volkswagen, Smart, and AMG. [Visit kunzmann.de](https://www.kunzmann.de/shop/de/) **Next Case Studies** [![The Telegraph](https://www.recombee.com/img/case-studies/small-covers/case-studies-the-telegraph.png)+35%CTR uplift on article detail pagesMedia Company](https://www.recombee.com/case-studies/the-telegraph) [![Apify](https://www.recombee.com/img/case-studies/small-covers/case-studies-apify.png)+6%SubscriptionsAI Marketplace](https://www.recombee.com/case-studies/apify) [![Diagnal](https://www.recombee.com/img/case-studies/small-covers/case-studies-diagnal.png)+35%Playbacks Across Video PlatformsVideo Streaming](https://www.recombee.com/case-studies/diagnal) [![9GAG](https://www.recombee.com/img/case-studies/small-covers/case-studies-9gag.png)+37%Post ViewsCross-Platform Entertainment Network](https://www.recombee.com/case-studies/9gag) [![Poslovi Infostud](https://www.recombee.com/img/case-studies/small-covers/case-studies-poslovi-infostud.png)+32%Application conversion on job listingsOnline Job Board](https://www.recombee.com/case-studies/poslovi-infostud) [![Perlego](https://www.recombee.com/img/case-studies/small-covers/case-studies-perlego.png)+44%CTRDigital Learning](https://www.recombee.com/case-studies/perlego) [![Slickdeals](https://www.recombee.com/img/case-studies/small-covers/case-studies-slickdeals.png)+70%CTR to Detail Page ViewsE-commerce + Deal Aggregators](https://www.recombee.com/case-studies/slickdeals) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima.png)+34%Video views on VOD platformMedia Company](https://www.recombee.com/case-studies/ftv-prima) [![Audiomack](https://www.recombee.com/img/case-studies/small-covers/case-studies-audiomack.png)+206%Monthly plays from recommendationsMusic](https://www.recombee.com/case-studies/audiomack) [![Pet Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-pet-media-group.png)+17%CR from visitor to active buyerP2P Marketplaces](https://www.recombee.com/case-studies/pet-media-group) [![Pepper](https://www.recombee.com/img/case-studies/small-covers/case-studies-pepper.png)+21%Click-outs to affiliate links from websiteDeal Aggregators](https://www.recombee.com/case-studies/pepper) [![Crexi](https://www.recombee.com/img/case-studies/small-covers/case-studies-crexi.png)+40%In buy actions from the recommended propertiesReal Estate](https://www.recombee.com/case-studies/crexi) [![Triola](https://www.recombee.com/img/case-studies/small-covers/case-studies-triola.png)10%Total orders from recommendationsE-commerce](https://www.recombee.com/case-studies/triola) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima-content.png)+10%RecirculationMedia Company](https://www.recombee.com/case-studies/ftv-prima-content) [![Unfiltered Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-unfiltered-media-group.png)+50%CTRMedia Company](https://www.recombee.com/case-studies/unfiltered-media-group) [![Reliving](https://www.recombee.com/img/case-studies/small-covers/case-studies-reliving.png)+37%Place BidE-commerce + P2P Marketplaces](https://www.recombee.com/case-studies/reliving) [![Showmax](https://www.recombee.com/img/case-studies/small-covers/case-studies-showmax.png)70Countries and multiple languagesSVOD Service](https://www.recombee.com/case-studies/showmax) [![Mafra](https://www.recombee.com/img/case-studies/small-covers/case-studies-mafra.png)40% higherCTR of suggested articlesMedia Company](https://www.recombee.com/case-studies/mafra) [![itison](https://www.recombee.com/img/case-studies/small-covers/case-studies-itison.png)2 000%ROINewsletter Personalization + Deal Aggregators](https://www.recombee.com/case-studies/itison) [![Cooklist](https://www.recombee.com/img/case-studies/small-covers/case-studies-cooklist.png)+27%CTRE-commerce](https://www.recombee.com/case-studies/cooklist) [![Economia](https://www.recombee.com/img/case-studies/small-covers/case-studies-economia.png)+64%Click Through RateMedia Company](https://www.recombee.com/case-studies/economia) [![Segundamano](https://www.recombee.com/img/case-studies/small-covers/case-studies-segundamano.png)3x more conversionsfrom users who engage with recommendationsP2P Marketplaces](https://www.recombee.com/case-studies/segundamano) [![Design Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-design-group.png)+52%Number of PurchasesE-commerce](https://www.recombee.com/case-studies/design-group) --- # Product Recommendations to Achieve Optimal Click-Through Rate > Source: https://www.recombee.com/case-studies/cooklist > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [Success Stories](https://www.recombee.com/case-studies) # Personalized Search for Over a Million Recipes and Products Cooklist [E-commerce](https://www.recombee.com/domains/e-commerce) [Product Recommendations](https://www.recombee.com/product-recommendations), [Search Personalization](https://www.recombee.com/where-to-use#full-text-search) ![](https://www.recombee.com/img/case-studies/cooklist.png) Cooklist is a mobile app that keeps track of bought groceries and recommends recipes based on the purchased items. More than a hundred thousand users of the Cooklist mobile app access millions of recipes using products available in major retailers like Kroger, Walmart, and Target at over 20,000 locations nationwide. Recombee is helping Cooklist to optimize search results and personalize recipe recommendations for individual users. In addition, Recombee is able to incorporate those ingredients the user already has into the proposed recipes. Recombee was implemented in the core of the product leading to a 27% increase in click-throughs and a significantly increased number of page views. 27% Increase in CTR 1,000,000+ Recipes & products ### Situation * Over 1,000,000 recipes and products. * Hundreds of thousands of users. * Desire to find the right personalization tool that is able to match recipes with purchased groceries. ### Requirements * Personalization of recipe recommendations based on user behavior insights and the groceries they bought. * An engine that can both personalize the content and optimize the search results. * Real-time response in high traffic volumes. ### Solution * Personalized full-text search based on user interaction data and metadata. * Automatic personalized recommendations applied to recipes. * An ensemble of collaborative filtering and content-based modelling applied to achieve the optimal click-through rate. ### Benefits & Results * **27% increase** in click-through rate. * **Over a million** of recipes are personalized for users in real-time. * **A significant increase** in page views. ### Scenarios ![cooklist](https://www.recombee.com/img/case-studies/scenarios/cooklist-1.png) #### Search Personalization To personalize searches in the Cooklist app, a combination of search engine and recommender system was used. This solution improves the relevance of search results and helps narrow the individual searches to specific items. Recombee search personalization also works with synonyms and typos. Cooklist utilizes default settings using search:personalized logic that considers not only the search query but also personal user preferences. ![cooklist](https://www.recombee.com/img/case-studies/scenarios/cooklist-2.png) #### Recipes You Can Cook Now The second scenario applied to Cooklist aims to show users recipes that are based on ingredients they have already bought. In this scenario, Cooklist uses advanced business rules (filters) to recommend only those recipes featuring the desired ingredients. The user's ingredients are sent to Recombee in the ReQL (Recombee Query Language) filter. The filter is different for each user and Recombee matches them with tailored recipes in real-time. ![cooklist](https://www.recombee.com/img/case-studies/scenarios/cooklist-3.png) #### Similar Recipes The third scenario recommends alternative recipes to those currently viewed by using recombee:default logic. The items catalog is synced with many different product attributes such as brand, allergen, calories, cuisine, or image. This scenario uses an ensemble of collaborative filters and content-based recommendations. AI continuously optimizes this ensemble to automatically adapt to incoming data. "Recombee does a fantastic job of personalizing recipe recommendations and **optimizing search results** for users of the Cooklist app. Their documentation is good and the **support team is super helpful.** I highly recommend their service to anyone looking to personalize their service." **Daniel Vitiello** CEO at Cooklist.co ![Cooklist](https://www.recombee.com/img/logos/cooklist.png) ### About Cooklist Cooklist introduces entirely new thinking about the relationship between cooking and grocery shopping. The app connects with the user’s grocery store loyalty card and automatically adds all purchased items to a digital pantry. Each individual user is shown tailored recipes matched to their grocery purchases. Over 1 million recipes are filtered and matched to the items they have bought. Cooklist is an early-stage startup that aims to promote a healthier future and is backed by Mercury Fund, TechStars, RevTech, and other top investors. [Visit Cooklist](https://www.cooklist.co) **Next Case Studies** [![The Telegraph](https://www.recombee.com/img/case-studies/small-covers/case-studies-the-telegraph.png)+35%CTR uplift on article detail pagesMedia Company](https://www.recombee.com/case-studies/the-telegraph) [![Apify](https://www.recombee.com/img/case-studies/small-covers/case-studies-apify.png)+6%SubscriptionsAI Marketplace](https://www.recombee.com/case-studies/apify) [![Diagnal](https://www.recombee.com/img/case-studies/small-covers/case-studies-diagnal.png)+35%Playbacks Across Video PlatformsVideo Streaming](https://www.recombee.com/case-studies/diagnal) [![9GAG](https://www.recombee.com/img/case-studies/small-covers/case-studies-9gag.png)+37%Post ViewsCross-Platform Entertainment Network](https://www.recombee.com/case-studies/9gag) [![Poslovi Infostud](https://www.recombee.com/img/case-studies/small-covers/case-studies-poslovi-infostud.png)+32%Application conversion on job listingsOnline Job Board](https://www.recombee.com/case-studies/poslovi-infostud) [![Perlego](https://www.recombee.com/img/case-studies/small-covers/case-studies-perlego.png)+44%CTRDigital Learning](https://www.recombee.com/case-studies/perlego) [![Slickdeals](https://www.recombee.com/img/case-studies/small-covers/case-studies-slickdeals.png)+70%CTR to Detail Page ViewsE-commerce + Deal Aggregators](https://www.recombee.com/case-studies/slickdeals) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima.png)+34%Video views on VOD platformMedia Company](https://www.recombee.com/case-studies/ftv-prima) [![Audiomack](https://www.recombee.com/img/case-studies/small-covers/case-studies-audiomack.png)+206%Monthly plays from recommendationsMusic](https://www.recombee.com/case-studies/audiomack) [![Pet Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-pet-media-group.png)+17%CR from visitor to active buyerP2P Marketplaces](https://www.recombee.com/case-studies/pet-media-group) [![Pepper](https://www.recombee.com/img/case-studies/small-covers/case-studies-pepper.png)+21%Click-outs to affiliate links from websiteDeal Aggregators](https://www.recombee.com/case-studies/pepper) [![Crexi](https://www.recombee.com/img/case-studies/small-covers/case-studies-crexi.png)+40%In buy actions from the recommended propertiesReal Estate](https://www.recombee.com/case-studies/crexi) [![Triola](https://www.recombee.com/img/case-studies/small-covers/case-studies-triola.png)10%Total orders from recommendationsE-commerce](https://www.recombee.com/case-studies/triola) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima-content.png)+10%RecirculationMedia Company](https://www.recombee.com/case-studies/ftv-prima-content) [![Unfiltered Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-unfiltered-media-group.png)+50%CTRMedia Company](https://www.recombee.com/case-studies/unfiltered-media-group) [![Reliving](https://www.recombee.com/img/case-studies/small-covers/case-studies-reliving.png)+37%Place BidE-commerce + P2P Marketplaces](https://www.recombee.com/case-studies/reliving) [![Autohaus Kunzmann](https://www.recombee.com/img/case-studies/small-covers/case-studies-autohaus-kunzmann.png)+14%Conversion RateE-commerce](https://www.recombee.com/case-studies/autohaus-kunzmann) [![Showmax](https://www.recombee.com/img/case-studies/small-covers/case-studies-showmax.png)70Countries and multiple languagesSVOD Service](https://www.recombee.com/case-studies/showmax) [![Mafra](https://www.recombee.com/img/case-studies/small-covers/case-studies-mafra.png)40% higherCTR of suggested articlesMedia Company](https://www.recombee.com/case-studies/mafra) [![itison](https://www.recombee.com/img/case-studies/small-covers/case-studies-itison.png)2 000%ROINewsletter Personalization + Deal Aggregators](https://www.recombee.com/case-studies/itison) [![Economia](https://www.recombee.com/img/case-studies/small-covers/case-studies-economia.png)+64%Click Through RateMedia Company](https://www.recombee.com/case-studies/economia) [![Segundamano](https://www.recombee.com/img/case-studies/small-covers/case-studies-segundamano.png)3x more conversionsfrom users who engage with recommendationsP2P Marketplaces](https://www.recombee.com/case-studies/segundamano) [![Design Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-design-group.png)+52%Number of PurchasesE-commerce](https://www.recombee.com/case-studies/design-group) --- # Personalized Listing to Skyrocket Buy Actions & CTOR > Source: https://www.recombee.com/case-studies/crexi > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [Success Stories](https://www.recombee.com/case-studies) # Personalized Listings for the Fastest-Growing Commercial Real Estate Marketplace Crexi [Real Estate](https://www.recombee.com/domains/real-estate) [Product Recommendations](https://www.recombee.com/product-recommendations) ![](https://www.recombee.com/img/case-studies/crexi.png) Crexi is a **leading technology platform** streamlining the commercial real estate transaction process by connecting buyers, sellers, brokers, and tenants. It offers a **centralized marketplace** for property listings, deal management, data analysis, and marketing, **improving efficiency** and facilitating data-driven **decision-making in the industry.** Recombee **improved** Crexi's platform **on multiple touchpoints** through personalized property suggestions, emailing campaigns, and sophisticated search results **improving** overall **user experience and customer satisfaction.** +40% In buy actions from the recommended properties +10% In buy actions from the search results +178% In CTOR from the personalized emails ### Situation * The growing user base of buyers, brokers, and tenants * 20,000,000+ active users * The growing inventory of commercial properties * 316,000+ spaces for **lease** * 165,000+ properties for **sale** * Increasing competition & need for differentiation ### Requirements * The capability of handling a large volume of attributes, users, and interactions, ensuring a seamless experience as the platform grows. * Adaptability to users' changing preferences and behaviors in real-time. * Matching users to the most relevant properties/spaces using their site activity. * Ability to integrate and process diverse data sources. ### Solution A complex and diverse ensemble of incrementally-trained recommendation models to **satisfy both the buyers and sellers.** * Collaborative Filtering * Content-Based models * Utilizing the latest deep-learning approaches * Reinforcement Learning (Contextual Bandits) Development of multiple **ReQL geographical functions** to exactly match Crexi's product vision and expectations for search behavior. * Geographical Points, Polygons, and Radii * Geographical Containment **Optimization for multiple diverse “buy actions”** including: * “Contact Broker”, * “Submit an Offer”, * “Download OM\*”, * “Request Due Diligence”, and others. \*Offering memorandum is a key legal document used in the private placement of commercial real estate. ### Benefits & Results * **Similar Properties Personalization** * **40% increase** in buy actions from the recommended properties * **Search Personalization** * **10% increase** in buy actions from the search results * **Emailing** * **178% increase** in click-to-open rate (CTOR) from the personalized emails * **14% increase** in lease actions from users who received the personalized emails ### Scenarios ![crexi](https://www.recombee.com/img/case-studies/scenarios/crexi-1.png) #### Similar Properties/Spaces An **automatically curated list of similar listings** that closely match the currently shown property and user’s requirements. By leveraging **advanced machine learning algorithms,** the system analyzes various data points, such as location, property type, price range, etc. to provide precise recommendations. **+40%** in buy actions from the recommended properties ![crexi](https://www.recombee.com/img/case-studies/scenarios/crexi-2.png) #### Search Personalization Recombee, when integrated with real-time filtering based on **arbitrarily created spherical polygons** represented by geographical coordinates, can provide personalized recommendations tailored to client’s preferences and requirements. **Every single movement** of the map, zooming in or adjusting the filter, leads to a **completely new personalized list** of commercial properties. This results in an enhanced user experience that combines precise search areas with dynamically generated, **relevant property suggestions.** **+10%** in buy actions from the search results ![crexi](https://www.recombee.com/img/case-studies/scenarios/crexi-3.png) #### Recommended Properties Email marketing campaigns are an essential tool to reach out to prospective clients and maintain connections with existing ones. Delivering a personalized, targeted list of properties and spaces that resonate with the recipient's needs by **utilizing customer data from the website,** such as preferences and search behavior. **+178%** in click-to-open rate (CTOR) from the personalized emails **+14%** in lease actions from users who received the personalized emails “Our collaboration with Recombee has supported our platform's capabilities through its intelligent personalization algorithms and sophisticated search results. Our customers now receive highly relevant property recommendations that cater to their specific needs, with one notable email campaign seeing a 178% uplift in CTOR. Their commitment to excellence is evident in the 40% increase in listing engagements on our platform, contributing to our growth in the competitive real estate market. The team at Recombee is responsive, professional, and puts in the effort to ensure that our unique business model is supported.” ![Larkin Magner](https://www.recombee.com/img/case-studies/testimonials/crexi-1.png) **Larkin Magner** Director of Product Management at Crexi ![Crexi](https://www.recombee.com/img/logos/crexi.svg) ### About Crexi Commercial Real Estate Exchange, Inc. (Crexi) is revolutionizing how commercial real estate professionals transact by **accelerating deal velocity** and **democratizing access to properties** and industry data. The marketplace has been **transforming the industry since 2015,** creating a hub for stakeholders to market, analyze, and trade commercial property. With **millions of users** and **over 400,000 active commercial listings,** Crexi is the **fastest-growing** commercial real estate marketplace in the world. [Visit Crexi](https://www.crexi.com/) **Next Case Studies** [![The Telegraph](https://www.recombee.com/img/case-studies/small-covers/case-studies-the-telegraph.png)+35%CTR uplift on article detail pagesMedia Company](https://www.recombee.com/case-studies/the-telegraph) [![Apify](https://www.recombee.com/img/case-studies/small-covers/case-studies-apify.png)+6%SubscriptionsAI Marketplace](https://www.recombee.com/case-studies/apify) [![Diagnal](https://www.recombee.com/img/case-studies/small-covers/case-studies-diagnal.png)+35%Playbacks Across Video PlatformsVideo Streaming](https://www.recombee.com/case-studies/diagnal) [![9GAG](https://www.recombee.com/img/case-studies/small-covers/case-studies-9gag.png)+37%Post ViewsCross-Platform Entertainment Network](https://www.recombee.com/case-studies/9gag) [![Poslovi Infostud](https://www.recombee.com/img/case-studies/small-covers/case-studies-poslovi-infostud.png)+32%Application conversion on job listingsOnline Job Board](https://www.recombee.com/case-studies/poslovi-infostud) [![Perlego](https://www.recombee.com/img/case-studies/small-covers/case-studies-perlego.png)+44%CTRDigital Learning](https://www.recombee.com/case-studies/perlego) [![Slickdeals](https://www.recombee.com/img/case-studies/small-covers/case-studies-slickdeals.png)+70%CTR to Detail Page ViewsE-commerce + Deal Aggregators](https://www.recombee.com/case-studies/slickdeals) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima.png)+34%Video views on VOD platformMedia Company](https://www.recombee.com/case-studies/ftv-prima) [![Audiomack](https://www.recombee.com/img/case-studies/small-covers/case-studies-audiomack.png)+206%Monthly plays from recommendationsMusic](https://www.recombee.com/case-studies/audiomack) [![Pet Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-pet-media-group.png)+17%CR from visitor to active buyerP2P Marketplaces](https://www.recombee.com/case-studies/pet-media-group) [![Pepper](https://www.recombee.com/img/case-studies/small-covers/case-studies-pepper.png)+21%Click-outs to affiliate links from websiteDeal Aggregators](https://www.recombee.com/case-studies/pepper) [![Triola](https://www.recombee.com/img/case-studies/small-covers/case-studies-triola.png)10%Total orders from recommendationsE-commerce](https://www.recombee.com/case-studies/triola) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima-content.png)+10%RecirculationMedia Company](https://www.recombee.com/case-studies/ftv-prima-content) [![Unfiltered Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-unfiltered-media-group.png)+50%CTRMedia Company](https://www.recombee.com/case-studies/unfiltered-media-group) [![Reliving](https://www.recombee.com/img/case-studies/small-covers/case-studies-reliving.png)+37%Place BidE-commerce + P2P Marketplaces](https://www.recombee.com/case-studies/reliving) [![Autohaus Kunzmann](https://www.recombee.com/img/case-studies/small-covers/case-studies-autohaus-kunzmann.png)+14%Conversion RateE-commerce](https://www.recombee.com/case-studies/autohaus-kunzmann) [![Showmax](https://www.recombee.com/img/case-studies/small-covers/case-studies-showmax.png)70Countries and multiple languagesSVOD Service](https://www.recombee.com/case-studies/showmax) [![Mafra](https://www.recombee.com/img/case-studies/small-covers/case-studies-mafra.png)40% higherCTR of suggested articlesMedia Company](https://www.recombee.com/case-studies/mafra) [![itison](https://www.recombee.com/img/case-studies/small-covers/case-studies-itison.png)2 000%ROINewsletter Personalization + Deal Aggregators](https://www.recombee.com/case-studies/itison) [![Cooklist](https://www.recombee.com/img/case-studies/small-covers/case-studies-cooklist.png)+27%CTRE-commerce](https://www.recombee.com/case-studies/cooklist) [![Economia](https://www.recombee.com/img/case-studies/small-covers/case-studies-economia.png)+64%Click Through RateMedia Company](https://www.recombee.com/case-studies/economia) [![Segundamano](https://www.recombee.com/img/case-studies/small-covers/case-studies-segundamano.png)3x more conversionsfrom users who engage with recommendationsP2P Marketplaces](https://www.recombee.com/case-studies/segundamano) [![Design Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-design-group.png)+52%Number of PurchasesE-commerce](https://www.recombee.com/case-studies/design-group) --- # Design Group Lifts Purchases by 52% With Personalization > Source: https://www.recombee.com/case-studies/design-group > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [Success Stories](https://www.recombee.com/case-studies) # How Machine Learning Can Be Used for Boosting Ecommerce KPIs Design Group [E-commerce](https://www.recombee.com/domains/e-commerce) [Product Recommendations](https://www.recombee.com/product-recommendations) ![](https://www.recombee.com/img/case-studies/design-group.png) With the strategic goal to fully **personalize its offer and help their shoppers to discover products they desire in an efficient way Design Group** felt it was necessary to ‘step up’ the personalization game and chose Recombee to deliver a complex recommendation solution tailored to their needs. +52% Number of Purchases ### Situation * +10.000 Products. * Great focus on shopper experience. * Need to personalize Mooding’s product offer. ### Requirements * Recognize preferences of individual users. * Balance personalization and discovery of inspiring products. * Find similar products. * Real-time response in large traffic. ### Solution * Automatic personalized recommendations applied to product view (Also See and Other Interesting Products sections) as well as shopping cart (Others Have Also Bought section). * Model for every single user with real-time updates. * Automated feature engineering. ### Benefits & Results * **52% increase in number of purchases** since the implementation of Recombee solution. * **Increasing relevancy** of recommendations for individual store visitors. * **Savings in shop’s employees’ time** previously spent on continuous manual product selection. * **Savings in money** of spent on manual content selection. * **Positive impact** on shopper experience. ### Scenarios ![design-group](https://www.recombee.com/img/case-studies/scenarios/design-group-1.png) #### Product Page Recombee applied to Also see and Other interesting products sections. ![design-group](https://www.recombee.com/img/case-studies/scenarios/design-group-2.png) #### Shopping Cart Recombee applied to You have just seen and Others have also bought sections. “We have been looking all over for a flexible and powerful recommendation engine for our ecommerce sites. It has been really difficult to find a solution that could integrate many different data sources and be 100% customizable, but with Recombee we get this. Recombee has fundamentally changed how we are serving recommendations and has really helped us grow. In addition to this, they have great customer support and are always ready to help.” **Nicholas Blicker Larsen** CEO, Design Group ![Design Group](https://www.recombee.com/img/logos/design-group-cs.png) ### About Design Group Design Group is Danish retailer, encompassing leading online stores focusing on designers’ products (Moodings.com, justspotted.dk). While Moodings.com is online furniture store headquartered in Denmark, offering wide selection (+ 10.000 items) of interior design products from Danish as well as international brands, justspotted.dk successfully offers designers graphics and paintings. Both sites operate with one common goal, which is “to offer their clients a wide and inspiring design universe, so they can pick the designs that create mood, identity and expression of their homes.” Moodings, with one of the widest range of interior design in the market, aims to “unite the small designer store and the giant department store on one and the same platform and give the best shopping experience every time shoppers visit their universe”. [**design-group.dk**](http://design-group.dk) [**moodings.com**](http://moodings.com) [**justspotted.dk**](http://justspotted.dk) **Next Case Studies** [![The Telegraph](https://www.recombee.com/img/case-studies/small-covers/case-studies-the-telegraph.png)+35%CTR uplift on article detail pagesMedia Company](https://www.recombee.com/case-studies/the-telegraph) [![Apify](https://www.recombee.com/img/case-studies/small-covers/case-studies-apify.png)+6%SubscriptionsAI Marketplace](https://www.recombee.com/case-studies/apify) [![Diagnal](https://www.recombee.com/img/case-studies/small-covers/case-studies-diagnal.png)+35%Playbacks Across Video PlatformsVideo Streaming](https://www.recombee.com/case-studies/diagnal) [![9GAG](https://www.recombee.com/img/case-studies/small-covers/case-studies-9gag.png)+37%Post ViewsCross-Platform Entertainment Network](https://www.recombee.com/case-studies/9gag) [![Poslovi Infostud](https://www.recombee.com/img/case-studies/small-covers/case-studies-poslovi-infostud.png)+32%Application conversion on job listingsOnline Job Board](https://www.recombee.com/case-studies/poslovi-infostud) [![Perlego](https://www.recombee.com/img/case-studies/small-covers/case-studies-perlego.png)+44%CTRDigital Learning](https://www.recombee.com/case-studies/perlego) [![Slickdeals](https://www.recombee.com/img/case-studies/small-covers/case-studies-slickdeals.png)+70%CTR to Detail Page ViewsE-commerce + Deal Aggregators](https://www.recombee.com/case-studies/slickdeals) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima.png)+34%Video views on VOD platformMedia Company](https://www.recombee.com/case-studies/ftv-prima) [![Audiomack](https://www.recombee.com/img/case-studies/small-covers/case-studies-audiomack.png)+206%Monthly plays from recommendationsMusic](https://www.recombee.com/case-studies/audiomack) [![Pet Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-pet-media-group.png)+17%CR from visitor to active buyerP2P Marketplaces](https://www.recombee.com/case-studies/pet-media-group) [![Pepper](https://www.recombee.com/img/case-studies/small-covers/case-studies-pepper.png)+21%Click-outs to affiliate links from websiteDeal Aggregators](https://www.recombee.com/case-studies/pepper) [![Crexi](https://www.recombee.com/img/case-studies/small-covers/case-studies-crexi.png)+40%In buy actions from the recommended propertiesReal Estate](https://www.recombee.com/case-studies/crexi) [![Triola](https://www.recombee.com/img/case-studies/small-covers/case-studies-triola.png)10%Total orders from recommendationsE-commerce](https://www.recombee.com/case-studies/triola) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima-content.png)+10%RecirculationMedia Company](https://www.recombee.com/case-studies/ftv-prima-content) [![Unfiltered Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-unfiltered-media-group.png)+50%CTRMedia Company](https://www.recombee.com/case-studies/unfiltered-media-group) [![Reliving](https://www.recombee.com/img/case-studies/small-covers/case-studies-reliving.png)+37%Place BidE-commerce + P2P Marketplaces](https://www.recombee.com/case-studies/reliving) [![Autohaus Kunzmann](https://www.recombee.com/img/case-studies/small-covers/case-studies-autohaus-kunzmann.png)+14%Conversion RateE-commerce](https://www.recombee.com/case-studies/autohaus-kunzmann) [![Showmax](https://www.recombee.com/img/case-studies/small-covers/case-studies-showmax.png)70Countries and multiple languagesSVOD Service](https://www.recombee.com/case-studies/showmax) [![Mafra](https://www.recombee.com/img/case-studies/small-covers/case-studies-mafra.png)40% higherCTR of suggested articlesMedia Company](https://www.recombee.com/case-studies/mafra) [![itison](https://www.recombee.com/img/case-studies/small-covers/case-studies-itison.png)2 000%ROINewsletter Personalization + Deal Aggregators](https://www.recombee.com/case-studies/itison) [![Cooklist](https://www.recombee.com/img/case-studies/small-covers/case-studies-cooklist.png)+27%CTRE-commerce](https://www.recombee.com/case-studies/cooklist) [![Economia](https://www.recombee.com/img/case-studies/small-covers/case-studies-economia.png)+64%Click Through RateMedia Company](https://www.recombee.com/case-studies/economia) [![Segundamano](https://www.recombee.com/img/case-studies/small-covers/case-studies-segundamano.png)3x more conversionsfrom users who engage with recommendationsP2P Marketplaces](https://www.recombee.com/case-studies/segundamano) --- # Economia Drives Pageviews With Personalized News > Source: https://www.recombee.com/case-studies/economia > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [Success Stories](https://www.recombee.com/case-studies) # Partnered with Geneea to Personalize News for Economia Economia [Articles, News, Media](https://www.recombee.com/domains/articles-news-media) [Content Recommendations](https://www.recombee.com/content-recommendations) ![](https://www.recombee.com/img/case-studies/economia.png) In a continuous effort to improve its online products, Economia a.s., a Czech media company running a number of news portals and thematic websites, decided to explore the power of artificial intelligence for personalization. Based on a superb A/B test performance, Recombee’s news personalization system together with Geneea’s NLP engine were selected as the winner in a tender of several commercial solutions. Since then, our solution has been expanded to the centrum.cz portal with over 50 million recommendation requests per month. Deployment to several other large sites is planned within the next few months. 64% Increase Conversion Rate ### Situation * Hundreds of articles published every day. * 300,000 historical articles. * Static content selection managed by editors. ### Requirements * Recognize preferences of individual users. * Balance personalization and discovery of new content. * Find similar articles. * Real-time response in large traffic. ### Solution * Automatic personalized recommendations of articles, news and videos. * Model for every single user with real-time updates. * Automated feature engineering. * Integrated NLP for understanding article content to improve recommendations. ### Benefits & Results * **64 % increase in Conversion Rate** compared to recommendations by editors * **Saving editors’ time** previously spent on continuous manual content selection * **Saving money** spent on manual content selection * **Increasing relevancy** of content for individual readers * **Positive impact** on reader satisfaction: more articles read, longer website visits * **Improved advertisement revenue** ### Scenarios ![economia](https://www.recombee.com/img/case-studies/scenarios/economia-1.png) #### Aktuality.cz - Video Detail Recombee applied to Doporucujeme (“Recommended”) section. ![economia](https://www.recombee.com/img/case-studies/scenarios/economia-2.png) #### Aktuality.cz - Home Page Recombee applied to Doporucujeme (“Recommended”) section. ![economia](https://www.recombee.com/img/case-studies/scenarios/economia-3.png) #### Zena.cz - Article Detail Recombee applied to Mohlo by vas zajimat (“You could be interested in”) section. “Thanks to Recombee’s recommendation service and Geneea’s NLP, we were able to personalize news and articles for visitors of our major portals (aktualne.cz, volny.cz, atlas.cz), **increasing the number of pageviews by 64 percent**. We expand recommendations to other scenarios such as video recommendations or personalized galleries.” **Vojtech Kostelecky** Product Manager, economia ![Economia](https://www.recombee.com/img/logos/economia-cs.png) ### About economia Economia, a.s., a European Business Press member, is a major Czech publishing house specializing in economic and professional periodicals. The internet division of the company runs over 20 out of the most important news and professional websites in the country, including aktualne.cz and ihned.cz. Together, they serve 500M page-views per month. Economia takes great pride in the trustworthiness and objectivity of information it provides. It became a reliable source of information for 66 % of Czech business owners, top management and public administration representatives. [Visit economia](http://economia.ihned.cz/) **Next Case Studies** [![The Telegraph](https://www.recombee.com/img/case-studies/small-covers/case-studies-the-telegraph.png)+35%CTR uplift on article detail pagesMedia Company](https://www.recombee.com/case-studies/the-telegraph) [![Apify](https://www.recombee.com/img/case-studies/small-covers/case-studies-apify.png)+6%SubscriptionsAI Marketplace](https://www.recombee.com/case-studies/apify) [![Diagnal](https://www.recombee.com/img/case-studies/small-covers/case-studies-diagnal.png)+35%Playbacks Across Video PlatformsVideo Streaming](https://www.recombee.com/case-studies/diagnal) [![9GAG](https://www.recombee.com/img/case-studies/small-covers/case-studies-9gag.png)+37%Post ViewsCross-Platform Entertainment Network](https://www.recombee.com/case-studies/9gag) [![Poslovi Infostud](https://www.recombee.com/img/case-studies/small-covers/case-studies-poslovi-infostud.png)+32%Application conversion on job listingsOnline Job Board](https://www.recombee.com/case-studies/poslovi-infostud) [![Perlego](https://www.recombee.com/img/case-studies/small-covers/case-studies-perlego.png)+44%CTRDigital Learning](https://www.recombee.com/case-studies/perlego) [![Slickdeals](https://www.recombee.com/img/case-studies/small-covers/case-studies-slickdeals.png)+70%CTR to Detail Page ViewsE-commerce + Deal Aggregators](https://www.recombee.com/case-studies/slickdeals) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima.png)+34%Video views on VOD platformMedia Company](https://www.recombee.com/case-studies/ftv-prima) [![Audiomack](https://www.recombee.com/img/case-studies/small-covers/case-studies-audiomack.png)+206%Monthly plays from recommendationsMusic](https://www.recombee.com/case-studies/audiomack) [![Pet Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-pet-media-group.png)+17%CR from visitor to active buyerP2P Marketplaces](https://www.recombee.com/case-studies/pet-media-group) [![Pepper](https://www.recombee.com/img/case-studies/small-covers/case-studies-pepper.png)+21%Click-outs to affiliate links from websiteDeal Aggregators](https://www.recombee.com/case-studies/pepper) [![Crexi](https://www.recombee.com/img/case-studies/small-covers/case-studies-crexi.png)+40%In buy actions from the recommended propertiesReal Estate](https://www.recombee.com/case-studies/crexi) [![Triola](https://www.recombee.com/img/case-studies/small-covers/case-studies-triola.png)10%Total orders from recommendationsE-commerce](https://www.recombee.com/case-studies/triola) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima-content.png)+10%RecirculationMedia Company](https://www.recombee.com/case-studies/ftv-prima-content) [![Unfiltered Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-unfiltered-media-group.png)+50%CTRMedia Company](https://www.recombee.com/case-studies/unfiltered-media-group) [![Reliving](https://www.recombee.com/img/case-studies/small-covers/case-studies-reliving.png)+37%Place BidE-commerce + P2P Marketplaces](https://www.recombee.com/case-studies/reliving) [![Autohaus Kunzmann](https://www.recombee.com/img/case-studies/small-covers/case-studies-autohaus-kunzmann.png)+14%Conversion RateE-commerce](https://www.recombee.com/case-studies/autohaus-kunzmann) [![Showmax](https://www.recombee.com/img/case-studies/small-covers/case-studies-showmax.png)70Countries and multiple languagesSVOD Service](https://www.recombee.com/case-studies/showmax) [![Mafra](https://www.recombee.com/img/case-studies/small-covers/case-studies-mafra.png)40% higherCTR of suggested articlesMedia Company](https://www.recombee.com/case-studies/mafra) [![itison](https://www.recombee.com/img/case-studies/small-covers/case-studies-itison.png)2 000%ROINewsletter Personalization + Deal Aggregators](https://www.recombee.com/case-studies/itison) [![Cooklist](https://www.recombee.com/img/case-studies/small-covers/case-studies-cooklist.png)+27%CTRE-commerce](https://www.recombee.com/case-studies/cooklist) [![Segundamano](https://www.recombee.com/img/case-studies/small-covers/case-studies-segundamano.png)3x more conversionsfrom users who engage with recommendationsP2P Marketplaces](https://www.recombee.com/case-studies/segundamano) [![Design Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-design-group.png)+52%Number of PurchasesE-commerce](https://www.recombee.com/case-studies/design-group) --- # Content Recommendations Lead to Increase in CTR and Views > Source: https://www.recombee.com/case-studies/ftv-prima > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [Success Stories](https://www.recombee.com/case-studies) # Content Recommendations Across VOD, Magazines and HbbTV for a European Media Leader FTV Prima [Articles, News, Media](https://www.recombee.com/domains/articles-news-media) [Content Recommendations](https://www.recombee.com/content-recommendations) ![](https://www.recombee.com/img/case-studies/ftv-prima.png) FTV Prima faced the challenge of personalizing content **for millions of monthly visitors across diverse platforms.** The manual curation by the editorial staff during a **substantial traffic influx** complicated efforts to ensure a smooth user experience, hindering progress toward the overarching **goal of increasing watch time and improving user recirculation across their various platforms.** **Recombee helped Prima to fully automate its editorial content for multiple platforms** (HbbTV, online magazines, VOD) and deliver personalised content recommendations across its entire ecosystem, while **significantly increasing not only the volume of advertising consumed online, but also other key performance metrics.** +34% **Video views** on VOD platform +73% **Ads views** on VOD platform 2.3x **Higher click-through rate** from Linear TV to HbbTV ### Situation * A very large volume and wide variety of video content available across multiple platforms. * Content recommendations made manually predominantly by editorial staff. * Desire to improve the quality of personalization for millions of monthly visitors. * The huge volume of traffic consisting of millions of monthly visitors. ### Objectives * Increase content discovery and the number of titles consumed. * Increase the time spent consuming content to maximize advertising offers. * Increase visitors' recirculation on online magazines. * Real-time personalized content recommendations for every individual visitor. * Personalization across multiple platforms. ### Solution **Real-time recommendations across multiple platforms** FTV Prima. * Online magazines * VOD platform prima+ * Linear TV (HbbTV) Sophistically **tailored algorithms** to meet the specific requirements for different scenarios. **The unique connection of all platforms** into a single database allows the linking of user profile identities across platforms for unified recommendations. **Automated content recommendations.** Personalized recommendations on **multiple devices** (web, mobile apps). **More than 55 unique scenarios** with fully AI-driven recommendations on different platforms. ### Benefits & Results VOD Platform Data after the launch of prima+ * **+34%** in video views * **+73%** in ads views * **2.5x more** heavy users (more than 3 titles/month) Linear TV to HbbTV * **2.3x increase** in click-through rate Online Magazines * **1.5M click-through** recommendation boxes per month on CNN Prima News * **300k click-through** recommendations boxes per month on Zeny * **+10% recirculation** on CNN Prima News magazine ### Recombee Solution for FTV Prima ![Recombee + FTV Prima Solution](https://www.recombee.com/img/case-studies/schema/recombee-ftv-prima.svg) ### Unique Connection of All FTV Prima Platforms ![Unique Connection](https://www.recombee.com/img/case-studies/schema/recombee-ftv-prima-2.svg) A unified database enables FTV Prima to: * Connect users' profile identities across multiple platforms. * Work with a comprehensive set of data collected from different sources. * Deliver a highly personalized experience wherever the user appears. * Easily manage all recommendations across platforms from one place. ### Use Cases Scenarios for [VOD](#vod) [Magazines](#magazines) [LiveTV (HbbTV)](#livetv-hbbtv) #### VOD Platform Recommendations Real-time personalization of content and search for each user on the VOD platform prima+ with dozens of customized scenarios used in different places. ![ftv-prima](https://www.recombee.com/img/case-studies/scenarios/ftv-prima-1.png) #### Selected For You **Collaborative filtering models** and **popularity-based models** are used to recommend visitors personalized content on the home page right from the start to drive higher engagement. The row is **updated immediately after the user first interacts** (e.g. a click on the specific title to watch), and becomes more tailored as the user interacts with the platform. In the case of titles with episodes, Recombee takes into account the nature of each series in order to recommend the user the **most relevant episode at that particular moment.** ![ftv-prima](https://www.recombee.com/img/case-studies/scenarios/ftv-prima-2.png) #### Personalized Search When a user searches for specific titles, it is crucial to **assist them in finding the most relevant content** as swiftly as possible. This scenario is powered by Recombee's **fine-tuned search engine, which is bolstered by popularity models** to refine search results, ensuring that only the most relevant titles are recommended. ![ftv-prima](https://www.recombee.com/img/case-studies/scenarios/ftv-prima-3.png) #### Similar Movies This scenario is used specifically for similar movie recommendations on a movie details page. **The moment the user selects and clicks on a movie, a detail page for that movie appears and the user is also offered a personalized row showcasing similar titles he might be also interested in.** The similarity is ensured through a combination of tags (e.g., romantic movies), used by Prima to internally describe movies, and **collaborative filtering with content-based models. This approach considers the titles consumed by other users who have also watched the selected movie.** ![ftv-prima](https://www.recombee.com/img/case-studies/scenarios/ftv-prima-4.png) #### Most Watched By utilizing **popularity-based models and various custom business rules,** Recombee creates highly engaging rows that combine the most popular titles in the specified category (e.g., the most popular movies for kids). In this scenario, Recombee also delivers either long-term or short-term popular titles by identifying trending content. **The platforms selected by users (e.g., mobile device or desktop) determine the source of popularity for this scenario.** ##### Key Benefits * Recommending the most viewed content among mobile and desktop users ensures that suggestions are based on the viewing habits and preferences of a similar user demographic. * Recommendation models are improved by analyzing the most viewed content across different device users. ![ftv-prima](https://www.recombee.com/img/case-studies/scenarios/ftv-prima-5.png) #### Advanced Solution for Episodes and Series Recommendations Prima's team fully took advantage of the possibility of customizing the behavior of the recommender with their custom logic. In this scenario, Recombee detects 3 types of series and then recommends combination of episodes in one row, taking into account how the user might consume these types of series: * **Trending series** \- when the user wants to see the newest episode everyone else talks about * **Serialized series** \- when the user wants to continuously watch one episode after another * **Episodic series** \- episodes of a series are not connected to a continuous overarching story, and each episode can be watched independently ##### Custom logic 1. To ensure that users do not miss any news, the **latest episode is always recommended.** 2. **The next unwatched episode in order is recommended,** so the users can seamlessly continue watching another part of the story. 3. **The best-suited episodes are randomly recommended** to users. ##### Key Benefits of Advanced Solution for Episodes Recommendations * **Increase time spent** on the platform, resulting in higher ad views. * **Increase user satisfaction** with a highly personalised experience based on the way how users want to consume the content. * **High** engagement and views. #### Online Magazines Recommendations Driving traffic across Prima's various domains and fostering recirculation through article recommendations. ![ftv-prima](https://www.recombee.com/img/case-studies/scenarios/ftv-prima-6.png) #### Suggested For You Recombee helped FTV Prima to **create a unique interconnection of magazines** to maximize user engagement, recirculation and time spent consuming content. By a combination of text processing, real-time collaborative filtering, and content-based models, Recombee **provides magazine visitors with a mix of recommendations (1,2,3) from other domains** to support the focus on recirculation. The scenario is **used across different domains** including CNN News, Zoom, Cool, Fresh, Zeny and others. ##### Key Benefits of Cross-Domain Recommendations * **Drive** traffic between online magazines with recommendation boxes. * **Increase recirculation.** * **Provide** users with a wide range of content options. * **Increase time spent** on platforms and the number of ads viewed. #### Recommendations in Linear TV to HbbTV Recommendations of titles directly during live broadcasting through the HbbTV interface guide the user to the VOD channels. There, similar titles and shows are personalized based on their historical interactions. ![ftv-prima](https://www.recombee.com/img/case-studies/scenarios/ftv-prima-7.gif) #### Linear TV to HbbTV Recommendations To maintain user engagement with relevant content within the Prima+ ecosystem and prevent them from leaving, Recombee was implemented in a special project. Users watching linear TV were recommended to view similar content at the conclusion of the live broadcast through the interactive HbbTV box. After clicking the red button on their TV controller, users were directed to a recommended VOD channel that closely matched the title on Linear TV, determined by a combination of metadata and custom models. ##### Key Benefits of Cross-Platform Recommendations * **Preventing** users from leaving the ecosystem. * **Encourage** traditional TV users to discover the VOD platform. * **Individually personalised** experience across platforms. “Recombee allows us to modify how we want the recommendations to behave across our platforms in very specific use cases. The implementation of Recombee on the VOD platform prima+ helped us to increase video views by 34% and ad views by 73%, resulting in a significant rise in advertising revenue. Another major success has been the growth of recirculation, which is our top priority on online magazine platforms. We are currently discussing extending their recommendations also to our emails. Highly valued partnership!“ ![Jan Lajka](https://www.recombee.com/img/case-studies/testimonials/ftv-prima.png) **Jan Lajka** Chief Data Officer, FTV Prima ![FTV Prima](https://www.recombee.com/img/logos/w/ftv-prima.svg) ### About FTV Prima FTV Prima is one of **the biggest media companies in central Europe** and currently broadcasts 8 TV channels and 6 radio stations. It has a broad portfolio of websites, print magazines and is also a pioneer in combining VOD with HbbTV. Their most popular platform, prima+, is **visited by millions of viewers per month,** consuming a wide range of content, such as TV series, movies, or documentaries. FTV Prima has also **partnered with CNN International Commercial** and now broadcasts its own channel, CNN Prima News. [Visit iprima.cz](http://iprima.cz/) **Next Case Studies** [![The Telegraph](https://www.recombee.com/img/case-studies/small-covers/case-studies-the-telegraph.png)+35%CTR uplift on article detail pagesMedia Company](https://www.recombee.com/case-studies/the-telegraph) [![Apify](https://www.recombee.com/img/case-studies/small-covers/case-studies-apify.png)+6%SubscriptionsAI Marketplace](https://www.recombee.com/case-studies/apify) [![Diagnal](https://www.recombee.com/img/case-studies/small-covers/case-studies-diagnal.png)+35%Playbacks Across Video PlatformsVideo Streaming](https://www.recombee.com/case-studies/diagnal) [![9GAG](https://www.recombee.com/img/case-studies/small-covers/case-studies-9gag.png)+37%Post ViewsCross-Platform Entertainment Network](https://www.recombee.com/case-studies/9gag) [![Poslovi Infostud](https://www.recombee.com/img/case-studies/small-covers/case-studies-poslovi-infostud.png)+32%Application conversion on job listingsOnline Job Board](https://www.recombee.com/case-studies/poslovi-infostud) [![Perlego](https://www.recombee.com/img/case-studies/small-covers/case-studies-perlego.png)+44%CTRDigital Learning](https://www.recombee.com/case-studies/perlego) [![Slickdeals](https://www.recombee.com/img/case-studies/small-covers/case-studies-slickdeals.png)+70%CTR to Detail Page ViewsE-commerce + Deal Aggregators](https://www.recombee.com/case-studies/slickdeals) [![Audiomack](https://www.recombee.com/img/case-studies/small-covers/case-studies-audiomack.png)+206%Monthly plays from recommendationsMusic](https://www.recombee.com/case-studies/audiomack) [![Pet Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-pet-media-group.png)+17%CR from visitor to active buyerP2P Marketplaces](https://www.recombee.com/case-studies/pet-media-group) [![Pepper](https://www.recombee.com/img/case-studies/small-covers/case-studies-pepper.png)+21%Click-outs to affiliate links from websiteDeal Aggregators](https://www.recombee.com/case-studies/pepper) [![Crexi](https://www.recombee.com/img/case-studies/small-covers/case-studies-crexi.png)+40%In buy actions from the recommended propertiesReal Estate](https://www.recombee.com/case-studies/crexi) [![Triola](https://www.recombee.com/img/case-studies/small-covers/case-studies-triola.png)10%Total orders from recommendationsE-commerce](https://www.recombee.com/case-studies/triola) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima-content.png)+10%RecirculationMedia Company](https://www.recombee.com/case-studies/ftv-prima-content) [![Unfiltered Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-unfiltered-media-group.png)+50%CTRMedia Company](https://www.recombee.com/case-studies/unfiltered-media-group) [![Reliving](https://www.recombee.com/img/case-studies/small-covers/case-studies-reliving.png)+37%Place BidE-commerce + P2P Marketplaces](https://www.recombee.com/case-studies/reliving) [![Autohaus Kunzmann](https://www.recombee.com/img/case-studies/small-covers/case-studies-autohaus-kunzmann.png)+14%Conversion RateE-commerce](https://www.recombee.com/case-studies/autohaus-kunzmann) [![Showmax](https://www.recombee.com/img/case-studies/small-covers/case-studies-showmax.png)70Countries and multiple languagesSVOD Service](https://www.recombee.com/case-studies/showmax) [![Mafra](https://www.recombee.com/img/case-studies/small-covers/case-studies-mafra.png)40% higherCTR of suggested articlesMedia Company](https://www.recombee.com/case-studies/mafra) [![itison](https://www.recombee.com/img/case-studies/small-covers/case-studies-itison.png)2 000%ROINewsletter Personalization + Deal Aggregators](https://www.recombee.com/case-studies/itison) [![Cooklist](https://www.recombee.com/img/case-studies/small-covers/case-studies-cooklist.png)+27%CTRE-commerce](https://www.recombee.com/case-studies/cooklist) [![Economia](https://www.recombee.com/img/case-studies/small-covers/case-studies-economia.png)+64%Click Through RateMedia Company](https://www.recombee.com/case-studies/economia) [![Segundamano](https://www.recombee.com/img/case-studies/small-covers/case-studies-segundamano.png)3x more conversionsfrom users who engage with recommendationsP2P Marketplaces](https://www.recombee.com/case-studies/segundamano) [![Design Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-design-group.png)+52%Number of PurchasesE-commerce](https://www.recombee.com/case-studies/design-group) --- # FTV Prima Boosts Engagement with Content Recommendations > Source: https://www.recombee.com/case-studies/ftv-prima-content > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [Success Stories](https://www.recombee.com/case-studies) # Content Recommendations Across Online Magazines for a European Media Leader FTV Prima [Articles, News, Media](https://www.recombee.com/domains/articles-news-media) [Content Recommendations](https://www.recombee.com/content-recommendations) ![](https://www.recombee.com/img/case-studies/ftv-prima-content.png) FTV Prima faced the challenge of delivering **personalized content across its vast portfolio of online magazines while managing millions of monthly visitors.** The manual curation process by editorial teams struggled to keep pace with growing traffic, affecting user engagement and recirculation efforts. These challenges complicated the overarching **goal of increasing content discovery and maximizing advertising revenue.** **Recombee provided a fully automated solution** to deliver real-time, personalized article recommendations across Prima's online platforms. **The result was a significant boost in content consumption, recirculation, and ad impressions,** all while maintaining editorial oversight. +10% **Recirculation** on CNN Prima News 1.5M **Clicks through recommendation boxes monthly** on CNN Prima News 300k **Clicks through recommendation boxes monthly** on Zeny Higher **Time Spent per visitor** across all magazine platforms ### Situation * A large volume of articles spread across diverse domains such as CNN Prima News and Zeny. * Manual editorial recommendations could not scale to meet the growing audience demand. * Desire to boost recirculation, time spent on site, and ad impressions. ### Objectives * Increase content discovery and consumption. * Enhance user recirculation across related domains. * Deliver personalized recommendations in real time for each visitor. * Maintain editorial oversight while automating recommendations. ### Solution Recombee implemented a robust recommendation engine tailored to FTV Prima’s online magazine platforms. **Real-Time Personalization** Recommendation boxes featured "Suggested for You" sections based on browsing behavior, collaborative filtering, and text analysis. Content updated instantly as users interacted with the platform, ensuring relevant suggestions. **Cross-Domain Recommendations** Articles from different Prima domains were cross-promoted, encouraging users to explore related content across sites like CNN Prima News, Zeny, and Zoom. **Editorial Oversight** Editors retained the ability to adjust recommendations, promote specific articles, and align content with editorial goals. **AI-Powered Text Analysis** Sophisticated algorithms processed article metadata and content tags to enhance contextual relevance. ### Benefits & Results CNN Prima News * **+10%** Recirculation rates * **1.5M Click-throughs monthly** via recommendation boxes Zeny Magazine * **300k Click-throughs monthly** via recommendation boxes * **Improved engagement** with curated lifestyle content. Cross-Domain Synergy * **Unified user profiles** allowed seamless article recommendations across domains. * **Increased** session duration and overall ad impressions. ### Unique Connection of All FTV Prima Platforms ![Unique Connection](https://www.recombee.com/img/case-studies/schema/recombee-ftv-prima-2.svg) A unified database enables FTV Prima to: * **Connect** users' profile identities across multiple platforms. * **Work** with a comprehensive set of data collected from different sources. * **Deliver** a highly personalized experience wherever the user appears. * **Easily manage** all recommendations across platforms from one place. ### Recombee's Solution in Action ![ftv-prima-content](https://www.recombee.com/img/case-studies/scenarios/ftv-prima-content-1.png) #### Suggested For You Personalized recommendations presented a mix of trending, popular, and contextually relevant articles for each visitor. ##### Key Benefits * **Boosted user engagement** by surfacing articles of interest in real time. * **Increased time spent** on-site and repeat visits. ![ftv-prima-content](https://www.recombee.com/img/case-studies/scenarios/ftv-prima-content-2.png) #### Cross-Domain Recommendations Encouraged users to explore related articles from other Prima domains, fostering recirculation. ##### Key Benefits * **Drove traffic** between Prima’s platforms. * **Improved exposure** to diverse content categories. ![ftv-prima-content](https://www.recombee.com/img/case-studies/scenarios/ftv-prima-content-3.png) #### Trending Now Highlighted short-term popular articles to capitalize on emerging interests. ##### Key Benefits * **Increased content consumption** during peak traffic periods. * **Enhanced user satisfaction** by delivering timely and relevant content. ### Editorial Control FTV Prima maintained editorial oversight through Recombee's "humans-in-the-loop" feature. #### Editors Could * **Adjust** algorithm priorities for specific campaigns. * **Boost** visibility of high-priority content. * **Ensure** alignment with Prima's editorial vision. ![Recombee Insights](https://www.recombee.com/img/domains/wwd-insights.png) ### Real-Time Analytics & Insights Recombee’s Insights tool empowers editors and product teams to dive deep into real-time data via the Admin UI. #### Key Features * **Access** detailed insights on recommended articles, popular topics, and more. * **Visualize** data, create custom reports, and easily share with the team. * **Choose** from a range of analytical views or build custom graphs and reports. * Use analytics to adjust rules and align the recommender engine with **your product vision.** [Explore Insights](https://www.recombee.com/features/real-time-analytics-insights) “Recombee allows us to modify how we want the recommendations to behave across our platforms in very specific use cases. A major success has been the growth of recirculation, which is our top priority on online magazine platforms. We are currently discussing extending their recommendations also to our emails. Highly valued partnership!“ ![Jan Lajka](https://www.recombee.com/img/case-studies/testimonials/ftv-prima.png) **Jan Lajka** Chief Data Officer, FTV Prima ![FTV Prima](https://www.recombee.com/img/logos/w/ftv-prima.svg) ### About FTV Prima FTV Prima is one of Central Europe's leading media companies, with a portfolio including **8 TV channels, 6 radio stations,** and a **range of online and print magazines.** Its diverse platforms serve millions of users monthly, offering news, lifestyle, and entertainment content. A unified database enables FTV Prima to: * **Connect** users’ profile identities across multiple platforms * **Work** with a comprehensive set of data collected from different sources * **Deliver** a highly personalized experience wherever the user appears * **Easily manage** all recommendations across platforms from one place [Visit iprima.cz](http://iprima.cz/) **Next Case Studies** [![The Telegraph](https://www.recombee.com/img/case-studies/small-covers/case-studies-the-telegraph.png)+35%CTR uplift on article detail pagesMedia Company](https://www.recombee.com/case-studies/the-telegraph) [![Apify](https://www.recombee.com/img/case-studies/small-covers/case-studies-apify.png)+6%SubscriptionsAI Marketplace](https://www.recombee.com/case-studies/apify) [![Diagnal](https://www.recombee.com/img/case-studies/small-covers/case-studies-diagnal.png)+35%Playbacks Across Video PlatformsVideo Streaming](https://www.recombee.com/case-studies/diagnal) [![9GAG](https://www.recombee.com/img/case-studies/small-covers/case-studies-9gag.png)+37%Post ViewsCross-Platform Entertainment Network](https://www.recombee.com/case-studies/9gag) [![Poslovi Infostud](https://www.recombee.com/img/case-studies/small-covers/case-studies-poslovi-infostud.png)+32%Application conversion on job listingsOnline Job Board](https://www.recombee.com/case-studies/poslovi-infostud) [![Perlego](https://www.recombee.com/img/case-studies/small-covers/case-studies-perlego.png)+44%CTRDigital Learning](https://www.recombee.com/case-studies/perlego) [![Slickdeals](https://www.recombee.com/img/case-studies/small-covers/case-studies-slickdeals.png)+70%CTR to Detail Page ViewsE-commerce + Deal Aggregators](https://www.recombee.com/case-studies/slickdeals) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima.png)+34%Video views on VOD platformMedia Company](https://www.recombee.com/case-studies/ftv-prima) [![Audiomack](https://www.recombee.com/img/case-studies/small-covers/case-studies-audiomack.png)+206%Monthly plays from recommendationsMusic](https://www.recombee.com/case-studies/audiomack) [![Pet Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-pet-media-group.png)+17%CR from visitor to active buyerP2P Marketplaces](https://www.recombee.com/case-studies/pet-media-group) [![Pepper](https://www.recombee.com/img/case-studies/small-covers/case-studies-pepper.png)+21%Click-outs to affiliate links from websiteDeal Aggregators](https://www.recombee.com/case-studies/pepper) [![Crexi](https://www.recombee.com/img/case-studies/small-covers/case-studies-crexi.png)+40%In buy actions from the recommended propertiesReal Estate](https://www.recombee.com/case-studies/crexi) [![Triola](https://www.recombee.com/img/case-studies/small-covers/case-studies-triola.png)10%Total orders from recommendationsE-commerce](https://www.recombee.com/case-studies/triola) [![Unfiltered Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-unfiltered-media-group.png)+50%CTRMedia Company](https://www.recombee.com/case-studies/unfiltered-media-group) [![Reliving](https://www.recombee.com/img/case-studies/small-covers/case-studies-reliving.png)+37%Place BidE-commerce + P2P Marketplaces](https://www.recombee.com/case-studies/reliving) [![Autohaus Kunzmann](https://www.recombee.com/img/case-studies/small-covers/case-studies-autohaus-kunzmann.png)+14%Conversion RateE-commerce](https://www.recombee.com/case-studies/autohaus-kunzmann) [![Showmax](https://www.recombee.com/img/case-studies/small-covers/case-studies-showmax.png)70Countries and multiple languagesSVOD Service](https://www.recombee.com/case-studies/showmax) [![Mafra](https://www.recombee.com/img/case-studies/small-covers/case-studies-mafra.png)40% higherCTR of suggested articlesMedia Company](https://www.recombee.com/case-studies/mafra) [![itison](https://www.recombee.com/img/case-studies/small-covers/case-studies-itison.png)2 000%ROINewsletter Personalization + Deal Aggregators](https://www.recombee.com/case-studies/itison) [![Cooklist](https://www.recombee.com/img/case-studies/small-covers/case-studies-cooklist.png)+27%CTRE-commerce](https://www.recombee.com/case-studies/cooklist) [![Economia](https://www.recombee.com/img/case-studies/small-covers/case-studies-economia.png)+64%Click Through RateMedia Company](https://www.recombee.com/case-studies/economia) [![Segundamano](https://www.recombee.com/img/case-studies/small-covers/case-studies-segundamano.png)3x more conversionsfrom users who engage with recommendationsP2P Marketplaces](https://www.recombee.com/case-studies/segundamano) [![Design Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-design-group.png)+52%Number of PurchasesE-commerce](https://www.recombee.com/case-studies/design-group) --- # itison Boosts Newsletter ROI Through Email Personalization > Source: https://www.recombee.com/case-studies/itison > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [Success Stories](https://www.recombee.com/case-studies) # Hi-Tech Email Product Recommendations For itison itison [Travel, Trips](https://www.recombee.com/domains/travel-trips) [Newsletter Recommendations](https://www.recombee.com/specialized-recommendations) ![](https://www.recombee.com/img/case-studies/itison.png) itison is a leading events and deals website offering their users unparalleled savings on their favorite activities. Recombee has provided itison with a AI-powered recommendation solution for weekly emailing to personalize each individual’s experience and appeal to their interests. 2000% ROI From Newsletter Personalization ### Situation * (+) 4500 Business Partners. * Audience of 1.2 Million Subscribers. * Need to utilize data to conduct mass marketing campaigns. ### Requirements * Need to understanding the machine learning aspect of an online platform * Capable of personalizing weekly deals and event news to subscribers. * Real-time response in large traffic. * The email recommendations must provide hyper-personalized offers for each subscriber to spark their interest. ### Solution * Recombees solution for itison was creating batch recommendations for weekly emailing campaigns. * An ensemble of collaborative filtering and content-based models were implemented to achieve the optimal click through rate. * Real-time model updates under continuous new data inflow will ensure the newest available data are being taken into account when generating emails. ### Benefits & Results * After partnering with Recombee, itison experienced a positive **20-fold (2 000%) return on investment.** * The personalized recommendation emails are creating better experiences for the subscribers and itison. E-commerce **conversion rate is 25% higher** than before. * Weekly personalized emails are being sent to itison’s subscribers, leading to about a **20% increase in traffic.** * Implementation of Recombee’s recommendations led to a **25% increase in e-commerce conversion rate.** ### Scenarios ![itison](https://www.recombee.com/img/case-studies/scenarios/itison-1.png) #### Scenario Example The email recommendations are providing hyper-personalized offers for each subscriber to spark their interest. The model optimization capabilities of Recombee’s recommender solution lead to an increase in conversion rates and improve user experience. “The recommendation powered email outperforms a good number of our editorial emails and it does so consistently. Thanks to Recombee’s recommendation service applied to our personalized emailing we have increased the e-commerce conversion rates by 25%, achieving 2,000% ROI. We are very excited about that.” **Gavin Montague** Head of Development, itison ![itison](https://www.recombee.com/img/logos/itison.png) ### About itison itison is one of the largest deals and events management websites in Scotland. Building on 15 years of experience in the market, itison offers diverse and exciting daily deals and event invitations throughout Scotland and Northern England. Some of these deals include a sight-seeing flight, an award-winning 8 course meal, or 3 nights in a luxury hotel. itison partners with over 4,500 quality businesses to provide an audience of over 1.2M subscribers with the best places and deals in their city through recommendations from the itison experts. itison is responsible for some of the UK’s most successful marketing campaigns. They have created and sold out major events, while conducting global media campaigns. [Visit itison](https://www.itison.com/) **Next Case Studies** [![The Telegraph](https://www.recombee.com/img/case-studies/small-covers/case-studies-the-telegraph.png)+35%CTR uplift on article detail pagesMedia Company](https://www.recombee.com/case-studies/the-telegraph) [![Apify](https://www.recombee.com/img/case-studies/small-covers/case-studies-apify.png)+6%SubscriptionsAI Marketplace](https://www.recombee.com/case-studies/apify) [![Diagnal](https://www.recombee.com/img/case-studies/small-covers/case-studies-diagnal.png)+35%Playbacks Across Video PlatformsVideo Streaming](https://www.recombee.com/case-studies/diagnal) [![9GAG](https://www.recombee.com/img/case-studies/small-covers/case-studies-9gag.png)+37%Post ViewsCross-Platform Entertainment Network](https://www.recombee.com/case-studies/9gag) [![Poslovi Infostud](https://www.recombee.com/img/case-studies/small-covers/case-studies-poslovi-infostud.png)+32%Application conversion on job listingsOnline Job Board](https://www.recombee.com/case-studies/poslovi-infostud) [![Perlego](https://www.recombee.com/img/case-studies/small-covers/case-studies-perlego.png)+44%CTRDigital Learning](https://www.recombee.com/case-studies/perlego) [![Slickdeals](https://www.recombee.com/img/case-studies/small-covers/case-studies-slickdeals.png)+70%CTR to Detail Page ViewsE-commerce + Deal Aggregators](https://www.recombee.com/case-studies/slickdeals) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima.png)+34%Video views on VOD platformMedia Company](https://www.recombee.com/case-studies/ftv-prima) [![Audiomack](https://www.recombee.com/img/case-studies/small-covers/case-studies-audiomack.png)+206%Monthly plays from recommendationsMusic](https://www.recombee.com/case-studies/audiomack) [![Pet Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-pet-media-group.png)+17%CR from visitor to active buyerP2P Marketplaces](https://www.recombee.com/case-studies/pet-media-group) [![Pepper](https://www.recombee.com/img/case-studies/small-covers/case-studies-pepper.png)+21%Click-outs to affiliate links from websiteDeal Aggregators](https://www.recombee.com/case-studies/pepper) [![Crexi](https://www.recombee.com/img/case-studies/small-covers/case-studies-crexi.png)+40%In buy actions from the recommended propertiesReal Estate](https://www.recombee.com/case-studies/crexi) [![Triola](https://www.recombee.com/img/case-studies/small-covers/case-studies-triola.png)10%Total orders from recommendationsE-commerce](https://www.recombee.com/case-studies/triola) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima-content.png)+10%RecirculationMedia Company](https://www.recombee.com/case-studies/ftv-prima-content) [![Unfiltered Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-unfiltered-media-group.png)+50%CTRMedia Company](https://www.recombee.com/case-studies/unfiltered-media-group) [![Reliving](https://www.recombee.com/img/case-studies/small-covers/case-studies-reliving.png)+37%Place BidE-commerce + P2P Marketplaces](https://www.recombee.com/case-studies/reliving) [![Autohaus Kunzmann](https://www.recombee.com/img/case-studies/small-covers/case-studies-autohaus-kunzmann.png)+14%Conversion RateE-commerce](https://www.recombee.com/case-studies/autohaus-kunzmann) [![Showmax](https://www.recombee.com/img/case-studies/small-covers/case-studies-showmax.png)70Countries and multiple languagesSVOD Service](https://www.recombee.com/case-studies/showmax) [![Mafra](https://www.recombee.com/img/case-studies/small-covers/case-studies-mafra.png)40% higherCTR of suggested articlesMedia Company](https://www.recombee.com/case-studies/mafra) [![Cooklist](https://www.recombee.com/img/case-studies/small-covers/case-studies-cooklist.png)+27%CTRE-commerce](https://www.recombee.com/case-studies/cooklist) [![Economia](https://www.recombee.com/img/case-studies/small-covers/case-studies-economia.png)+64%Click Through RateMedia Company](https://www.recombee.com/case-studies/economia) [![Segundamano](https://www.recombee.com/img/case-studies/small-covers/case-studies-segundamano.png)3x more conversionsfrom users who engage with recommendationsP2P Marketplaces](https://www.recombee.com/case-studies/segundamano) [![Design Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-design-group.png)+52%Number of PurchasesE-commerce](https://www.recombee.com/case-studies/design-group) --- # Real-Time Collaborative Filtering & Text Processing > Source: https://www.recombee.com/case-studies/mafra > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [Success Stories](https://www.recombee.com/case-studies) # Outperforming and Replacing In-House Recommender System of News Articles Mafra [Articles, News, Media](https://www.recombee.com/domains/articles-news-media) [Content Recommendations](https://www.recombee.com/content-recommendations) ![](https://www.recombee.com/img/case-studies/mafra.png) Headquartered in Prague, Mafra is the largest media group on the Czech market delivering the latest local and foreign news. Its online activities reach up to 8 million monthly users, with up to 45 million recommendation requests per month. Recombee’s smart content recommendations helped Mafra personalize news to each of their readers across three online media outlets ([iDnes](https://www.idnes.cz/), [Lidovky](https://www.lidovky.cz/), [Expres](https://www.expres.cz/)) and emailing to their premium users. Implementation of Recombee’s sophisticated system has led to a significant increase in both click-through rate and readers’ satisfaction. 40% Higher CTR of suggested articles 25% Improvement in emailing CTOR ### Situation * Desire to find the right content personalization solution (replacing in-house solution) * Potential to deploy across various media brands * Huge traffic consisting of millions of monthly readers ### Requirements * Targeting specific content to the people who will find it most relevant * Retaining consumers’ trust by providing unbiased information and recommendations * Email personalization for iDNES Premium users ### Solution * **Above 65 distinctive scenarios** on different websites and various platforms * **Enriching the cultural life of consumers** by encouraging the discovery of new content and information * **Real-time** collaborative filtering, text processing and reinforcement learning * Overcoming small amount of fully sorted articles to still provide **highly personalized emails** ### Benefits & Results * **Highly outperformed** internal read-next recommender system * **40% increase** in click-through rate of suggested articles * **25% improvement** in emailing click-to-open rate compared to the previous solution * **Time saved** by automatically organized content and **increase in user engagement** ### Scenarios ![mafra](https://www.recombee.com/img/case-studies/scenarios/mafra-1.png) #### Read Next Items to User scenario used across three different media outlets - idnes.cz, lidovky.cz and expres.cz. Personalized recommendations at the bottom of each article to maximize user engagement and important KPIs like ATS (average time spent) and PVs (page views) per session. Thanks to a sophisticated combination of real-time collaborative filtering, text-processing, and reinforcement learning models, Recombee is able to immediately respond to newly published articles, and even automatically identify breaking news, making sure that users are always offered highly relevant content that they shouldn't miss. ![mafra](https://www.recombee.com/img/case-studies/scenarios/mafra-2.png) #### Emailing for Premium Users The scenario used in the emailing campaigns recommending fresh daily premium content for subscribed users. Significant increase of the CTOR, despite a relatively small set of articles being sorted for each user every day (there are about 20 daily Premium articles). The solution benefits from the live recommendations deployed on-site, where a lot of feedback is collected about user engagement with the freshly published articles just before the campaign is executed. Thanks to Recombee real-time collaborative filtering and reinforcement learning models, even a few of the exposure to the newly published articles bring enough data for an excellent performance. “We conducted A/B testings of multiple recommendation engines to find the best content personalization solution. Out of all solutions, only **Recombee outperformed our internal read-next recommendations** of news articles. After long-lasting A/B testing, Recombee achieved **40% higher CTR of suggested articles,** which ultimately led to deployment of the solution to most of our news sites (iDNES, Lidovky, Expres).“ **Petr Kelin** Manager at MAFRA, a.s. ![Mafra](https://www.recombee.com/img/logos/mafra.png) ### About Mafra The MAFRA Media Group reaches its diverse pool of readers and site visitors through print and internet media. Offering exclusive content and quality entertainment with fresh news for the local demography, Mafra provides clarity and easy orientation of the latest events. Their portfolio covers products from every spectrum of the media market, including newspapers, magazines, video portals, tv stations, virtual operators, and radios. [Visit Mafra](http://mafra.cz/) **Next Case Studies** [![The Telegraph](https://www.recombee.com/img/case-studies/small-covers/case-studies-the-telegraph.png)+35%CTR uplift on article detail pagesMedia Company](https://www.recombee.com/case-studies/the-telegraph) [![Apify](https://www.recombee.com/img/case-studies/small-covers/case-studies-apify.png)+6%SubscriptionsAI Marketplace](https://www.recombee.com/case-studies/apify) [![Diagnal](https://www.recombee.com/img/case-studies/small-covers/case-studies-diagnal.png)+35%Playbacks Across Video PlatformsVideo Streaming](https://www.recombee.com/case-studies/diagnal) [![9GAG](https://www.recombee.com/img/case-studies/small-covers/case-studies-9gag.png)+37%Post ViewsCross-Platform Entertainment Network](https://www.recombee.com/case-studies/9gag) [![Poslovi Infostud](https://www.recombee.com/img/case-studies/small-covers/case-studies-poslovi-infostud.png)+32%Application conversion on job listingsOnline Job Board](https://www.recombee.com/case-studies/poslovi-infostud) [![Perlego](https://www.recombee.com/img/case-studies/small-covers/case-studies-perlego.png)+44%CTRDigital Learning](https://www.recombee.com/case-studies/perlego) [![Slickdeals](https://www.recombee.com/img/case-studies/small-covers/case-studies-slickdeals.png)+70%CTR to Detail Page ViewsE-commerce + Deal Aggregators](https://www.recombee.com/case-studies/slickdeals) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima.png)+34%Video views on VOD platformMedia Company](https://www.recombee.com/case-studies/ftv-prima) [![Audiomack](https://www.recombee.com/img/case-studies/small-covers/case-studies-audiomack.png)+206%Monthly plays from recommendationsMusic](https://www.recombee.com/case-studies/audiomack) [![Pet Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-pet-media-group.png)+17%CR from visitor to active buyerP2P Marketplaces](https://www.recombee.com/case-studies/pet-media-group) [![Pepper](https://www.recombee.com/img/case-studies/small-covers/case-studies-pepper.png)+21%Click-outs to affiliate links from websiteDeal Aggregators](https://www.recombee.com/case-studies/pepper) [![Crexi](https://www.recombee.com/img/case-studies/small-covers/case-studies-crexi.png)+40%In buy actions from the recommended propertiesReal Estate](https://www.recombee.com/case-studies/crexi) [![Triola](https://www.recombee.com/img/case-studies/small-covers/case-studies-triola.png)10%Total orders from recommendationsE-commerce](https://www.recombee.com/case-studies/triola) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima-content.png)+10%RecirculationMedia Company](https://www.recombee.com/case-studies/ftv-prima-content) [![Unfiltered Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-unfiltered-media-group.png)+50%CTRMedia Company](https://www.recombee.com/case-studies/unfiltered-media-group) [![Reliving](https://www.recombee.com/img/case-studies/small-covers/case-studies-reliving.png)+37%Place BidE-commerce + P2P Marketplaces](https://www.recombee.com/case-studies/reliving) [![Autohaus Kunzmann](https://www.recombee.com/img/case-studies/small-covers/case-studies-autohaus-kunzmann.png)+14%Conversion RateE-commerce](https://www.recombee.com/case-studies/autohaus-kunzmann) [![Showmax](https://www.recombee.com/img/case-studies/small-covers/case-studies-showmax.png)70Countries and multiple languagesSVOD Service](https://www.recombee.com/case-studies/showmax) [![itison](https://www.recombee.com/img/case-studies/small-covers/case-studies-itison.png)2 000%ROINewsletter Personalization + Deal Aggregators](https://www.recombee.com/case-studies/itison) [![Cooklist](https://www.recombee.com/img/case-studies/small-covers/case-studies-cooklist.png)+27%CTRE-commerce](https://www.recombee.com/case-studies/cooklist) [![Economia](https://www.recombee.com/img/case-studies/small-covers/case-studies-economia.png)+64%Click Through RateMedia Company](https://www.recombee.com/case-studies/economia) [![Segundamano](https://www.recombee.com/img/case-studies/small-covers/case-studies-segundamano.png)3x more conversionsfrom users who engage with recommendationsP2P Marketplaces](https://www.recombee.com/case-studies/segundamano) [![Design Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-design-group.png)+52%Number of PurchasesE-commerce](https://www.recombee.com/case-studies/design-group) --- # Driving up to 477% Higher Onboarding Engagement for a Global Digital Library > Source: https://www.recombee.com/case-studies/perlego > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # [Success Stories](https://www.recombee.com/case-studies) # Driving up to 477% Higher Onboarding Engagement for a Global Digital Library Perlego ![](https://www.recombee.com/img/case-studies/perlego.png) **Perlego** is a digital learning library offering access to over 1.5 million academic, professional, and non-fiction books through a single subscription. Serving students, researchers, and lifelong learners worldwide, the platform helps users discover relevant educational content across a vast and continuously growing catalogue. To boost early activation and long-term retention, Perlego partnered with **Recombee** to deliver reliable, personalized recommendations across key user journeys. +477% in 2+ book saves during onboarding +90% in books saved through homepage recommendation carousels ### Situation * A large and diverse academic content catalog made personalized discovery increasingly difficult to deliver effectively to students and institutions * Existing recommendation system had reliability issues and limited scalability * Heavy engineering dependency slowed iteration and blocked broader use across teams * Need to support multiple personalization use cases across the user journey ### Objectives * Improve reliability and scalability of recommendations * Enable self-serve control for product, marketing, and publisher-driven promotion * Support personalization across key journeys, from onboarding to homepage discovery * Increase early engagement and book-saving behavior as a driver of retention ### Solution Recombee supports personalized content recommendations across Perlego’s digital library. Personalization influences which titles are shown across key touchpoints, from onboarding to homepage discovery, adapting in real time as users interact with the platform. **Fast implementation across key user journeys** A flexible recommendation setup enabled rapid deployment across the sign-up funnel and personalized homepage discovery. Self-serve capabilities reduced engineering dependency and allowed multiple teams to iterate independently across use cases. **Reliable personalization at scale** A robust recommendation engine ensured consistent performance across a large and diverse academic content catalog, eliminating missing or failed recommendations and improving discovery reliability. **Context-aware recommendations from first session** Recommendations use contextual signals such as entry points and early user behavior to deliver relevant suggestions from the first interaction, directly supporting early activation and the key behavior of book saving. **Full control for product, marketing, and publisher-driven promotion** A self-serve setup enables teams to manage filtering, boosting, and experimentation independently. This includes publisher boosting capabilities, allowing strategic promotion of new or priority titles within the library experience while preserving recommendation relevance. **Analytics and performance visibility** Built-in reporting provides visibility into recommendation performance, helping teams understand impact, track engagement and activation metrics, and continuously optimize strategies over time. ### Benefits & Results #### Sign-Up Funnel * **+3%** overall conversion rate * **+283%** users saving at least 1 book * **+477%** users saving at least 2 books * **+375%** users saving 3+ books #### Library Homepage * **+44%** CTR * **+40%** book opens * **+90%** books saved ### Recombee’s Solution in Action ![perlego](https://www.recombee.com/img/case-studies/scenarios/perlego-1.png) ### Sign-up Funnel Personalization Similar Books One of the most important early personalization touchpoints within the sign-up flow, where users are prompted to build their first bookshelf. Recommendations are generated based on contextual signals such as the entry book page and early user intent, ensuring highly relevant suggestions from the first interaction. As users engage, recommendations adapt based on behavioral signals and interaction history, becoming increasingly tailored to their interests. **Key Benefits** * Increase early activation through first-session book saves * Improve onboarding conversion with relevant, contextual recommendations * Strengthen long-term retention by driving early engagement behaviors ![perlego](https://www.recombee.com/img/case-studies/scenarios/perlego-2.png) ### Homepage Personalization Recommended for You A core discovery surface within the library experience, delivering personalized recommendation carousels on the homepage. This scenario combines collaborative filtering with behavioral signals to surface relevant academic content across a large and diverse catalog, helping users continue discovery after login. It also supports controlled content promotion through publisher boosting, enabling selected titles and new releases to be surfaced within the library experience while maintaining relevance. **Key Benefits** * Improve content discovery across a broad catalog * Increase engagement through personalized browsing experiences * Enable controlled content promotion via publisher boosting * Drive ongoing content consumption beyond initial session entry “With Recombee, we were able to roll out personalized recommendations quickly across key journeys like the sign-up funnel and homepage discovery experience. What stood out immediately was how fast we started seeing impact.” ![Julia Unite](https://www.recombee.com/img/case-studies/testimonials/perlego.png) **Julia Unite** Product Manager ![Perlego](https://www.recombee.com/img/logos/perlego.svg) ### About Perlego Perlego is a global digital learning library offering access to over 1.5 million academic, professional, and non-fiction books through a single subscription. Serving students, researchers, and lifelong learners worldwide, the platform helps users discover relevant educational content across a vast and continuously growing catalogue. [Visit Perlego](https://www.perlego.com) **Next Case Studies** [![The Telegraph](https://www.recombee.com/img/case-studies/small-covers/case-studies-the-telegraph.png)+35%CTR uplift on article detail pagesMedia Company](https://www.recombee.com/case-studies/the-telegraph) [![Apify](https://www.recombee.com/img/case-studies/small-covers/case-studies-apify.png)+6%SubscriptionsAI Marketplace](https://www.recombee.com/case-studies/apify) [![Diagnal](https://www.recombee.com/img/case-studies/small-covers/case-studies-diagnal.png)+35%Playbacks Across Video PlatformsVideo Streaming](https://www.recombee.com/case-studies/diagnal) [![9GAG](https://www.recombee.com/img/case-studies/small-covers/case-studies-9gag.png)+37%Post ViewsCross-Platform Entertainment Network](https://www.recombee.com/case-studies/9gag) [![Poslovi Infostud](https://www.recombee.com/img/case-studies/small-covers/case-studies-poslovi-infostud.png)+32%Application conversion on job listingsOnline Job Board](https://www.recombee.com/case-studies/poslovi-infostud) [![Slickdeals](https://www.recombee.com/img/case-studies/small-covers/case-studies-slickdeals.png)+70%CTR to Detail Page ViewsE-commerce + Deal Aggregators](https://www.recombee.com/case-studies/slickdeals) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima.png)+34%Video views on VOD platformMedia Company](https://www.recombee.com/case-studies/ftv-prima) [![Audiomack](https://www.recombee.com/img/case-studies/small-covers/case-studies-audiomack.png)+206%Monthly plays from recommendationsMusic](https://www.recombee.com/case-studies/audiomack) [![Pet Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-pet-media-group.png)+17%CR from visitor to active buyerP2P Marketplaces](https://www.recombee.com/case-studies/pet-media-group) [![Pepper](https://www.recombee.com/img/case-studies/small-covers/case-studies-pepper.png)+21%Click-outs to affiliate links from websiteDeal Aggregators](https://www.recombee.com/case-studies/pepper) [![Crexi](https://www.recombee.com/img/case-studies/small-covers/case-studies-crexi.png)+40%In buy actions from the recommended propertiesReal Estate](https://www.recombee.com/case-studies/crexi) [![Triola](https://www.recombee.com/img/case-studies/small-covers/case-studies-triola.png)10%Total orders from recommendationsE-commerce](https://www.recombee.com/case-studies/triola) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima-content.png)+10%RecirculationMedia Company](https://www.recombee.com/case-studies/ftv-prima-content) [![Unfiltered Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-unfiltered-media-group.png)+50%CTRMedia Company](https://www.recombee.com/case-studies/unfiltered-media-group) [![Reliving](https://www.recombee.com/img/case-studies/small-covers/case-studies-reliving.png)+37%Place BidE-commerce + P2P Marketplaces](https://www.recombee.com/case-studies/reliving) [![Autohaus Kunzmann](https://www.recombee.com/img/case-studies/small-covers/case-studies-autohaus-kunzmann.png)+14%Conversion RateE-commerce](https://www.recombee.com/case-studies/autohaus-kunzmann) [![Showmax](https://www.recombee.com/img/case-studies/small-covers/case-studies-showmax.png)70Countries and multiple languagesSVOD Service](https://www.recombee.com/case-studies/showmax) [![Mafra](https://www.recombee.com/img/case-studies/small-covers/case-studies-mafra.png)40% higherCTR of suggested articlesMedia Company](https://www.recombee.com/case-studies/mafra) [![itison](https://www.recombee.com/img/case-studies/small-covers/case-studies-itison.png)2 000%ROINewsletter Personalization + Deal Aggregators](https://www.recombee.com/case-studies/itison) [![Cooklist](https://www.recombee.com/img/case-studies/small-covers/case-studies-cooklist.png)+27%CTRE-commerce](https://www.recombee.com/case-studies/cooklist) [![Economia](https://www.recombee.com/img/case-studies/small-covers/case-studies-economia.png)+64%Click Through RateMedia Company](https://www.recombee.com/case-studies/economia) [![Segundamano](https://www.recombee.com/img/case-studies/small-covers/case-studies-segundamano.png)3x more conversionsfrom users who engage with recommendationsP2P Marketplaces](https://www.recombee.com/case-studies/segundamano) [![Design Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-design-group.png)+52%Number of PurchasesE-commerce](https://www.recombee.com/case-studies/design-group) --- # Driving 32% Higher Job Application Conversion for Serbia’s Leading Employment Platform > Source: https://www.recombee.com/case-studies/poslovi-infostud > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # [Success Stories](https://www.recombee.com/case-studies) # Driving 32% Higher Job Application Conversion for Serbia’s Leading Employment Platform Poslovi Infostud [Job Boards, HR, Networking](https://www.recombee.com/domains/jobs-boards-hr-networking) [Content Recommendations](https://www.recombee.com/content-recommendations) ![](https://www.recombee.com/img/case-studies/poslovi-infostud.png) **Poslovi Infostud** is the leading online job search and recruitment platform in Serbia, helping job seekers discover relevant roles and enabling employers to reach qualified candidates at scale. With over 4 million monthly visits from both anonymous and registered users, delivering relevant recommendations across every stage of the job-seeking journey is central to the platform’s core mission. In the first week of A/B testing against the existing recommender, **Recombee** exceeded the predefined +20% success threshold across all measured metrics, leading to the decision to fully migrate. +32% Application conversion on job listings +200% Recommendation-to-job engagement among non-logged-in users +40.3% Thank-you page recommendation conversion ### Situation * Existing in-house recommender was functional but difficult to scale without significant engineering investment * Recommendation development relied on rebuilding individual scenarios internally * Recommendations needed to work for both logged-in users with interaction history and anonymous visitors with limited behavioral data * High-intent locations, such as job detail pages and post-application thank-you pages, required relevant recommendations to increase engagement and continue the journey ### Objectives * Support multiple recommendation scenarios, including item-to-item, user-to-item, and item-to-user recommendations * Improve scalability and handle large traffic volumes across multiple website placements and communication channels * Reduce engineering dependency with a self-serve admin interface for configuring recommendation logic, previewing results, monitoring performance, and managing scenarios * Deliver relevant personalization for both logged-in and non-logged-in users ### Solution Recombee supports personalized job discovery across Infostud’s employment platform. Recommendations adapt to different user contexts, supporting browsing, job search, and application journeys with relevant suggestions in real time. **Full-Funnel Personalization** A flexible recommendation setup supports multiple scenarios, including user-to-job, job-to-job, and candidate-to-job recommendations. This enables relevant discovery across key touchpoints, from browsing job listings to high-intent application stages. **Cold-start & Anonymous Optimization** The solution delivers relevant recommendations for both logged-in users and anonymous visitors by leveraging real-time behavioral signals. Even with limited user history, recommendations adapt to available context to surface relevant job opportunities. **Dynamic Profile Enrichment** Personalized experiences for returning users are powered by interaction history and profile data. By analyzing user behavior and preferences, recommendations align with individual interests and improve job discovery over time. **Iterative Conversion Strategy** A scalable recommendation infrastructure enables continuous optimization through testing, performance monitoring, and data-driven improvements. This supports ongoing refinement of recommendation logic to increase engagement and application conversion rates. ### Benefits & Results * **+32%** Application conversion on job listings * **+200%** Recommendation-to-job engagement among non-logged-in users * **+40.3%** Thank-you page recommendation conversion * Improved job discovery across both anonymous and logged-in user journeys * Reduced engineering dependency with self-serve personalization management * Faster iteration and optimization across multiple recommendation scenarios ### Recombee’s Solution in Action ![poslovi-infostud](https://www.recombee.com/img/case-studies/scenarios/poslovi-infostud-1.png) ### Job-Page Continuation Similar Jobs A key discovery touchpoint on individual job pages, appearing at the moment when candidates decide whether to continue exploring or leave the platform. Recommendations are generated using contextual signals such as the viewed role, job attributes, and candidate interactions to surface highly relevant alternative opportunities. By keeping candidates engaged beyond the initial job view, this scenario helps extend the discovery journey and increases the chance of finding the right match. **Key Benefits** * Reduce candidate drop-off on individual job pages * Increase exploration through contextually relevant role alternatives * Extend browsing sessions and improve overall engagement ![poslovi-infostud](https://www.recombee.com/img/case-studies/scenarios/poslovi-infostud-2.png) ### Feed & Thank You Page Jobs for You A personalized discovery surface across the candidate feed and post-application thank-you page, helping users continue exploring relevant opportunities throughout their journey. Recommendations combine behavioral signals, previous interactions, and fresh job activity to deliver the most relevant roles based on each candidate’s evolving interests. By maintaining engagement after key actions such as applying for a role, this scenario encourages continued discovery and increases opportunities for successful matches. **Key Benefits** * Drive continued engagement after application completion * Improve job discovery through personalized recommendations * Increase return visits by keeping candidates connected to relevant opportunities ![poslovi-infostud](https://www.recombee.com/img/case-studies/scenarios/poslovi-infostud-3.png) ### Candidate Profile Matching Candidates to Job A reverse recommendation scenario designed to help recruiters and platforms identify the most relevant candidates for open roles. Recombee analyzes job requirements, candidate profiles, and behavioral signals to match suitable candidates with relevant opportunities beyond traditional search-based workflows. These recommendations can be activated across multiple channels, including notifications, pop-ups, and other platform touchpoints, enabling proactive candidate discovery. **Key Benefits** * Improve candidate-job matching accuracy * Reduce manual search effort for recruiters * Enable proactive talent discovery across multiple channels “Within the first week of A/B testing, Recombee exceeded our success criteria across all key metrics and gave us the confidence to migrate from our internal recommender.” ![Milan Tokic](https://www.recombee.com/img/case-studies/testimonials/poslovi-infostud.png) **Milan Tokic** AI Engineer ![Poslovi Infostud](https://www.recombee.com/img/logos/poslovi-infostud.svg) ### About Poslovi Infostud **Poslovi Infostud** is the leading online job search and recruitment platform in Serbia, helping job seekers discover relevant roles and enabling employers to reach qualified candidates at scale. With over 4 million monthly visits from both anonymous and registered users, delivering relevant recommendations across every stage of the job-seeking journey is central to the platform’s core mission. [Visit Poslovi Infostud](https://poslovi.infostud.com) **Next Case Studies** [![The Telegraph](https://www.recombee.com/img/case-studies/small-covers/case-studies-the-telegraph.png)+35%CTR uplift on article detail pagesMedia Company](https://www.recombee.com/case-studies/the-telegraph) [![Apify](https://www.recombee.com/img/case-studies/small-covers/case-studies-apify.png)+6%SubscriptionsAI Marketplace](https://www.recombee.com/case-studies/apify) [![Diagnal](https://www.recombee.com/img/case-studies/small-covers/case-studies-diagnal.png)+35%Playbacks Across Video PlatformsVideo Streaming](https://www.recombee.com/case-studies/diagnal) [![9GAG](https://www.recombee.com/img/case-studies/small-covers/case-studies-9gag.png)+37%Post ViewsCross-Platform Entertainment Network](https://www.recombee.com/case-studies/9gag) [![Perlego](https://www.recombee.com/img/case-studies/small-covers/case-studies-perlego.png)+44%CTRDigital Learning](https://www.recombee.com/case-studies/perlego) [![Slickdeals](https://www.recombee.com/img/case-studies/small-covers/case-studies-slickdeals.png)+70%CTR to Detail Page ViewsE-commerce + Deal Aggregators](https://www.recombee.com/case-studies/slickdeals) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima.png)+34%Video views on VOD platformMedia Company](https://www.recombee.com/case-studies/ftv-prima) [![Audiomack](https://www.recombee.com/img/case-studies/small-covers/case-studies-audiomack.png)+206%Monthly plays from recommendationsMusic](https://www.recombee.com/case-studies/audiomack) [![Pet Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-pet-media-group.png)+17%CR from visitor to active buyerP2P Marketplaces](https://www.recombee.com/case-studies/pet-media-group) [![Pepper](https://www.recombee.com/img/case-studies/small-covers/case-studies-pepper.png)+21%Click-outs to affiliate links from websiteDeal Aggregators](https://www.recombee.com/case-studies/pepper) [![Crexi](https://www.recombee.com/img/case-studies/small-covers/case-studies-crexi.png)+40%In buy actions from the recommended propertiesReal Estate](https://www.recombee.com/case-studies/crexi) [![Triola](https://www.recombee.com/img/case-studies/small-covers/case-studies-triola.png)10%Total orders from recommendationsE-commerce](https://www.recombee.com/case-studies/triola) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima-content.png)+10%RecirculationMedia Company](https://www.recombee.com/case-studies/ftv-prima-content) [![Unfiltered Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-unfiltered-media-group.png)+50%CTRMedia Company](https://www.recombee.com/case-studies/unfiltered-media-group) [![Reliving](https://www.recombee.com/img/case-studies/small-covers/case-studies-reliving.png)+37%Place BidE-commerce + P2P Marketplaces](https://www.recombee.com/case-studies/reliving) [![Autohaus Kunzmann](https://www.recombee.com/img/case-studies/small-covers/case-studies-autohaus-kunzmann.png)+14%Conversion RateE-commerce](https://www.recombee.com/case-studies/autohaus-kunzmann) [![Showmax](https://www.recombee.com/img/case-studies/small-covers/case-studies-showmax.png)70Countries and multiple languagesSVOD Service](https://www.recombee.com/case-studies/showmax) [![Mafra](https://www.recombee.com/img/case-studies/small-covers/case-studies-mafra.png)40% higherCTR of suggested articlesMedia Company](https://www.recombee.com/case-studies/mafra) [![itison](https://www.recombee.com/img/case-studies/small-covers/case-studies-itison.png)2 000%ROINewsletter Personalization + Deal Aggregators](https://www.recombee.com/case-studies/itison) [![Cooklist](https://www.recombee.com/img/case-studies/small-covers/case-studies-cooklist.png)+27%CTRE-commerce](https://www.recombee.com/case-studies/cooklist) [![Economia](https://www.recombee.com/img/case-studies/small-covers/case-studies-economia.png)+64%Click Through RateMedia Company](https://www.recombee.com/case-studies/economia) [![Segundamano](https://www.recombee.com/img/case-studies/small-covers/case-studies-segundamano.png)3x more conversionsfrom users who engage with recommendationsP2P Marketplaces](https://www.recombee.com/case-studies/segundamano) [![Design Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-design-group.png)+52%Number of PurchasesE-commerce](https://www.recombee.com/case-studies/design-group) --- # Real-Time AI-Based Personalization Increased ROI > Source: https://www.recombee.com/case-studies/reliving > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [Success Stories](https://www.recombee.com/case-studies) # Real-Time Product Recommendations for the Vintage Gems Marketplace Reliving [E-commerce](https://www.recombee.com/domains/e-commerce), [P2P Marketplaces](https://www.recombee.com/domains/p2p-marketplaces) [Product Recommendations](https://www.recombee.com/product-recommendations) ![](https://www.recombee.com/img/case-studies/reliving.png) Reliving encourages people to live by sustainability principles, offering a wide range of curated second-hand interior design pieces. The marketplace strives to become the #1 player in Europe for high-quality, affordable vintage furniture. Recombee's AI product recommendations allowed Reliving to personalize the unique shopping experience for their customers while increasing place bids, cart additions and thus the ROI. 37% In Place Bid 6% In Add to Cart ### Situation * The steady growth of both the user base and products available on the platform. * Frequently changing catalog with one-time purchases. * A big part of traffic driven directly to the product detail pages from search engines. ### Requirements * Personalize the customer experience by helping each user find what they are looking for. * Easy setup and implementation of the solution. * Possibility to influence the recommendations to prefer specific products over the others. ### Solution * **“Maybe You Also Like”** section on the product detail page. * **Default model ensemble** continuously optimized by AI to automatically adapts to incoming data. * **Custom-made business rules** to boost specific labels. ### Benefits & Results * **37% increase** in the “Place Bid” supporting the healthy competition between visitors to purchase a specific product. * **6% increase** in “Add to cart” which is the equivalent of a conversion. * Plans to deploy Recombee on more places over the website in the future. ### Scenarios ![reliving](https://www.recombee.com/img/case-studies/scenarios/reliving-1.png) #### Maybe You Also Like When you are looking at specific products on the Reliving website, you are also getting a list of similar items, helping you find the right design piece for your interior. This is achieved thanks to the boosting rule and ensemble of content-based and collaborative filtering models included in the **recombee:default logic.** “Instead of the lengthy and costly process of building an in-house personalization machine, we seamlessly implemented Recombee's AI-powered recommender engine to improve our services. Applying their e-commerce tailored scenarios, we have registered steady improvements ever since, with **a 6% increase in 'Add to Cart' and a 37% increase in 'Place Bid'.** Thanks to their easy and intuitive integration, Recombee was an obvious choice from the range of personalization solutions.” **Vincent van Leeuwen** Co-Founder & CTO/CPO at Reliving ![Reliving](https://www.recombee.com/img/logos/reliving.png) ### About Reliving Reliving was founded in 2019 with the vision to become the first place where you look when in search of an interior design piece. The platform advocates an environmentally friendly alternative to traditional furniture sales, providing beautiful, quality items that are already out there. The ultimate goal? Pursuit of a world where fast furniture disappears and furniture has a longer lifecycle. [Visit Reliving](https://www.reliving.nl) **Next Case Studies** [![The Telegraph](https://www.recombee.com/img/case-studies/small-covers/case-studies-the-telegraph.png)+35%CTR uplift on article detail pagesMedia Company](https://www.recombee.com/case-studies/the-telegraph) [![Apify](https://www.recombee.com/img/case-studies/small-covers/case-studies-apify.png)+6%SubscriptionsAI Marketplace](https://www.recombee.com/case-studies/apify) [![Diagnal](https://www.recombee.com/img/case-studies/small-covers/case-studies-diagnal.png)+35%Playbacks Across Video PlatformsVideo Streaming](https://www.recombee.com/case-studies/diagnal) [![9GAG](https://www.recombee.com/img/case-studies/small-covers/case-studies-9gag.png)+37%Post ViewsCross-Platform Entertainment Network](https://www.recombee.com/case-studies/9gag) [![Poslovi Infostud](https://www.recombee.com/img/case-studies/small-covers/case-studies-poslovi-infostud.png)+32%Application conversion on job listingsOnline Job Board](https://www.recombee.com/case-studies/poslovi-infostud) [![Perlego](https://www.recombee.com/img/case-studies/small-covers/case-studies-perlego.png)+44%CTRDigital Learning](https://www.recombee.com/case-studies/perlego) [![Slickdeals](https://www.recombee.com/img/case-studies/small-covers/case-studies-slickdeals.png)+70%CTR to Detail Page ViewsE-commerce + Deal Aggregators](https://www.recombee.com/case-studies/slickdeals) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima.png)+34%Video views on VOD platformMedia Company](https://www.recombee.com/case-studies/ftv-prima) [![Audiomack](https://www.recombee.com/img/case-studies/small-covers/case-studies-audiomack.png)+206%Monthly plays from recommendationsMusic](https://www.recombee.com/case-studies/audiomack) [![Pet Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-pet-media-group.png)+17%CR from visitor to active buyerP2P Marketplaces](https://www.recombee.com/case-studies/pet-media-group) [![Pepper](https://www.recombee.com/img/case-studies/small-covers/case-studies-pepper.png)+21%Click-outs to affiliate links from websiteDeal Aggregators](https://www.recombee.com/case-studies/pepper) [![Crexi](https://www.recombee.com/img/case-studies/small-covers/case-studies-crexi.png)+40%In buy actions from the recommended propertiesReal Estate](https://www.recombee.com/case-studies/crexi) [![Triola](https://www.recombee.com/img/case-studies/small-covers/case-studies-triola.png)10%Total orders from recommendationsE-commerce](https://www.recombee.com/case-studies/triola) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima-content.png)+10%RecirculationMedia Company](https://www.recombee.com/case-studies/ftv-prima-content) [![Unfiltered Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-unfiltered-media-group.png)+50%CTRMedia Company](https://www.recombee.com/case-studies/unfiltered-media-group) [![Autohaus Kunzmann](https://www.recombee.com/img/case-studies/small-covers/case-studies-autohaus-kunzmann.png)+14%Conversion RateE-commerce](https://www.recombee.com/case-studies/autohaus-kunzmann) [![Showmax](https://www.recombee.com/img/case-studies/small-covers/case-studies-showmax.png)70Countries and multiple languagesSVOD Service](https://www.recombee.com/case-studies/showmax) [![Mafra](https://www.recombee.com/img/case-studies/small-covers/case-studies-mafra.png)40% higherCTR of suggested articlesMedia Company](https://www.recombee.com/case-studies/mafra) [![itison](https://www.recombee.com/img/case-studies/small-covers/case-studies-itison.png)2 000%ROINewsletter Personalization + Deal Aggregators](https://www.recombee.com/case-studies/itison) [![Cooklist](https://www.recombee.com/img/case-studies/small-covers/case-studies-cooklist.png)+27%CTRE-commerce](https://www.recombee.com/case-studies/cooklist) [![Economia](https://www.recombee.com/img/case-studies/small-covers/case-studies-economia.png)+64%Click Through RateMedia Company](https://www.recombee.com/case-studies/economia) [![Segundamano](https://www.recombee.com/img/case-studies/small-covers/case-studies-segundamano.png)3x more conversionsfrom users who engage with recommendationsP2P Marketplaces](https://www.recombee.com/case-studies/segundamano) [![Design Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-design-group.png)+52%Number of PurchasesE-commerce](https://www.recombee.com/case-studies/design-group) --- # Generating 160 Million Product Recommendations per Month > Source: https://www.recombee.com/case-studies/segundamano > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [Success Stories](https://www.recombee.com/case-studies) # Bringing Personalization to Advertising for 160 Million Monthly Views Segundamano [P2P Marketplaces](https://www.recombee.com/domains/p2p-marketplaces) [Product Recommendations](https://www.recombee.com/product-recommendations) ![](https://www.recombee.com/img/case-studies/segundamano.png) Segundamano is one of the largest online marketplaces in Mexico offering exceptional product diversity. Due to the high number of products available, and the constantly changing content, it was not practical to manually browse and manage each and every recommendation given to consumers. We solved this personalization challenge by providing Segundamano with a high-tech recommender engine built on collaborative filtering-based models. 3x more conversions from users who engage with recommendations ### Situation * Difficult to sort through high amount of product offerings. * 4 million products/month. * Highly heterogeneous products. ### Requirements * Recognize preferences of individual users. * Balance personalization and discovery of products. * Find similar products. * Real-time response in large traffic. ### Solution * Deep learning image models to determine similar items. * Advanced processing of Spanish text titles and descriptions. * A combination of collaborative-filtering and robust text mining models. * Real-time model updates and recommendations under 100 ms. * High performance without compromising quality. ### Benefits & Results * Automatic personalized recommendations applied to Product Detail Pages (_‘También te podría interesar’/‘You may also like’_ recommendation box) * 10% item views attributed to Recombee recommendations * **3x more Conversions** from users who engaged with recommendations * **4x more Interactions** from users who engaged with recommendations * Real-time product personalization ### Scenarios ![segundamano](https://www.recombee.com/img/case-studies/scenarios/segundamano-1.png) #### También te podría interesar (you may also like) The product detail view recommendations take into consideration similar user interactions, purchase history, and different product attributes enhanced by image processing capabilities of Recombee’s solution. This scenario’s main goal is to show users relevant content while showing users similar products to widen their shopping options. “Our developers love Recombee documentation as well as quick and valuable technical support. We see Recombee’s “recommendationsToUser“ algorithm as a great option to start offering a personalized experience on our site.” **Marco Alvarez** Product Manager at Segundamano ![segundamano](https://www.recombee.com/img/logos/segundamano.png) ### About Segundamano Segundamano is the largest online classifieds ads platform in Mexico. With its business model similar to Craigslist in the US, Segundamano seeks to be the best meeting point between sellers and buyers allowing them to create mutually beneficial relationships. Segundamano’s platform offers an array of products such as furniture, electronics, and clothing, as well as real estate ads and job offers. The online marketplace belongs to Adevinta, a Norwegian multinational company that is one of the most ambitious technology and product companies in Europe, serving more than 250 million users worldwide. [Visit Segundamano](https://www.segundamano.mx/) **Next Case Studies** [![The Telegraph](https://www.recombee.com/img/case-studies/small-covers/case-studies-the-telegraph.png)+35%CTR uplift on article detail pagesMedia Company](https://www.recombee.com/case-studies/the-telegraph) [![Apify](https://www.recombee.com/img/case-studies/small-covers/case-studies-apify.png)+6%SubscriptionsAI Marketplace](https://www.recombee.com/case-studies/apify) [![Diagnal](https://www.recombee.com/img/case-studies/small-covers/case-studies-diagnal.png)+35%Playbacks Across Video PlatformsVideo Streaming](https://www.recombee.com/case-studies/diagnal) [![9GAG](https://www.recombee.com/img/case-studies/small-covers/case-studies-9gag.png)+37%Post ViewsCross-Platform Entertainment Network](https://www.recombee.com/case-studies/9gag) [![Poslovi Infostud](https://www.recombee.com/img/case-studies/small-covers/case-studies-poslovi-infostud.png)+32%Application conversion on job listingsOnline Job Board](https://www.recombee.com/case-studies/poslovi-infostud) [![Perlego](https://www.recombee.com/img/case-studies/small-covers/case-studies-perlego.png)+44%CTRDigital Learning](https://www.recombee.com/case-studies/perlego) [![Slickdeals](https://www.recombee.com/img/case-studies/small-covers/case-studies-slickdeals.png)+70%CTR to Detail Page ViewsE-commerce + Deal Aggregators](https://www.recombee.com/case-studies/slickdeals) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima.png)+34%Video views on VOD platformMedia Company](https://www.recombee.com/case-studies/ftv-prima) [![Audiomack](https://www.recombee.com/img/case-studies/small-covers/case-studies-audiomack.png)+206%Monthly plays from recommendationsMusic](https://www.recombee.com/case-studies/audiomack) [![Pet Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-pet-media-group.png)+17%CR from visitor to active buyerP2P Marketplaces](https://www.recombee.com/case-studies/pet-media-group) [![Pepper](https://www.recombee.com/img/case-studies/small-covers/case-studies-pepper.png)+21%Click-outs to affiliate links from websiteDeal Aggregators](https://www.recombee.com/case-studies/pepper) [![Crexi](https://www.recombee.com/img/case-studies/small-covers/case-studies-crexi.png)+40%In buy actions from the recommended propertiesReal Estate](https://www.recombee.com/case-studies/crexi) [![Triola](https://www.recombee.com/img/case-studies/small-covers/case-studies-triola.png)10%Total orders from recommendationsE-commerce](https://www.recombee.com/case-studies/triola) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima-content.png)+10%RecirculationMedia Company](https://www.recombee.com/case-studies/ftv-prima-content) [![Unfiltered Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-unfiltered-media-group.png)+50%CTRMedia Company](https://www.recombee.com/case-studies/unfiltered-media-group) [![Reliving](https://www.recombee.com/img/case-studies/small-covers/case-studies-reliving.png)+37%Place BidE-commerce + P2P Marketplaces](https://www.recombee.com/case-studies/reliving) [![Autohaus Kunzmann](https://www.recombee.com/img/case-studies/small-covers/case-studies-autohaus-kunzmann.png)+14%Conversion RateE-commerce](https://www.recombee.com/case-studies/autohaus-kunzmann) [![Showmax](https://www.recombee.com/img/case-studies/small-covers/case-studies-showmax.png)70Countries and multiple languagesSVOD Service](https://www.recombee.com/case-studies/showmax) [![Mafra](https://www.recombee.com/img/case-studies/small-covers/case-studies-mafra.png)40% higherCTR of suggested articlesMedia Company](https://www.recombee.com/case-studies/mafra) [![itison](https://www.recombee.com/img/case-studies/small-covers/case-studies-itison.png)2 000%ROINewsletter Personalization + Deal Aggregators](https://www.recombee.com/case-studies/itison) [![Cooklist](https://www.recombee.com/img/case-studies/small-covers/case-studies-cooklist.png)+27%CTRE-commerce](https://www.recombee.com/case-studies/cooklist) [![Economia](https://www.recombee.com/img/case-studies/small-covers/case-studies-economia.png)+64%Click Through RateMedia Company](https://www.recombee.com/case-studies/economia) [![Design Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-design-group.png)+52%Number of PurchasesE-commerce](https://www.recombee.com/case-studies/design-group) --- # Increase Average Order Value | Targito > Source: https://www.recombee.com/case-studies/triola > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [Success Stories](https://www.recombee.com/case-studies) # Boosting Lingerie & Swimwear Sales by 10% with AI Personalization Triola [E-commerce](https://www.recombee.com/domains/e-commerce) [Product Recommendations](https://www.recombee.com/product-recommendations) ![](https://www.recombee.com/img/case-studies/triola.png) Triola is a Czech producer of several lines of bras and swimwear designed for women of all shapes and sizes, with attention to functionality and comfort. With their large catalog of products in multiple styles and fits, Triola discovered their visitors were struggling to find the desired items which lead to high bounce rates. In collaboration with partner Targito, Recombee was applied to help shoppers find the right product in the right style and size as quickly as possible. By improving the user experience, Triola increased 10% in shopping cart volume and achieved a stunning cost-revenue ratio of 3.7%. 10% Total Orders from Recommendations +10% In Average Order Value 3.7% Cost-Revenue Ratio ### Situation * Triola offers a vast catalog of lingerie and swimwear featuring various sizes, styles, and fits. * Shoppers often struggled to find the right product. * Triola was experiencing high bounce rates with new and returning customers. * Typical customers had only one or two favorite styles from an entire pool of options. * Triola’s goal was to provide the best possible user experience to each shopper. ### Requirements * Personalization strategy based on user behavior insights and product attributes. * Leverage user data from both physical stores and online behavior to provide tailored recommendations. * Showcases relevant products to each shopper, including the shopper's favorite style and fit. * Real-time response to new users to ensure all users receive personalized journeys from the start. ### Solution Personalized homepage, detail page, shopping cart, search, and emailing. Solution considering data collected in **physical stores.** High emphasis on recommending the right fit for each shopper. **Search results** optimized based on the user’s behavior and product attributes. Targito + Recombee Solution for Triola Targito’s clients enjoy a simplified method of integration. The user’s interactions and attributes are seamlessly sent from Targito to Recombee providing a quick and easy integration. ### Benefits & Results * **10%** of total orders **from recommendations.** * **+10%** in average order value * The Cost-revenue ratio of **3.7%** * Rise in shoppers’ satisfaction; decreased bounce rate, longer shopping sessions, and larger volume of frequently returning customers. * Guarantee future purchased products will be a good fit to help Triola create a foundation of loyal and returning customers. ### Targito + Recombee Solution for Triola Targito’s clients enjoy a simplified method of integration. The user’s interactions and attributes are seamlessly sent from Targito to Recombee providing a quick and easy integration. ![Recombee + Targito](https://www.recombee.com/img/case-studies/schema/recombee-targito.svg) “At Targito, our priority is to maximize campaign performance, especially conversion, for our clients with as little effort as possible. Thanks to the already resolved integration of the Recombee solution, we were able to offer visitors a wider and, at the same time, better-targeted range of products on our Triola website. This approach did not require any major modifications on Triola's part, which would have implied a large investment, both time and money. We connected everything through our Targito data platform, which Triola has been using for a long time. Previously, Targito was only used for sophisticated sending of email campaigns and automation with overlap into social networks. Thanks to Recombee and Targito's partnership, the connection is very simple, and we can use data from multiple sources. The result is the addition of another highly profitable channel to the ecosystem of personalized campaigns.” ![Pavel Šolc](https://www.recombee.com/img/case-studies/testimonials/triola-2.png) **Pavel Solc** Business Development Manager at Targito ### AI-Powered Product Recommendations & Improve Customer Conversion Funnel ![Funnel](https://www.recombee.com/img/case-studies/schema/funnel.svg) ### Scenarios ![triola](https://www.recombee.com/img/case-studies/scenarios/triola-1.png) #### Personalized Search The “personalized search” scenario employs a combination of a search engine and a recommender system. Leveraging Recombee's personalized logic, Triola places the right product selection in front of the right user based on the user’s interaction and metadata. ![triola](https://www.recombee.com/img/case-studies/scenarios/triola-2.png) #### Category Sorting Automatic set-up of items on the bras and swimming suit main pages to provide a seamless shopping experience. Similar products to previously purchased items are displayed on the top of the page, making it easier for shoppers to find items that match their style and fit preferences. ![triola](https://www.recombee.com/img/case-studies/scenarios/triola-3.png) #### Relevant Products The scenario is found on the product detail page and uses a unique ensemble of image and text processing models to recommend products that are similar or may be bought together with the currently viewed one. ![triola](https://www.recombee.com/img/case-studies/scenarios/triola-4.png) #### You May Also Like The “You May Also Like” scenario is commonly used for Cross-selling and utilizes an automatically AI optimized ensemble of both content-based and collaborative filtering models. The complementary products prompt shoppers to make more purchases, e.g. by recommending matching underwear to the given bra. “At Triola, we faced the challenge of optimizing the user experience on our website, given our extensive catalog of products of numerous shapes and sizes. Knowing our customers typically don't use filters to find what they are looking for, our goal was to guide them by showcasing on the homepage the items they previously bought and suggesting similar products. That's why we decided to work with Recombee, and the results have been outstanding. The company has witnessed a 10% increase in total orders from recommendations, a 10% increase in average order value, and an overall cost-revenue ratio of 3.7%.” ![Lenka Sobotkova](https://www.recombee.com/img/case-studies/testimonials/triola-1.png) **Lenka Sobotkova** E-commerce Manager at Triola ![Targito](https://www.recombee.com/img/logos/targito.svg) ### About Targito Targito is one of the largest email marketing technologies and services providers in the CEE region, setting the direction of email marketing for the last decade. The company is working to bring simplicity, connectivity, and efficiency to the hectic data-driven world. In addition to advanced email campaign automation and personalization in creating, distributing, and evaluating e-newsletters, the platform offers integration of all online communication with customers. Targito clients are composing of E-commerce platforms and together with Recombee provide personalized experience since 2021. Targito’s clients include: Zoot, Karcher, Bonami and 100+ E-commerce companies. [Visit Targito](https://www.targito.com/en/) ![Triola](https://www.recombee.com/img/logos/triola.svg) ### About Triola Triola is a Czech producer of lingerie and swimwear with more than 100 years of tradition. With its emphasis on the quality of materials and customer needs, the brand translates years of experience into modern production processes and fashionable design. Triola is one of the few companies in the Czech and Slovak markets to offer several lines of bras and swimwear for women of various shapes, with a wide range of sizes and attention to functionality and comfort. [Visit Triola](https://www.triola.cz/) **Next Case Studies** [![The Telegraph](https://www.recombee.com/img/case-studies/small-covers/case-studies-the-telegraph.png)+35%CTR uplift on article detail pagesMedia Company](https://www.recombee.com/case-studies/the-telegraph) [![Apify](https://www.recombee.com/img/case-studies/small-covers/case-studies-apify.png)+6%SubscriptionsAI Marketplace](https://www.recombee.com/case-studies/apify) [![Diagnal](https://www.recombee.com/img/case-studies/small-covers/case-studies-diagnal.png)+35%Playbacks Across Video PlatformsVideo Streaming](https://www.recombee.com/case-studies/diagnal) [![9GAG](https://www.recombee.com/img/case-studies/small-covers/case-studies-9gag.png)+37%Post ViewsCross-Platform Entertainment Network](https://www.recombee.com/case-studies/9gag) [![Poslovi Infostud](https://www.recombee.com/img/case-studies/small-covers/case-studies-poslovi-infostud.png)+32%Application conversion on job listingsOnline Job Board](https://www.recombee.com/case-studies/poslovi-infostud) [![Perlego](https://www.recombee.com/img/case-studies/small-covers/case-studies-perlego.png)+44%CTRDigital Learning](https://www.recombee.com/case-studies/perlego) [![Slickdeals](https://www.recombee.com/img/case-studies/small-covers/case-studies-slickdeals.png)+70%CTR to Detail Page ViewsE-commerce + Deal Aggregators](https://www.recombee.com/case-studies/slickdeals) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima.png)+34%Video views on VOD platformMedia Company](https://www.recombee.com/case-studies/ftv-prima) [![Audiomack](https://www.recombee.com/img/case-studies/small-covers/case-studies-audiomack.png)+206%Monthly plays from recommendationsMusic](https://www.recombee.com/case-studies/audiomack) [![Pet Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-pet-media-group.png)+17%CR from visitor to active buyerP2P Marketplaces](https://www.recombee.com/case-studies/pet-media-group) [![Pepper](https://www.recombee.com/img/case-studies/small-covers/case-studies-pepper.png)+21%Click-outs to affiliate links from websiteDeal Aggregators](https://www.recombee.com/case-studies/pepper) [![Crexi](https://www.recombee.com/img/case-studies/small-covers/case-studies-crexi.png)+40%In buy actions from the recommended propertiesReal Estate](https://www.recombee.com/case-studies/crexi) [![FTV Prima](https://www.recombee.com/img/case-studies/small-covers/case-studies-ftv-prima-content.png)+10%RecirculationMedia Company](https://www.recombee.com/case-studies/ftv-prima-content) [![Unfiltered Media Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-unfiltered-media-group.png)+50%CTRMedia Company](https://www.recombee.com/case-studies/unfiltered-media-group) [![Reliving](https://www.recombee.com/img/case-studies/small-covers/case-studies-reliving.png)+37%Place BidE-commerce + P2P Marketplaces](https://www.recombee.com/case-studies/reliving) [![Autohaus Kunzmann](https://www.recombee.com/img/case-studies/small-covers/case-studies-autohaus-kunzmann.png)+14%Conversion RateE-commerce](https://www.recombee.com/case-studies/autohaus-kunzmann) [![Showmax](https://www.recombee.com/img/case-studies/small-covers/case-studies-showmax.png)70Countries and multiple languagesSVOD Service](https://www.recombee.com/case-studies/showmax) [![Mafra](https://www.recombee.com/img/case-studies/small-covers/case-studies-mafra.png)40% higherCTR of suggested articlesMedia Company](https://www.recombee.com/case-studies/mafra) [![itison](https://www.recombee.com/img/case-studies/small-covers/case-studies-itison.png)2 000%ROINewsletter Personalization + Deal Aggregators](https://www.recombee.com/case-studies/itison) [![Cooklist](https://www.recombee.com/img/case-studies/small-covers/case-studies-cooklist.png)+27%CTRE-commerce](https://www.recombee.com/case-studies/cooklist) [![Economia](https://www.recombee.com/img/case-studies/small-covers/case-studies-economia.png)+64%Click Through RateMedia Company](https://www.recombee.com/case-studies/economia) [![Segundamano](https://www.recombee.com/img/case-studies/small-covers/case-studies-segundamano.png)3x more conversionsfrom users who engage with recommendationsP2P Marketplaces](https://www.recombee.com/case-studies/segundamano) [![Design Group](https://www.recombee.com/img/case-studies/small-covers/case-studies-design-group.png)+52%Number of PurchasesE-commerce](https://www.recombee.com/case-studies/design-group) --- # News and Media Personalization | Recombee AI > Source: https://www.recombee.com/blog/breaking-the-news-the-role-of-ai-in-modern-journalism > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # Breaking the News: The Role of AI in Modern Journalism ![](https://www.recombee.com/img/blog/authors/tanalancova.png) Tana Lancova Mar 14, 2023 Artificial Intelligence (AI) has rapidly transformed the media industry in recent years. From automated news production to trend analysis and personalized content recommendations, AI has brought significant changes to the way media is created, distributed, and consumed. ![](https://www.recombee.com/img/blog/breaking-the-news-the-role-of-ai-in-modern-journalism/main.png) AI algorithms can help news organizations quickly process large amounts of data to identify patterns, analyze and capture trends, verify facts, detect fake news, etc. AI can also generate headlines and illustrations, enrich the article with external sources, convert articles into spoken form and vice versa. Beyond these assistant tasks, AI can be used to generate content. Simple templated sports or stock market texts, daily summaries of important news in the form of long-form summaries, are generated by modern language models rapidly replacing rule-based AI systems. Moreover, AI can also be used to write texts in any genre on any topic, ready for publication with just a little editorial tweaking. There are several AI writing tools that you can try such as [ChatGPT](https://chat.openai.com/), [Smodin Author](https://smodin.io/writer), [Writesonic](https://writesonic.com/), [Jasper](https://www.jasper.ai/), [Article Forge](https://www.articleforge.com/), [Ink](https://inkforall.com/), [AI Writer](https://ai-writer.com/), [Word AI](https://wordai.com/) or [You.com](https://you.com/search?q=how+to+write+well&&tbm=youwrite&cfr=write&). However, AI in media also comes with its own set of challenges and concerns. One major issue is the potential for generative AI to fabricate and amplify misleading and harmful content with the consequence of [radicalizing society](https://www.nbcnews.com/tech/misinformation/facebook-lawsuit-africa-content-moderation-violence-rcna61530) or [manipulating public opinion](https://www.motherjones.com/politics/2021/05/facebook-bot-farm/). In order to reduce such risks, researchers and engineers have developed tools that can recognize AI generated content which may be [inaccurate](https://www.reuters.com/technology/google-cautions-against-hallucinating-chatbots-report-2023-02-11/) or even [completely](https://www.forbes.com/sites/lanceeliot/2022/08/24/ai-ethics-lucidly-questioning-this-whole-hallucinating-ai-popularized-trend-that-has-got-to-stop/?sh=58a90c4b77df) fabricated. While some tools are already available such as [OpenAI Classifier](https://platform.openai.com/ai-text-classifier) or [Groover](https://grover.allenai.org/detect) from the Allen institute, they are not perfect as the task of identifying AI generated content is much more difficult than, for example, detecting counterfeit items. As generative AI flourishes, detection methods will always be lagging behind and we will continue to rely on experienced [fact checkers](https://www.poynter.org/fact-checking/2022/how-will-automated-fact-checking-work/) who are rapidly adopting [AI tools](https://thetrustedweb.org/ai-powered-tools-for-fighting-fake-news/). Another key area where AI is transforming media is content distribution. As the prevalence of online editions of content has risen, people have gained access to a virtually unlimited supply of content. But sorting through all that information can be overwhelming. AI algorithms analyze a person's viewing habits, preferences, and behavior to provide highly personalized content recommendations. This increases user engagement and helps users discover new and relevant content they may have missed otherwise. In Recombee, we work with a number of public and private media organizations. We provide them with a recommender system as a service which distributes unique and tailored content to their users. We need to be sure that our recommendations are not biased or harmful. In our [previous blogpost](https://www.recombee.com/blog/repsys-opensource-library-for-interactive-evaluation-of-recommendation-systems), we demonstrated how algorithmic biases can be detected and reduced. Another potentially harmful feature of algorithmic recommendation is the [emergence of filter bubbles](https://link.springer.com/article/10.1007/s11280-022-01031-4) and the negative [effect](https://arxiv.org/pdf/1906.08772.pdf) on the audience that might be a [bit overstated](https://journals.sagepub.com/doi/pdf/10.1177/08944393221149290) in mainstream news. Fortunately, our experiments confirm that users prefer not to be locked in the filter bubbles and our algorithms help them to explore. Media companies should be aware of this risk and should avoid using simple collaborative filtering algorithms or content based methods optimized for simple criteria such as the offline recall. ![](https://www.recombee.com/img/blog/breaking-the-news-the-role-of-ai-in-modern-journalism/01.png) Filter bubbles: when a recommender system is not optimized for supporting users in content exploration, users are locked into filter bubbles (image from [this paper](https://link.springer.com/article/10.1007/s11280-022-01031-4)). From our perspective, the implementation of AI algorithms in the news industry has had a positive impact. Through the use of personalised recommendation technology, our customers have observed that content producers are able to reach their target audience more efficiently and effectively. This technology allows for niche content, which might not have been prominently featured on the front page, to receive ample attention by connecting with the relevant audience through website recommendations or personalised newsletters. ## From Theory to Practice: Insights from Professionals on the Frontlines We wanted to know how the media organizations across Europe see the role of AI in their environment. That is why we asked media professionals to share their ideas, experiences, and concerns about using AI in the media industry. We are happy to share some interesting thoughts from the very inspiring conversations we had in preparation for the Discussion panel in the [AI Mellontology (futurology) Symposium](http://icarus.csd.auth.gr/ai-mellontology-e-symposium-2022/). We paid particular attention to public media, because, by its very nature, it has the most demanding requirements for AI. Since it can still be a sensitive topic, some have asked us to remain anonymous. We respect their wishes and have anonymized their comments using their initials. As one our correspondent (let's call them J from a big public radio company) points out: We need to clarify and explore _“how companies and governments deal with AI and what our ethical guidelines are about it.”_ K (from a multinational telecommunications company) sums up: _“We are not like Netflix.”_ ### AI’s Efficiency Artificial intelligence is considered a very promising tool, or, more precisely, approach. If we focus only on content, then almost all public media successfully use it for data preparation, automatic content generation, and content distribution. The automatic message location, metadata extraction, and fact-checking save a lot of time. Or it can be used for "automatic song recognition", as mentioned by John Patrick Rott from [Hessischer Rundfunk](https://www.hr.de/index.html). _“It could help the newsroom get more time and really focus on journalist work. Editors should work on reports where they bring in their view and where we can use their expertise as journalists; since writing normal reports is a little bit of a waste of time”_ J highlights the benefits of using AI to generate content automatically. On the other hand, this also brings complexities because, as J continues, _"We will have to teach people to distinguish between artificially generated text by a synthetic medium and real quality content prepared by a journalist."_ ### Transparency and Responsibility in Recommendation Systems for Public Media Concerning the distribution of content, public service media, in particular, has very strict requirements for recommendation systems in terms of transparency, credibility, and the possibility of editorial intervention. Cristina Kadar from [Neue Zürcher Zeitung](https://www.nzz.ch/), who joined our panel, impressed us with her presentation. As demonstrated in her presentation slide, the team achieved significantly higher user engagement rates while also prioritizing ethical considerations such as mitigating algorithmic bias and upholding their responsibilities as upholding their high journalistic standards as the Swiss-German newspaper of record. ![](https://www.recombee.com/img/blog/breaking-the-news-the-role-of-ai-in-modern-journalism/02.png) Cristina Kadar presented her insights at AI Mentology Symposium in Greece ### Considerations for User Information and Metrics in Recommender Systems Very careful consideration must also be given to what user information the Recommender is allowed to use and how to properly evaluate user actions. _“Before implementing a recommender system, it's important to define the actions that will be used to measure success. Will it be based solely on clicks, or should other factors, such as time spent on a page, be taken into consideration?_ _This can be a challenging task, as it requires careful consideration of what metrics will accurately reflect user engagement and satisfaction with the recommended content”,_ according to Simone Spaccarotella from [BBC.](https://www.bbc.com/) _“The purpose of making data available is not to simply release it without any direction, but rather to use it strategically to motivate people towards a specific goal or direction. It's important to choose Key Performance Indicators (KPIs) that align with the transformation you wish to achieve. By selecting the right KPIs, you can motivate individuals to take actions that will drive the desired transformation and ultimately achieve your goals,”_ emphasizes F from a big public news organization. ### Challenges Beyond AI: Adapting to New Platforms and Changes in Information Consumption AI is not the only challenge that traditional public service media must face. The development of new platforms and a shift in the way we consume information are also bringing new pressures for change. _“In the past, it was possible to set the agenda and direct users towards content within your ecosystem by simply moving them around. However, with the rise of streaming platforms, the majority of users (approximately 80%) are now making decisions about what to watch before even arriving on the platform. As a result, the role of platforms has shifted from being a leader in content direction to being more of a facilitator in helping users find the content they are already seeking”,_ concludes J. When you are not only distributing your content, but also allow third party content to your newsfeed, there are even more challenges involved. Karel Koupil from [Seznam](https://www.seznam.cz/) pointed out in his talk at AI Mentology Symposium, _"After giving third-party content producers access to our homepage, we immediately noticed an influx of click-bait content. In order to maintain the quality of our platform, we had to take swift action to address the issue."_ However, eliminating click-bait content completely is also not the best solution. _"Many people are primarily seeking entertainment when consuming content, and are often aware that some click-bait content may not offer substantial value. However, the presence of a small portion of such content can actually improve their loyalty and even increase engagement with more serious reporting. This is because a degree of variety can help break up the monotony and provide a more engaging and enjoyable experience for readers. Ultimately, striking the right balance between entertainment and informative content is key to keeping audiences engaged and loyal over the long term."_ concludes Karel. These interviews and presentations have been very enlightening for us and we have gained a better understanding of the challenges that not only public relations professionals encounter. Many thanks for that. We do not aim to cover the entire AI ecosystem in media organizations. In our next blog post, we will explain how we can assist editors and media professionals in gaining a better understanding of AI recommendation systems and their readers, enabling them to gain more insight into the produced recommendations. And if you share our interest in the intersection of artificial intelligence and media, we highly recommend delving into [this](https://www.mediaroad.eu/archives/30328) insightful and comprehensive study on the subject. We are also speaking at the [Data Technology Seminar](https://tech.ebu.ch/events/dts2023) organized by the [European Broadcasting Union.](https://www.ebu.ch/home) If you are interested in meeting us in person in Geneva please contact us at [pavel.kordik@recombee.com](mailto:pavel.kordik@recombee.com) or at [tana.lancova@recombee.com](mailto:tana.lancova@recombee.com) Recommendation Engine Personalization ## Next Articles [![](https://www.recombee.com/img/blog/inductive-matrix-completion-how-to-improve-recommendations-for-cold-start-users-and-items-by-incorporating-their-attributes.png)](https://www.recombee.com/blog/inductive-matrix-completion-how-to-improve-recommendations-for-cold-start-users-and-items-by-incorporating-their-attributes) ### [Inductive Matrix Completion: How to Improve Recommendations for Cold Start Users and Items by Incorporating Their Attributes](https://www.recombee.com/blog/inductive-matrix-completion-how-to-improve-recommendations-for-cold-start-users-and-items-by-incorporating-their-attributes) Matrix completion (MC), the problem of recovering the missing entries of a partially observed matrix, has found use in a wide range of domains. Still, its potentially most successful application is as a collaborative filtering technique for recommender systems (RSs)... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Mar 20, 2023 Personalization Recommendation Engine [![](https://www.recombee.com/img/blog/innovative-personalization-features-for-2023.png)](https://www.recombee.com/blog/innovative-personalization-features-for-2023) ### [Innovative Personalization Features for 2023](https://www.recombee.com/blog/innovative-personalization-features-for-2023) The digital world is changing; users' expectations for personalization are increasing, and our Recombee features are continuously improving. One of our focuses is to support our clients in providing the best possible user experiences... ![](https://www.recombee.com/img/blog/authors/gabrielatakacova.png) Gabriela Takacova Feb 17, 2023 New Features Recommendation Engine [![](https://www.recombee.com/img/blog/recombee-item-segmentations.png)](https://www.recombee.com/blog/recombee-item-segmentations) ### [Recombee Item Segmentations](https://www.recombee.com/blog/recombee-item-segmentations) Item Segmentations are Recombee's original and elegant solution to various advanced tasks related to hierarchical and relational data. The feature provides a flexible way to group items (products or pieces of content) into segments... ![](https://www.recombee.com/img/blog/authors/tomasrehorek.png) Tomas Rehorek Jan 11, 2023 New Features Recommendation Engine --- # Linear Methods and Autoencoders in Recommender Systems | Blog > Source: https://www.recombee.com/blog/linear-methods-and-autoencoders-in-recommender-systems > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # Linear Methods and Autoencoders in Recommender Systems ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Sep 7, 2021 Linear regression is probably the simplest and surprisingly efficient machine learning method. It should be the method of your first choice, according to the famous [KISS](https://en.wikipedia.org/wiki/KISS_principle) principle. Also, it often works better than sophisticated methods, because it is quite difficult to overfit training data with linear models, thus their generalization performance is not hurt by [overfitting.](https://en.wikipedia.org/wiki/Overfitting) On the other hand, linear models are weak learners, meaning that they tend to underfit data, because of their limited capacity to capture complex relationships. Combining multiple linear models doesn’t help much either, because it is hard to enforce diversity among them (even with [negative correlation learning](https://www.jmlr.org/papers/volume6/brown05a/brown05a.pdf)) and the resulting ensemble is often close to a linear model again. Nevertheless, for sparse multidimensional input data, linear models have always been a good match, and hard to beat when used as a baseline. And in recommender systems, almost every dataset of user-item ratings is sparse and multidimensional. ![](https://www.recombee.com/img/blog/linear-methods-and-autoencoders-in-recommender-systems/autoencoder-2.png) [Autoencoder](https://en.wikipedia.org/wiki/Autoencoder) is a special model trained to reconstruct input vectors, meaning that input and output vectors should match, despite the [information bottleneck](https://en.wikipedia.org/wiki/Information_bottleneck_method) in the hidden layer. You train autoencoders to minimize this reconstruction error on unseen data vectors. The information bottleneck is implemented for example by reducing and subsequently increasing dimensionality of input vectors, by regularizing models to enforce sparsity in the hidden layer, or by a combination of such methods. In computer vision, autoencoders are successfully used to reconstruct images, where minimizing the reconstruction error can be further extended by employing a [perceptual loss](https://arxiv.org/pdf/2001.03444). Deep autoencoders are multilayered networks reducing the dimensionality of images (encoder) and subsequently increasing the dimensionality again (decoder). Those are typically realized by convolutional networks, because images have special properties (e.g. color of pixels in the neighborhood is strongly correlated). Such architectures are also utilized in recommender systems when producing compressed feature vectors (neural embeddings) of [items to be recommended from their images.](https://medium.com/recombee-blog/machine-learning-for-recommender-systems-part-2-deep-recommendation-sequence-prediction-automl-f134bc79d66b) ![](https://www.recombee.com/img/blog/linear-methods-and-autoencoders-in-recommender-systems/autoencoder-1.png) The output of a deep autoencoder should match the input as well as possible. To minimize the reconstruction error, various training techniques and architectures have been developed. The user-item matrix commonly used in Recommender systems (called the rating or implicit preference matrix) is significantly different from images. The absence of local similarities and autocorrelations in sparse multidimensional rating matrices makes utilizing autoencoders from computer vision difficult. ## Linear Autoencoders in Recommender Systems In the case of recommender systems, you can predict the rating of an item using linear models from the rating (or implicit feedback) matrix. Sparse Linear Method ([SLIM](http://glaros.dtc.umn.edu/gkhome/node/774)) trains a sparse weight matrix with a zero diagonal to “reconstruct” the original matrix. Meaning that a recommendation score of an item for a given user is computed as a weighted linear combination of other items that have been rated by the user. The size of the square weight matrix matches the number of items. Therefore it is a special case of sparse matrix factorization. SLIM can also be classified as a sparse autoencoder. The encoding part is an identity (no compression of the user vectors) and decoder is the weight matrix transpose. The only information bottleneck here is the zero diagonal of the weight matrix. In case there can be 1 in the diagonal of the weight matrix, any machine learning algorithm would quickly discover a trivial solution to reconstruction of the rating matrix - place 1 on the diagonal and 0 elsewhere. When diagonal weights are forced to be zero, the machine learning algorithm has to reconstruct the rating using similar items. ![](https://www.recombee.com/img/blog/linear-methods-and-autoencoders-in-recommender-systems/autoencoder-7.png) Here is an example of how the other items that have been rated by a particular user (223) are summed up weighted by their similarities (or co-occurrence) with predicted item (350). ![](https://www.recombee.com/img/blog/linear-methods-and-autoencoders-in-recommender-systems/autoencoder-3.png) SLIM training is straightforward. Reducing the squared reconstruction error (of the Rating matrix A) while regularizing the weight matrix (W) to enforce sparsity and keeping the diagonal zero to avoid a trivial solution. It has been shown later that the stochastic gradient descent works better than the optimization technique proposed in the original SLIM paper. The trained sparse weight matrix can be also used to generate explanations. This item is recommended, because the user rated items that obtained the highest weights in the column of the sparse weight matrix and are therefore considered most similar to the recommended item. Embarrassingly Shallow Autoencoders for Sparse Data ([EASE](https://arxiv.org/abs/1905.03375)) from Harald Steck are quite similar to SLIM, but the weight matrix is dense and training is done in one step by generalized [least squares optimization](https://en.wikipedia.org/wiki/Generalized_least_squares) using the matrix inversion. Both SLIM and EASE have difficulties with datasets where a number of items is large, because of the size of the weight matrix. For more details, have a look at the presentation of Harald Steck on [linear models in recommendation](https://slideslive.com/38943220/linear-autoencoders-for-recommender-systems). ![](https://www.recombee.com/img/blog/linear-methods-and-autoencoders-in-recommender-systems/autoencoder-4.png) The reconstruction of the rating matrix is intuitive - for each item, look at all other items the user has rated (except the predicted item) and increase/decrease the predicted rating based on the corresponding weights (similarity/dissimilarity to predicted item). While SLIM allows only non-negative weights in the Items-to-items matrix, EASE shows that allowing negative weights (dissimilarities) is beneficial. Note that it is also possible to use a [low rank approximation](https://en.wikipedia.org/wiki/Low-rank_approximation) of the rating matrix instead of the sparse items-to-items weight matrix. This solves the scalability issues, because the number of latent features can be as low as a couple of hundreds even for massive datasets. These linear methods ([matrix factorization](https://en.wikipedia.org/wiki/Matrix_factorization_%28recommender_systems%29) or [singular value decomposition](https://en.wikipedia.org/wiki/Singular_value_decomposition)) are quite popular in recommender systems. Again, these methods are linear autoencoders and we use their codes (latent embeddings) extensively in Recombee. ![](https://www.recombee.com/img/blog/linear-methods-and-autoencoders-in-recommender-systems/autoencoder-6.png) The lower is the dimensionality (k) of latent codes, the bigger is the information bottleneck. User vector (row of a rating matrix) is encoded and compressed into a user embedding, item vector (column) is encoded into a latent vector in the item latent matrix. Decoding involves multiplying corresponding user and item latent vectors for each element of the reconstructed rating matrix. Because the reconstruction is a linear operation, the encoder should be also linear even when latent matrices are trained by sophisticated optimization algorithms. For more details, check out [this great doctoral thesis](https://www.skoltech.ru/app/data/uploads/2018/09/Thesis-Fonarev-Final.pdf) on low rank approximation methods for matrix factorization including handling cold start vectors. ## Non-linear Autoencoders Shallow linear autoencoders are weak learners. The nonlinear relations can be modelled by more sophisticated autoencoders, for example [variational autoencoders](https://en.wikipedia.org/wiki/Variational_autoencoder) with fully connected layers. While the code (embedding) of linear autoencoders is just a linear projection of the rating matrix (similar to PCA, but minimizing the reconstruction error does not necessary lead to orthogonal latent features), non-linear autoencoders use multiple layers of neurons with e.g. sigmoid activation functions to compress and decompress the rating matrix. The size of the encoder and decoder does not need to grow proportionally with the size of the rating matrix, therefore their scalability is better than in case of EASE or SLIM models. ![](https://www.recombee.com/img/blog/linear-methods-and-autoencoders-in-recommender-systems/autoencoder-5.png) In the next post, we will introduce our VASP recommendation algorithm, taking the advantage from both linear and non-linear autoencoders. VASP is currently a [state of the art algorithm on the popular Movielens 20M dataset](https://paperswithcode.com/sota/collaborative-filtering-on-movielens-20m). Recommendation Engine ## Next Articles [![](https://www.recombee.com/img/blog/advancing-your-career-in-artificial-intelligence-with-prg-ai-and-recombee.png)](https://www.recombee.com/blog/advancing-your-career-in-artificial-intelligence-with-prg-ai-and-recombee) ### [Advancing Your Career in Artificial Intelligence with prg.ai and Recombee](https://www.recombee.com/blog/advancing-your-career-in-artificial-intelligence-with-prg-ai-and-recombee) At Recombee, we have always collaborated with academia — after all, five of our co-founders graduated from the Czech Technical University in Prague, one of the largest and oldest technical universities in Europe, and most of them hold a Ph.D. degree. ![](https://www.recombee.com/img/blog/authors/karenharazimova.png) Karen Harazimova Nov 21, 2021 [![](https://www.recombee.com/img/blog/recombee-xperience-kentico.png)](https://xperience.io/discover/blog/recombee-kentico-xperience-1-on-1-personalization) ### [Recombee and Kentico Xperience: Guide to One-On-One Personalization](https://xperience.io/discover/blog/recombee-kentico-xperience-1-on-1-personalization) Recombee expanded its integration options - and now is available at the Kentico Xperience platform! Analyzing different types of personalization, we look into why Kentiko chose our AI-powered recommendation engine over manual segmentation. ![](https://www.recombee.com/img/blog/authors/gabrielatakacova.png) Gabriela Takacova May 6, 2021 Personalization Recommendation Engine [![](https://www.recombee.com/img/blog/recombee-in-2020/main.png)](https://www.recombee.com/blog/recombee-in-2020) ### [Recombee in 2020: New Features and Improvements](https://www.recombee.com/blog/recombee-in-2020) We know that this year has been quite challenging for many people, including ourselves. However, today we want to focus entirely on the positive (no pun included) side of the year and the stuff we are the proudest of. ![](https://www.recombee.com/img/blog/authors/ondrejfiedler.png) Ondrej Fiedler Dec 30, 2020 New Features --- # Making Linear Autoencoders Work for Large Scale Recommendation Systems | Blog > Source: https://www.recombee.com/blog/making-linear-autoencoders-work-for-large-scale-recommendation-systems > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # Making Linear Autoencoders Work for Large Scale Recommendation Systems ![](https://www.recombee.com/img/blog/authors/vojtechvancura.png) Vojtech Vancura Aug 29, 2022 Linear autoencoders for collaborative filtering in recommender systems are simple and surprisingly accurate as we explained in our [blogpost on how linear methods work](https://www.recombee.com/blog/linear-methods-and-autoencoders-in-recommender-systems). The critical disadvantage of methods like [EASE](https://dl.acm.org/doi/abs/10.1145/3308558.3313710?casa_token=PX2Csdyy7ZwAAAAA:xXRaGi8aydGPeCLx6ieF19orNgB7204YJZGaZk2bw3UwThH215XTGIH0GFYsoNj4jLbRRCc8jAFq) is that they are not applicable to real-world problems, where the number of items to be recommended is high. In this blogpost, we will explain how we made linear methods applicable even to the largest customers of Recombee with tens of millions of recommendable items. **We [introduce ELSA](https://recsys.acm.org/recsys22/accepted-contributions/#content-tab-1-2-tab),** a scalable linear model that is basically a shallow autoencoder. What is interesting is that ELSA is not only more scalable, but it actually outperforms EASE and other linear methods in a number of tasks. We will first explain how ELSA is different and how we managed to make it scalable. ![](https://www.recombee.com/img/blog/making-linear-autoencoders-work-for-large-scale-recommendation-systems/1.png)![](https://www.recombee.com/img/blog/making-linear-autoencoders-work-for-large-scale-recommendation-systems/2.png)![](https://www.recombee.com/img/blog/making-linear-autoencoders-work-for-large-scale-recommendation-systems/3.png) The problem of linear autoencoders (EASE, SLIM) is that the **Item to Item weight matrix** grows significantly with the number of items and using a sparse representation does not help. It makes these methods practically applicable to problems with tens of thousands of items. However many real-world recommendation problems are orders of magnitude bigger than that. In ELSA, we were able to replace this matrix by a trick allowing us to use a smaller matrix (items x d) instead, where d is a small constant (e.g. 512) you can choose. This trick makes both memory and time complexity of ELSA grow linearly with the number of items. ![](https://www.recombee.com/img/blog/making-linear-autoencoders-work-for-large-scale-recommendation-systems/4.png) As you can see from our results on benchmarking data, the offline performance of ELSA is similar to EASE and it is beneficial to use ELSA when you have more than 20k recommendable items. What is surprising is that ELSA is consistently better than EASE or MF in online tests, probably thanks to a more robust training procedure. We will make ELSA source codes available for everyone. Also, our paper was **[accepted for publication at the RecSyS 2022](https://recsys.acm.org/recsys22/accepted-contributions/#content-tab-1-2-tab)** conference in Seattle, so you can meet us there in case you are interested in our research. Recommendation Engine ## Next Articles [![](https://www.recombee.com/img/blog/ai-powered-content-recommendations-with-a-headless-cms.png)](https://www.luminary.com/blog/ai-content-recommendations-headless-cms-recombee) ### [AI-Powered Content Recommendations With a Headless CMS](https://www.luminary.com/blog/ai-content-recommendations-headless-cms-recombee) Thanks to its API-first nature, it is quite straightforward to integrate your headless CMS with the most powerful AI-powered content recommendations available on the market. Luminary just did that with their own website, Kontent.ai and Recombee. ![](https://www.recombee.com/img/blog/authors/andythompson.png) Andy Thompson (Luminary) Aug 31, 2022 Partnerships [![](https://www.recombee.com/img/blog/new-features-for-a-better-personalization-experience.png)](https://www.recombee.com/blog/new-features-for-a-better-personalization-experience) ### [New Features for a Better Personalization Experience](https://www.recombee.com/blog/new-features-for-a-better-personalization-experience) Like most of the world, the majority of 2021 was spent on home office or in isolation - which left us with all the time to be invested in work (and Netflix :) ) and improving UX for our clients. We are now happy to share new features we can offer to reach new levels of personalization. ![](https://www.recombee.com/img/blog/authors/ondrejfiedler.png) Ondrej Fiedler Mar 07, 2022 New Features [![](https://www.recombee.com/img/blog/advancing-your-career-in-artificial-intelligence-with-prg-ai-and-recombee.png)](https://www.recombee.com/blog/advancing-your-career-in-artificial-intelligence-with-prg-ai-and-recombee) ### [Advancing Your Career in Artificial Intelligence with prg.ai and Recombee](https://www.recombee.com/blog/advancing-your-career-in-artificial-intelligence-with-prg-ai-and-recombee) At Recombee, we have always collaborated with academia — after all, five of our co-founders graduated from the Czech Technical University in Prague, one of the largest and oldest technical universities in Europe, and most of them hold a Ph.D. degree. ![](https://www.recombee.com/img/blog/authors/karenharazimova.png) Karen Harazimova Nov 21, 2021 --- # Item Discovery and Search | Recombee AI > Source: https://www.recombee.com/blog/modern-recommender-systems-part-1-introduction > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # Modern Recommender Systems - Part 1: Introduction ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Apr 17, 2023 How machine learning methods simplify item discovery and search. Modern Recommender Systems * [1. Introduction](https://www.recombee.com/blog/modern-recommender-systems-part-1-introduction) * [2. Data](https://www.recombee.com/blog/modern-recommender-systems-part-2-data) * [3. Objectives](https://www.recombee.com/blog/modern-recommender-systems-part-3-objectives) * 4. Coming Soon Over the last ten years, we have been working on an universal and domain agnostic recommender system. We have learned a lot of things along the way, as our recommender serves hundreds of customers including big brands in all domains you can imagine. In this blogpost, we will be not only sharing our insights, but also to give you a comprehensive overview of the technology behind nowadays recommenders that power almost every major site you use when you search for something online. Additionally, we will discuss related problems and ethical aspects of this technology. In this initial blog post, we will explore the emergence of recommender systems and clarify their role in advertising technology, which is often misunderstood as a single entity. Our goal is to highlight the various objectives that recommender systems fulfill for different stakeholders. In the subsequent blog posts, we will delve into the crucial data and signals that are essential for modern recommenders, followed by an in-depth exploration of the technology behind recommenders. Recommender systems have become pervasive and the most influential machine learning technology since everyone receives hundreds of recommendations on a daily basis, whether it's news to read, songs to listen to, movies to watch, items to purchase, or tweets to see. The number of recommendations for an average active online user has been growing steadily over the years. In the last decade, it has accelerated exponentially and the growth is far from saturation. **Youtube revenues in billions of dollars over the last few years** ![](https://www.recombee.com/img/blog/modern-recommender-systems-part-1-introduction/02.svg) It is hard to estimate precise numbers, but the trend is clear and the growth is driven mostly by following factors: 1. the number of internet users increases 2. average person spends more time online 3. more and more websites and online services adopt recommender systems. [Youtube revenues](https://www.omnicoreagency.com/youtube-statistics/) are dependent on growing active usage of their service and number of recommendations served. We observe a similar trend among our customers. Recommender systems flourish and they are getting better, but first let’s have a look how it started. ## History Recommender systems developed alongside information retrieval systems (IR) in the early seventies thanks to availability of computers and internet. Traditional IR systems were focusing on assisting users when searching large catalogs of items using text queries. Users often inserted their queries via shared computers in public libraries and the output (recommended books) was the same for everyone. No information about users was taken into account and the output was non-personalized. Also even user agnostic IR systems were gradually improved by sequential learning (Information retrieval: A sequential learning process, 1983) and learning to rank algorithms (Learning to rank using gradient descent, Burges, 2005). We describe these machine learning algorithms for IR systems in our [blogpost on personalized search](https://medium.com/recombee-blog/introduction-to-personalized-search-2b70eb5fa5ae). ![](https://www.recombee.com/img/blog/modern-recommender-systems-part-1-introduction/04.png) Personal computers and widespread internet connection enabled personalized recommendations based on history of user actions. One of the first recommendation systems relying exclusively on user historical interactions (explicit ratings of articles) was (GroupLens: An open architecture for collaborative filtering of netnews, 1992). Since then, recommender systems have developed and improved in many directions. Our ambition is to give you a brief overview of the most important techniques that are used in modern recommender systems. In general, nowadays personalized recommendation and search systems are combining many techniques (which will be described in this blogpost series) to optimize various objectives. ## Do Recommenders Really Spy on Users and Generate Targeted Ads? Many people link recommender systems to annoying targeted advertisements. However, such advertisements are often based on simple heuristics and do not use AI based recommender systems at all. ![](https://www.recombee.com/img/blog/modern-recommender-systems-part-1-introduction/01.png) A typical example is abandoned cart retargeting, a type of targeted advertising used by e-commerce websites to encourage users who have left items in their online shopping cart to complete their purchase. When a user adds items to their cart but does not complete the purchase, the e-commerce website may display ads for those specific items to the user as they browse the web or use social media. The list of abandoned products is often a simple reminder and advertisers do not use machine learning to compile it. Another example of targeted advertising is what users see when visiting online media websites, such as news outlets. These ads are also typically not generated by a recommender system, but auctioned by an AdTech platform based on context and user profiles. There are multiple vendors of AdTech platforms, with the most successful being global giants capable of large-scale user profiling. Moreover, most AdTech techniques rely on collecting and analyzing large amounts of data about a user's online activities and preferences in order to create a detailed user profile. This data may include information about a user's browsing history, search queries, online purchases, and online communication, among other things. People are concerned about the potential for this data to be collected and used for targeted ads without their knowledge or consent, or for it to be misused or mishandled in some way. This raises ethical concerns about the right to privacy and the protection of personal information. Recommender systems are primarily used to help users find relevant content on a website. These systems, as opposed to AdTech techniques, typically only consider anonymised user's past interactions with the content on the website and do not utilize external data from other websites or user attributes. As a result, there are likely fewer ethical concerns associated with recommender systems compared to AdTech systems, which may use a wider range of data to personalize ads. ## Goals The goal of classical Information Retrieval methods is to **make search faster and more accurate** by assuming that users know what they are looking for and can formulate it in a query. On the other hand, recommender systems do not require a query and instead focus on inspiring users or helping them to discover new items that they may not even know about. Modern recommender systems incorporate both search and discovery components, allowing users to start typing a query and receive intelligent suggestions to navigate their recommendations in the desired direction. The ambition of modern recommendation systems is to **improve the user experience and optimize user engagement**. While the user is typically the main beneficiary of the recommender system, the goals of the system and the objectives it is set to optimize are defined by the owners, designers and developers of the system rather than by users. Recommenders are often set up to optimize for certain metrics, such as click-through rate or conversion rate and user experience or engagement are not optimized directly because even developers of recommender systems often face the challenge of accurately measuring sophisticated and often subjective metrics such as long-term user engagement. ### Product Owner Perspective ![](https://www.recombee.com/img/blog/modern-recommender-systems-part-1-introduction/06.png) Making a great product so the user base grows and users are actively enjoying the product seems like the right strategy. However there is an extra aspect product owners have in mind and that is increasing the revenue. Subscription based services optimize revenue by maximizing the number of loyal subscribers which is inline with improving quality of product and user experience. Many online businesses however generate revenue by selling advertisements or generating leads for partner sites. ![](https://www.recombee.com/img/blog/modern-recommender-systems-part-1-introduction/03.png) Selling advertisements enables many businesses to offer their product for free. To increase revenue from displaying ads, one typically maximizes the number of page views. This goal can be achieved by tweaking the recommendation system to optimize this objective. In such scenarios, recommenders typically give more visibility to content that has potential to generate additional page views, such as online photo galleries. Product owners need to find a good balance between increasing short term revenues from advertisements and growing the number of active users in the long term. Too aggressive emphasis on maximizing page views typically might lead to significant decrease of user loyalty and less revenues from ads in the long term. No boosting is the other extreme, that is typically adopted only by fully subscription based services. In relation to the business model, apart from fully subscription based services and “free” ads-sponsored services, media organizations started to offer a [combination](https://nypost.com/2021/08/04/spotify-tests-99-cent-a-month-subscription-with-ads/) of these two models - ads sponsored subscription based service. Product owners are motivated to convert “freemium” users into ads sponsored or full subscribers. Recommendations for freemium users can be adjusted in a way that they are getting more content either behind the paywall or content that convinced similar users to subscribe in the past. When a significant portion of revenues is coming from generating leads for partner sites (e.g. in case of online aggregators), the goal is to recommend third-party content that is associated with highest provisions. Again, tweaking the recommender too much in this direction leads to a significant decrease of user engagement in the longer term. Similar situation holds for online retailers willing to maximize revenues by boosting visibility of high margin products. Please note that all these “tricks” are used even by manually curated websites. Recommender systems just help product owners to use them more systematically and efficiently. Instead of displaying high margin content to everyone, they can just increase the probability that relevant users see it. ### Content Producer Perspective ![](https://www.recombee.com/img/blog/modern-recommender-systems-part-1-introduction/07.png) Various websites and services offer content created by artists (such as songs, movies, podcasts, paintings, and poetry), writers (such as books, articles, and news), vendors (such as real-estate listings, products, and online marketplace listings), or other entities (such as job offers). The ultimate aim of these content producers is to attract an audience that will engage with and appreciate their content. They rely on recommender systems to distribute their content to the right users. Many creators aspire to achieve bestseller status, and they rely on recommender engines to make that possible. However, they are also wary of receiving negative reviews and would prefer their content to be directed towards an audience that will respond positively. As such, they expect recommender systems to accurately identify the ideal audience and maximize the likelihood of their content being seen or purchased. Creators often measure their success based on the popularity of their content, and some even tailor their content to maximize certain metrics, such as number of page views generated, time users spent reading the content, and so on. ### User Perspective ![](https://www.recombee.com/img/blog/modern-recommender-systems-part-1-introduction/08.png) Users expect recommender systems to help them reach their goals. However their goals can change even within a single session and possibly be also affected by recommendations. Imagine yourself reading some serious news or educational materials. For some time you are fully engaged, you are actively looking for related content and expect the recommender and search to support you in this process. Slowly, as you get tired, you would like to read some less serious content and get entertained. This is quite a significant change of your goals and objectives to optimize. A good recommender system can deal with such situations even though it is hard to notice that the user goal has changed from the data available. When you visit an e-commerce website, sometimes you are in the mood just to browse interesting products in the catalog and get inspired. This is again a very different objective from the situation when you are about to buy any suitable shoes as fast as possible. You can help users in reaching their goals by having more recommendation scenarios available (e.g. “get inspired” and “your favorites”). Typical user is looking for the best value products, however it is not always the case. Some users prefer premium quality products and their price sensitivity is low, while other users have limited budgets and prefer low-end products. Again, more scenarios (e.g. “best value products for you”, “your premium products”, “cheapest picks for you”) can be available. Advanced recommenders are also able to recommend scenarios (see [Recombee Item Segmentations](https://www.recombee.com/blog/recombee-item-segmentations)) so you do not display irrelevant “cheapest picks for you” to users, who never bought low-end products. The task of the recommender system is to predict user intents in real-time and support them in reaching their goals. Such a task is however very difficult in the environment, where users are not very keen to provide explicit feedback and even implicit historical interactions are limited. Such data needs to be available to enable the recommender system aligning with user goals and intents. ## Problems and Ethical Aspects When the goals of all stakeholders are aligned, the objective of the recommender system is clear and its deployment is straightforward. The problem arises when product owners give emphasis to objectives that work against goals of other stakeholders (users or content producers). ![](https://www.recombee.com/img/blog/modern-recommender-systems-part-1-introduction/05.png) One particular example can be a job board site that publishes job positions and is rewarded for the number of applicants for each position. When the recommender system and personalized search are set to maximize solely this objective, users will be recommended positions they are most likely to apply for, no matter if they have any chance to get accepted. Companies will get a high number of applications for their positions they need to evaluate, which is good. The relevance of candidates will be however quite low, because many of them applied to several positions and they do not have much chance to be accepted. If the recommender system and personalized search are set to optimize for successful applications directly, there will be much less frustration among all participants. Users will get recommended positions where they have a high chance to get accepted. Companies will get less but much more relevant applicants for their positions, so they save time evaluating them. The job board company will have a better product that works well for both users (applicants) and content creators (companies). However such change often involves adjusting not just the recommender objectives, but also changing the business model of the job board company and collecting data about selected applicants for historically offered positions. Even more widespread example is already mentioned optimization for the number of page views in media. Such a strategy might help product owners to generate more revenue from advertisements that are displayed to users with every single page view. Imagine that a car seller orders a certain number of ads that should be displayed with articles in the auto category. Then the product owner will boost the probability that the recommender system suggests these articles to as many users as possible. These users will ask themselves, why are they getting so many articles about cars? And some content producers will be in even more difficult situations. Their insightful articles will not get enough visibility and do not even reach relevant audiences. It is because users do not generate so many page views reading long insightful articles and the recommender engine will favor other shorter articles that make more revenue from ads for the product owner. Why should the product owner sacrifice revenues from additional page views? From our experience it is always better to take into consideration longer term criteria as well. It has a positive effect on user loyalty and increasing revenues in the longer term. Also, many media companies introduced subscription-based models enabling them to optimize recommendations directly to user engagement. Non subscribed users are more likely to subscribe when they are recommended relevant content that is unique or insightful. Recommenders here are often set to balance uplift in page views with the number of new subscriptions. There are still open questions such as: Is it ethical for product owners to tweak recommenders and optimize solely their objectives? Should users be informed about these tweaks? And how? In my opinion, products that significantly deviate from interests of their users or content creators are doomed to lose their market share anyways. **Here are 5 key takeaways from the blog post:** 1. Recommender systems have evolved from traditional information retrieval systems in the early seventies to personalized recommendation and search systems that combine many techniques to optimize various objectives, such as improving engagement, user satisfaction, and revenue. 2. Recommender systems have become an important machine learning technology and are used to make personalized recommendations to users, but they are often confused with AdTech platforms that generate targeted ads. 3. Recommender systems primarily help users find relevant content on a website and typically only consider a user's past interactions with the content on the website, while AdTech platforms may use a wider range of data to personalize ads, which raises ethical concerns about privacy and the protection of personal information. 4. The number of recommendations an average active online user receives has been growing steadily over the years, driven by the amount of time people spend online, and the adoption of recommender systems by more websites and online services. 5. Recommender systems can be configured to optimize different criteria and it is important that they are aligned with objectives of users in order to win their long term engagement. In the next part, we will discuss in detail the data and signals powering recommender systems. And even more importantly, how we can evaluate recommender systems and measure if they deviate from goals they were set to reach. Recommendation Engine Personalization ## Next Articles [![](https://www.recombee.com/img/blog/the-ai-revolution-in-the-media-industry.png)](https://www.recombee.com/blog/the-ai-revolution-in-the-media-industry) ### [The AI (R)Evolution in the Media Industry](https://www.recombee.com/blog/the-ai-revolution-in-the-media-industry) In today's digital age, personalization has become the cornerstone of the media industry. Whether it's tailoring content recommendations, refining marketing strategies, or enhancing user experiences... ![](https://www.recombee.com/img/blog/authors/annapetr.png) Anna Dolezelova & Petr Popov Oct 23, 2023 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/explaining-recommender-systems-to-product-owners.png)](https://pavelkordik.substack.com/p/explaining-recommender-systems-to) ### [Explaining Recommender Systems to Product Owners](https://pavelkordik.substack.com/p/explaining-recommender-systems-to) In my presentation at the Data Technology Seminar organized by the European Broadcasting Union, I have focused on demonstrating that recommender systems can actually help public media organizations to better fulfill their role in society and reduce content distribution biases. ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Mar 27, 2023 Personalization Recommendation Engine [![](https://www.recombee.com/img/blog/inductive-matrix-completion-how-to-improve-recommendations-for-cold-start-users-and-items-by-incorporating-their-attributes.png)](https://www.recombee.com/blog/inductive-matrix-completion-how-to-improve-recommendations-for-cold-start-users-and-items-by-incorporating-their-attributes) ### [Inductive Matrix Completion: How to Improve Recommendations for Cold Start Users and Items by Incorporating Their Attributes](https://www.recombee.com/blog/inductive-matrix-completion-how-to-improve-recommendations-for-cold-start-users-and-items-by-incorporating-their-attributes) Matrix completion (MC), the problem of recovering the missing entries of a partially observed matrix, has found use in a wide range of domains. Still, its potentially most successful application is as a collaborative filtering technique for recommender systems (RSs)... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Mar 20, 2023 Personalization Recommendation Engine --- # Modern Recommender Systems - Part 2: Data > Source: https://www.recombee.com/blog/modern-recommender-systems-part-2-data > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # Modern Recommender Systems - Part 2: Data ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Mar 07, 2024 Data used by modern recommenders and how we can measure progress towards goals. Modern Recommender Systems * [1. Introduction](https://www.recombee.com/blog/modern-recommender-systems-part-1-introduction) * [2. Data](https://www.recombee.com/blog/modern-recommender-systems-part-2-data) * [3. Objectives](https://www.recombee.com/blog/modern-recommender-systems-part-3-objectives) * 4. Coming Soon ## Data Is Crucial Data plays an essential role in the functioning of a recommender system, as it is the primary source of information used to generate accurate and personalized recommendations. In this blogpost, we will discuss the importance of data for recommender systems, the various types of data sources used, and how data can be used to improve the accuracy and effectiveness of recommendations. The data that can be used for recommendations can be categorized into 1) Item catalog 2) User catalog and 3) History of user X item interactions. ![](https://www.recombee.com/img/blog/modern-recommender-systems-part-2-data/1.png) Attributes of items are stored in item catalog, user catalog holds information about users and there are several types of user to item interactions that are recorded in different contexts. ### Item Catalog First of all, it is good to know what we can recommend to users. A database of all items is called an item catalog. In this catalog, we store not only items that can be recommended (active items), but also historical items that were recommended in the past and are not available to users any more. Those historical items are important when measuring similarity of users who interacted with them in the past. Attributes of items help recommenders understand how items are related and which are more similar than others. Here are a few examples of most important item attributes (or item properties). * **Categories** \- Items can be categorized into distinct groups, however you might also come with a hierarchical system of categories where one item can belong to multiple categories. Categories can be used to create item segments so you can recommend particular categories to a given user. You can also filter out items from recommendation based on their category labels or boost probability that items from a particular set of categories are recommended to a user. * **Text descriptions** \- When you recommend articles, the text of the article can be used in a text description attribute of the item. Modern recommenders have capabilities to process text using advanced neural networks. Similarities of **text neural item embeddings** can be very important especially when recommending cold start items that do not have many interactions yet. * **Images** \- Modern recommenders can use multiple images of an item to create an **image neural item embedding.** Again, such information is super important for recommendation systems especially when images play a significant role for users (e.g online art gallery) or when interactions and text descriptions are missing. Imagine an online marketplace where users can upload images of items for sale. As they use their smartphones, it is not likely that they will also add rich and informative text descriptions. Another example would be a real-estate portal, where users like to find similar listings based on images of properties. Or a fashion e-commerce site that decided to utilize visual similarity to recommend alternatives from the product catalog. ### User Catalog Similarly to item catalog, user catalog holds attributes and properties of users. Most important user attributes are the following: * **Location of user** \- Geographic location of users is important in recommendation scenarios, when users are interested in items that are located nearby (such as real estate, job or event recommendation). Even users with no interaction history can then get relevant recommendations such as popular items in their region. * **User search history** \- One can suggest relevant items based on historical user search queries. Also, user search history is instrumental for personalized query suggestions, where reminding users about their past similar queries is very helpful. * **User bio, interests or skills** \- In some domains, it is important to take into consideration not just user interactions with items, but also additional background information that can reveal user interests and help to select relevant items. Again, this is particularly important in cold start scenarios where we need to recommend to users without historical interactions. ### Problems and Challenges of User Catalog User catalogs, while important for personalizing recommendations in modern recommender systems, face several significant challenges. These issues primarily revolve around data privacy concerns, user identification difficulties, and the dynamic nature of user attributes. Addressing these challenges is crucial for maintaining the effectiveness and trustworthiness of recommender systems. #### Data Privacy Concerns In the context of increasing data privacy concerns, it's crucial for recommender systems to responsibly collect and utilize user data to deliver optimal user experiences and enhance product offerings. Regulatory frameworks like the GDPR provide essential guidelines for data handling, yet these should be viewed not as obstacles but as opportunities to foster trust and transparency in the digital ecosystem. Responsible recommenders are pivotal in striking a balance between personalization and privacy. By employing data minimization strategies, pseudo-anonymizing user information, and ensuring robust data protection measures, recommender systems can offer highly personalized experiences without compromising user privacy. #### User Identification Difficulties Accurately identifying users is fundamental to creating and maintaining useful user profiles. However, several issues complicate user identification: * **Multiple Profiles:** Users may create multiple accounts on the same platform, leading to fragmented data that hinders a unified view of user preferences and behavior. * **Shared Profiles:** Accounts shared among several users, common in streaming services and online shopping platforms, present a challenge in discerning individual preferences, resulting in less personalized recommendations. * **Cross-Device Identification:** Users frequently access services across multiple devices, making it challenging to link these interactions to a single user profile accurately. These identification challenges can lead to inaccuracies in user profiles, impacting the relevance of recommendations and potentially diminishing user satisfaction. #### Maintaining Up-To-Date User Attributes User preferences, interests, and even geographic locations can change over time. Keeping user attributes up-to-date is important for the accuracy of recommender systems. * **Changing Preferences and Interests:** As users evolve, so do their preferences and interests. A recommendation system that fails to adapt to these changes may continue suggesting irrelevant items, leading to user disengagement. * **Skills and Professional Changes:** In domains like job recommendation systems, users' skills and professional interests may develop, requiring the system to adapt to these changes to remain relevant. * **Mood Variability:** User mood, which can influence content preference (such as music or movies), varies significantly. Capturing and adapting to these transient states poses an additional layer of complexity. These challenges require online platforms to implement mechanisms for regularly updating user catalog and explicit user preferences. The alternative solution is to reduce reliance on user attributes and let recommender systems infer preferences of users from their interactions with items, incorporating feedback loops, and employing adaptive algorithms capable of adjusting to changes in user behavior and preferences. Where subscription based services can typically supply recommender system with rich user profiles, online platforms that rely on advertising revenue can have as much as 70 percent of anonymous active users with short and recent interaction history. For such users, recommender systems rely on simple session based algorithms such as multi armed bandits. When an anonymous user logs into the platform, the recommender system should be able to [merge](https://docs.recombee.com/api#set-view-portion) browsing histories. Modern platforms should be able to balance personalization with privacy and transparency. As recommender systems evolve, so too must the strategies for managing user catalogs, ensuring that they continue to offer relevant, timely, and engaging recommendations in a privacy-conscious manner. Nice inspiration are [recent developments in managing personal profiles](https://openai.com/blog/memory-and-new-controls-for-chatgpt) for large language models. ### User to Item Interactions Interactions of users with items is the most important data source for recommender systems. In extreme cases, reasonable recommendations can be produced exclusively based on the interaction (or rating) matrix, where user to item interactions are typically stored. Such recommendations can be computed for anonymous users interacting with anonymous items meaning that neither item attributes nor user attributes are used. ![](https://www.recombee.com/img/blog/modern-recommender-systems-part-2-data/2.png) User interactions with items are collected in different scenarios, some of which are powered by a recommender system. There are a variety of user interactions that can be used to derive implicit feedback for recommender systems. These include ratings, browsing history, clicks, interactions with content (such as watching a video or liking a post), purchase history, search history and more. The data collected from these interactions can then be used to create user profiles and model user behavior, which can then be used to create personalized recommendations. ![](https://www.recombee.com/img/blog/modern-recommender-systems-part-2-data/3.png) Typical time sequence is that the recommender system is requested for recommendations to a particular user in some scenario. It returns a personalized list of items that is subsequently displayed to the user. When a user engages with some item from the list, it is important to inform the recommender about the user interaction and if the interaction is based on a particular recommendation. For some scenarios, feedback is almost imminent, for other scenarios, it might take days (e.g. personalized newsletter sent by email). ### Problems With Collecting User Feedback Collecting and interpreting user feedback accurately is a cornerstone for the efficiency of recommender systems. However, several challenges complicate this process, impacting the quality of recommendations. Among these challenges, caching recommendations, biased user interactions, and the lack of explicit user feedback are particularly significant. **Caching Recommendations and Its Impact:** To economize on the costs associated with recalculating recommendations for frequent users, some platforms employ a strategy of caching recommendations. This method can lead to reduced costs, improved response times, and provides users the opportunity to explore recommended items more thoroughly. However, this practice introduces a significant issue: users may repeatedly encounter the same items. If the recommender system is not notified of these repeated exposures and cannot adjust accordingly, it misses the critical opportunity to refine recommendations based on the user's demonstrated lack of interest in these repeated items. This oversight often results in a decline in user experience, as the system fails to recognize and adapt to the evolving preferences of the user. **Biased User Interactions:** Bias in user interactions can significantly skew the data that recommender systems rely on. One form of bias, editorial bias, occurs when some recommendation scenarios are curated by editors and presented the same way to all users. Users typically click on several items from these curated lists, creating an artificial interaction similarity among items that are not genuinely similar. This phenomenon can mislead the recommender system into overestimating the relevance of certain items, thereby distorting the recommendation process. **Lack of User Feedback:** Addressing the challenge of collecting user feedback, it's important to acknowledge that most users are reluctant to provide explicit feedback, such as rating items with stars or indicating likes and dislikes. A critical challenge for recommender systems is the absence of explicit or even implicit feedback in many scenarios. For instance, when users are recommended a list of articles and only read the excerpts without further interacting, they may still be satisfied with the recommendations. However, the recommender system receives no feedback signal to reflect this satisfaction. Similarly, in "autoplay" scenarios for music or short videos, users often continue to watch or listen without active engagement, reacting only if the recommendation is particularly unsuitable. This passive consumption can falsely signal to the RS that the user is engaged, leading to misinterpretations of user interest and satisfaction. Additionally, in scenarios where a recommender system generates a vast array of items but presents only a select few to the user, it becomes essential for the system to recognize that users may not view the recommendations positioned lower on the list. Misinterpreting a user's non-interaction with these less-visible items as a lack of interest can skew the system's perception of user preferences. Furthermore, there are instances where recommendations may not be seen by the user at all, such as when they are placed far down on a webpage and the user does not scroll sufficiently to encounter them. In such cases, the system's assumption that the user has seen and disregarded these recommendations is flawed. Ideally, recommendations should be requested and displayed to the user dynamically, minimizing the time gap between generation and presentation to ensure that users are exposed to relevant recommendations in a timely manner. To effectively address these challenges, it's critical to enhance the quality of feedback loops and the accuracy of data provided to the recommender system. The more precise and comprehensive the user feedback, the more tailored the recommendations can be. For instance, [tracking engagement metrics](https://docs.recombee.com/api#set-view-portion) such as the portions of a video watched, segments of a song listened to, or parts of an article read can offer deeper insights into user preferences. Additionally, recommender systems need to employ advanced techniques to identify and correct biases, improve data quality, and develop methods for gauging user satisfaction beyond their immediate interactions. Furthermore, fostering an environment of transparency and encouraging users to offer direct and explicit feedback on the recommendations they receive can significantly improve the feedback loop, thereby elevating the overall performance of the recommender system. ## Conclusion Data stands at the core of modern recommender systems, fueling the generation of personalized and precise recommendations. The effectiveness of these systems hinges on their ability to leverage diverse data sources, including item catalogs, user catalogs, and user-item interactions. By understanding the attributes of both items and users, along with their interaction history, recommender systems can navigate the complexities of personalization, privacy, and changing user preferences to provide relevant recommendations. However, challenges such as difficulty of user identification, and the dynamic nature of user attributes necessitate advanced strategies to maintain the data useful for improving user experience. Furthermore, accurate collection and interpretation of user feedback are essential for refining recommendation algorithms and enhancing user satisfaction. Here are main takeaways from the article: * **Data Categorization:** Recommender systems rely on item catalogs, user catalogs, and the history of user-item interactions to generate recommendations. * **Item Catalog Importance:** Attributes stored in the item catalog, like categories, text descriptions, and images, help in understanding item relationships and preferences for better recommendations. * **User Catalog Challenges:** Data privacy, user identification difficulties, and the need for up-to-date user attributes present significant challenges in maintaining useful and actual user profiles. * **User to Item Interactions:** The most crucial data source for recommender systems, enabling the creation of personalized recommendations based on user behavior. * **Feedback Collection Challenges:** Issues such as caching recommendations, biased user interactions, and lack of explicit feedback pose challenges to the effectiveness of recommender systems. * **Privacy and Personalization Balance:** Modern platforms must navigate the delicate balance between providing personalized experiences and respecting user privacy. * **Advanced Data Quality Strategies:** Employing advanced techniques to address biases, improve data quality, and adapt to user behavior changes is essential for the continued relevance and effectiveness of recommender systems. Recommendation Engine Personalization ## Next Articles [![](https://www.recombee.com/img/blog/elevate-your-personalization-strategy-with-recombees-innovative-features.png)](https://www.recombee.com/blog/elevate-your-personalization-strategy-with-recombees-innovative-features) ### [Elevate Your Personalization Strategy with Recombee's Innovative Features](https://www.recombee.com/blog/elevate-your-personalization-strategy-with-recombees-innovative-features) The digital landscape and customer preferences and behavior are changing faster than ever now. To help our clients stay on top of the game, our team has focused on developing innovative features... ![](https://www.recombee.com/img/blog/authors/janvaluch.png) Jan Valuch Mar 13, 2024 New Features Recommendation Engine [![](https://www.recombee.com/img/blog/recombee-real-time-ai-recommendations-as-the-new-destination-in-segment.png)](https://www.recombee.com/blog/recombee-real-time-ai-recommendations-as-the-new-destination-in-segment) ### [Recombee Real-Time AI Recommendations as the New Destination in Segment](https://www.recombee.com/blog/recombee-real-time-ai-recommendations-as-the-new-destination-in-segment) Segment has enabled its users to enjoy Recombee personalization services without the need to leave their platform and with minimum coding involved. With a few simple clicks, domains using Segment can upgrade their services to maximize the digital experience for their customers. ![](https://www.recombee.com/img/blog/authors/adelasloupenska.png) Adela Sloupenska Mar 05, 2024 Personalization Integrations Partnerships [![](https://www.recombee.com/img/blog/is-this-comment-useful-enhancing-personalized-recommendations-by-considering-user-rating-uncertainty.png)](https://www.recombee.com/blog/is-this-comment-useful-enhancing-personalized-recommendations-by-considering-user-rating-uncertainty) ### [Is This Comment Useful? Enhancing Personalized Recommendations by Considering User Rating Uncertainty](https://www.recombee.com/blog/is-this-comment-useful-enhancing-personalized-recommendations-by-considering-user-rating-uncertainty) Picture this: you're on the hunt for the perfect new smartphone, browsing through your favourite online electronics store. The online store’s recommendation engine pops up with what it thinks could be your possible next gadget love... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Mar 01, 2024 Recommendation Engine Personalization --- # Modern Recommender Systems - Part 3: Objectives > Source: https://www.recombee.com/blog/modern-recommender-systems-part-3-objectives > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # Modern Recommender Systems - Part 3: Objectives ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Sep 03, 2025 Learning objectives of recommender systems and personalized search. Modern Recommender Systems * [1. Introduction](https://www.recombee.com/blog/modern-recommender-systems-part-1-introduction) * [2. Data](https://www.recombee.com/blog/modern-recommender-systems-part-2-data) * [3. Objectives](https://www.recombee.com/blog/modern-recommender-systems-part-3-objectives) * 4. Coming Soon In the first two parts of this book, we laid the groundwork: understanding what modern recommender systems are and how data fuels them. But knowing the machinery and the raw material is not enough—you also need to know the destination. Every recommendation engine is, at its core, an optimizer: it learns patterns from interactions and then pushes its predictions toward a goal. Early systems optimized implicitly for popularity or similarity, but today’s platforms face far more complex and sometimes conflicting objectives: keeping users engaged without overwhelming them, balancing diversity with relevance, and driving business value without eroding trust. In this third part, we shine a spotlight on these objectives—the “north stars” that guide recommender systems and personalized search. By clarifying what is being optimized, we uncover the hidden logic behind why different platforms make the recommendations they do, and set the stage for translating these goals into concrete learning tasks in the next chapter. ## Examples of Learning Objectives To give you an impression how broad objectives in recommender systems and personalized search can be, we will start with examples in various domains. Content streaming services (e.g., music, video, podcasts) prioritize objectives centered around user engagement and retention: * **Maximizing User Engagement**: Keeping users actively consuming content (e.g., total view time, session duration, content completion). * **Reducing Churn Rate**: Minimizing users canceling subscriptions or ceasing to use the service. * **Accelerating Content Discovery**: Helping users easily find new, enjoyable content, showcasing catalog breadth. * **Balancing Mainstream vs. Niche Content Exposure**: Promoting diverse content to cater to varied tastes and support a healthy content ecosystem. * **User Satisfaction and Perceived Value**: Ensuring users feel recommendations are enjoyable and justify subscription costs or time spent. * **Supporting Creator Ecosystem**: Ensuring fair exposure and monetization for content creators. In general, subscription-based content streaming services focus on optimizing user satisfaction with the service. For free users, the main objective would be to convert them into subscribers (e.g., by recommending highly relevant content beyond the paywall). For ad-powered content streaming services, watch time maximization might be a good strategy to increase revenue from displaying ads. Controversies are discussed in [Part 1, Introduction](https://www.recombee.com/blog/modern-recommender-systems-part-1-introduction). E-commerce platforms typically prioritize objectives focused on driving sales and enhancing customer value: * **Increasing Conversion Rates**: Maximizing the percentage of users who make a purchase after viewing a recommendation or visiting the site. * **Increasing Average Order Value (AOV)**: Encouraging users to purchase more items or higher-value items per transaction. * **Reducing Cart Abandonment**: Minimizing instances where users add items to their cart but leave without completing the purchase. * **Maximizing Customer Lifetime Value (CLV)**: Fostering long-term customer loyalty and repeat purchases through sustained relevance. * **Improving Product Discovery Across the Catalog**: Helping users find relevant products beyond popular items or their immediate search. * **Optimizing Inventory Turnover**: Promoting overstocked items or those nearing end-of-season, balancing business needs with user experience. * **Waking-up Inactive Customers**: Offering targeted discounts or suggesting highly relevant products with limited availability. In e-commerce, it is more about making the customer buy products rather than any other objectives. However, some e-commerce platforms and marketplaces are focusing on generating traffic for other e-shops rather than selling directly. Their focus is therefore shifted towards producing outclicks, especially when purchases associated with outclicks are not reported and rewarded by partner sites. In other domains, objectives might be even more complex. Imagine job boards or dating sites that need to optimize for satisfaction of multiple parties under constraints. Also, there are general level objectives that apply to most scenarios where users interact with personalized recommendations or search. User Satisfaction and Task Completion optimize for successful user sessions that result in items found in a reasonable time (Time-to-Result Optimization). One might also optimize for Relevance, Quality, Diversity, and Freshness of items. In all scenarios, we strive for Abandonment Reduction (e.g., search query, cart, watch next recommendation), which might lead to unsuccessful user sessions. Note that online platforms observe just partial user feedback signals, so all these objectives are typically optimized in a noisy environment. One might ask how particular objectives are defined for a specific online platform and individual use cases. Typically, this is done through careful analysis of user needs, business requirements, and strategic objectives. These goals typically emerge from stakeholder discussions, user research, and business strategy sessions. The objectives of modern recommender and search systems involve multiple stakeholders often with conflicting interests (e.g. users, content creators, editors, business). Effective optimization seeks to balance and align these. We can broadly categorize these critical objectives as follows. ## Taxonomy of Learning Objectives To make sense of the many and often competing goals in recommender systems, it helps to group them into broader categories. This taxonomy of learning objectives highlights four perspectives—user, content, business, and product—that together capture the full landscape of what modern systems are designed to optimize. ![](https://www.recombee.com/img/blog/modern-recommender-systems-part-3-objectives/1.png) ### User Objectives Focused primarily on satisfying and engaging end-users, these metrics capture how effectively the system delivers personalized experiences. * **Engagement**: Measures the extent of active user interactions (clicks, views, session length, and return visits) indicating user interest and commitment to the content. * **Relevance**: Ensures recommended items align closely with user preferences, past behavior, and explicit user feedback. * **Novelty and Serendipity**: Goes beyond traditional relevance to introduce users to fresh, unexpected yet satisfying recommendations, keeping user experiences interesting and avoiding monotonous or predictable content. * **Trust and Transparency**: Users prefer transparent, explainable recommendations that build trust and confidence in the system's decisions, especially important in sensitive or high-stakes scenarios. ### Content Objectives These objectives ensure the breadth, richness, and balanced representation of available content. * **Diversity**: Guarantees variety within recommendations, preventing repetition, echo chambers, or overly similar content. * **Coverage**: Refers to the proportion of the content catalog effectively recommended and utilized, ensuring both niche and popular items have a fair opportunity for exposure. * **Freshness**: Prioritizes new or timely content, critical for domains where recency significantly impacts user satisfaction (e.g., news, trends, social media). * **Locality**: Ensures content relevance based on geographic, cultural, or regional context, where content that's highly relevant for users in one area may be irrelevant or inappropriate for others (e.g., local news, regional events, location-specific services, cultural content). ### Business Objectives Reflecting economic and strategic goals of an online platform, these metrics typically justify the investment in a recommender system or personalized search solution. * **Profitability**: Recommendations should directly or indirectly enhance revenue by increasing sales, upselling, cross-selling, or improving monetization opportunities. * **Cost Efficiency**: Systems should optimize resource utilization, reducing computational costs and data processing overhead. * **User Retention and Loyalty**: Strong recommendation systems support long-term customer relationships, reducing churn and boosting lifetime customer value. ### Product Objectives These objectives ensure that the recommender system contributes positively to the overall product experience, reputation, and ethical considerations. * **Speed and Responsiveness**: Recommendations must be fast and timely, ensuring that latency does not degrade user experience, especially critical in real-time scenarios. * **Brand Consistency**: Recommendations must align with the overall brand identity, supporting brand image and maintaining consistent messaging and quality expectations. * **Fairness and Ethics**: Recommenders should proactively avoid biases, stereotypes, or unfair treatment of user groups or content providers. Fairness also encompasses equitable representation and opportunities for less prominent content providers. * **Compliance**: Systems must responsibly handle user data and adhere to legal/ethical frameworks (e.g., GDPR, AI Acts), ensuring privacy and lawful processing. * **Security**: Systems must be resilient to malicious activities (e.g., attacks, hacking), safeguarding integrity, data, and reliability. ## Balancing Objectives in Real-World Recommender Systems The key to operationalizing these diverse objectives is aligning them with measurable metrics that can, in turn, be optimized through specific machine learning tasks (see next chapter). For example, user engagement might be measured via session length and click-through rates, while content diversity could be quantified using intra-list similarity scores. The star plot above illustrates how different objective categories like user objectives (engagement, relevance, novelty), content objectives (diversity, coverage, freshness), business objectives (profitability, cost efficiency, retention), and product/ethical objectives (speed, brand consistency, fairness) form a multi-dimensional optimization space. Modern recommender systems rarely optimize a single goal—they balance several at once, often through multi-stage pipelines. Netflix, for example, blends candidate generation for relevance with [re-ranking for freshness and diversity](https://www.vamsitalkstech.com/ai/industry-spotlight-engineering-the-ai-factory-inside-netflixs-ai-infrastructure-part-3/), applies business filters, and enforces fairness constraints, all within milliseconds. Spotify faces similar challenges: its personalized playlists mix familiar tracks with exploration of new or lesser-known artists to [keep listeners engaged without creating monotony](https://newsroom.spotify.com/2023-03-06/responsibly-balancing-what-goes-into-your-personalized-recommendations/). The danger of over-optimizing one objective is very real. A Spotify [field experiment](https://arxiv.org/abs/2003.08203) showed that personalized podcast recommendations increased streams by 29% but reduced listening diversity by more than 11%, meaning users became more engaged but in narrower, less healthy patterns. This highlights how optimizing purely for engagement can harm long-term satisfaction. Similarly, academic and industry discussions frequently warn against [filter bubbles](https://www.recombee.com/blog/breaking-the-news-the-role-of-ai-in-modern-journalism) and popularity bias, where too much focus on relevance or clicks leads to [stale, repetitive experiences](https://www.music-tomorrow.com/blog/fairness-and-diversity-in-music-recommendation-algorithms) and reduces fairness in content exposure. I also witnessed this problem firsthand with one customer. They boosted recommendations that generated higher out-click revenue, which initially looked like a win for the business. But over time, user engagement declined because the system was pushing too aggressively toward monetizable items at the expense of user satisfaction. In the end, they had to scale back the boost significantly to protect the overall experience. The lesson across these cases is that objectives are interconnected: relevance, diversity, engagement, revenue, and fairness must be optimized in concert. ## Customizing Recombee to Meet your Objectives Recombee's recommendation engine uses modular Logics (algorithms/ensembles) and Scenarios (named use-cases) to optimize for a wide range of objectives. ### Scenarios A [Scenario](https://docs.recombee.com/scenarios) in Recombee represents a specific place in the application where recommendations are shown, such as a box on a product detail page, a watch-next screen, or a newsletter slot. Each Scenario defines the context and purpose of recommendations for that particular use case, creating a named configuration that can be easily managed by product or editorial teams within the Recombee web interface. When an application requests recommendations using a particular Scenario ID, Recombee executes the defined configuration to deliver a precisely tailored and contextually appropriate list of items for that specific use case. ### Logics At the heart of every Scenario is a [Logic](https://docs.recombee.com/recommendation_logics)—a named ensemble of recommendation models. Recombee provides a variety of Logics that are either universal or domain-specific, enabling targeted optimization for each industry. Many of these Logics have additional parameters for tuning their behavior (e.g., whether to recommend already watched content or not). #### Universal Logics These are applicable across domains and address general-purpose personalization: * [recombee:personal](https://docs.recombee.com/recommendation_logics#recombee-personal) – Personalized ranking of items for a user, based on the user's interaction history and user properties, typically used on homepages or dashboards. * [recombee:similar](https://docs.recombee.com/recommendation_logics#recombee-similar) – Items similar to a given item (both interaction-wise and content-wise), commonly used on detail pages. * [recombee:popular](https://docs.recombee.com/recommendation_logics#recombee-popular) – Items that get a lot of interactions within the whole user base, or within a specific user segment. #### Domain-Specific Logics Recombee also provides Logics fine-tuned for specific verticals: * Video & OTT: [video:watch-next](https://docs.recombee.com/recommendation_logics#video-watch-next), [video:continue-watching](https://docs.recombee.com/recommendation_logics#video-continue-watching), [video:editors-picks](https://docs.recombee.com/recommendation_logics#video-editors-picks), etc. * News & Media: [news:daily-news](https://docs.recombee.com/recommendation_logics#news-daily-news), [news:trending](https://docs.recombee.com/recommendation_logics#news-trending), [news:categories-for-you](https://docs.recombee.com/recommendation_logics#news-categories-for-you), etc. * E-commerce: [ecommerce:cross-sell](https://docs.recombee.com/recommendation_logics#ecommerce-cross-sell), [ecommerce:similar-products](https://docs.recombee.com/recommendation_logics#ecommerce-similar-products), [ecommerce:bestseller](https://docs.recombee.com/recommendation_logics#ecommerce-bestseller), etc. These Logics incorporate domain-specific behaviors, signals, and diversity models out of the box. ### Custom Settings and Rules In addition to selecting an appropriate Logic for each Scenario, Recombee allows fine-tuning each recommendation request through various custom settings and rules that help align the system with specific objectives: * [Filters](https://docs.recombee.com/scenarios#filters) – rules to limit which items can appear (e.g., hide out-of-stock products, recommend only articles from certain categories and of a certain age). * [Boosters](https://docs.recombee.com/scenarios#boosters) – rules that bias the recommender engine toward recommending certain items more (e.g., promote discounted items or recent articles). * [Constraints](https://docs.recombee.com/scenarios#constraints) – rules that enforce diversity in recommended items (e.g., limit the number of items from a single brand in a recommendation). These customizable elements allow organizations to adapt the recommendation behavior to their specific business requirements, editorial policies, and user experience goals without modifying the underlying machine learning models. ## How Recombee Logics & Scenarios Optimize Diverse Objectives ### User Objectives Focused primarily on satisfying and engaging end-users, these metrics capture how effectively the system delivers personalized experiences that create value for the people actually using the platform. * **Engagement**: Infinite feed scenarios with fresh content after refresh on next visit. Customer Lifetime Value (CLV) optimization through sustained interaction patterns. Logics like video:continue-watching and news:daily-news maintain user interest across sessions, while the automatic exploration algorithms prevent content fatigue. * **Relevance**: Automatic optimization through recombee:personal and similar logics that learn from user behavior patterns and preferences. Many logics that do not have “personal” explicitly stated in their names still utilize smart algorithms to ensure recommended items are relevant for a particular user. * **Novelty and Serendipity**: Automatic optimization through user history analysis and exploration algorithms that introduce users to unexpected but relevant content. Diversity constraints prevent filter bubbles, while logics like video:editors-picks surface curated content users might not discover organically. * **Trust and Transparency**: Built-in data protection (avoiding external data enrichment) with comprehensive tools and insights for recommender system operators and editors to understand and explain system behavior. Clear scenario naming and logic selection help users understand why certain content is being recommended. ### Content Objectives These objectives ensure the breadth, richness, and balanced representation of available content, preventing the marginalization of niche or emerging content while maintaining editorial quality and brand standards. * **Diversity**: Automatic diversity optimization through exploratory algorithms and configurable constraints that ensure recommendation slates are diverse across multiple dimensions (genre, topic, creator, recency). Constraints prevent over-concentration of similar items, while boosters can promote underrepresented categories. * **Coverage**: Automatic recall-coverage tradeoff optimization ensuring niche users discover niche content, preventing the long-tail from being overlooked. Special algorithms like beeFormer are capable of recommending content without interactions using semantic attribute similarity. * **Freshness**: All news logics incorporate automatic exploration of recent content. Dedicated scenarios with time-based filters ensure timely content surfacing, while boosters can prioritize newly published items. This prevents recommendations from becoming stale and ensures users stay current with latest developments. * **Locality**: Boosting content by geographic distance and user location preferences, enabling region-specific and culturally relevant recommendations. Filters can restrict content to specific regions, while location-aware logics surface content that resonates with local interests and cultural context. See Recombee online [blogpost](https://www.recombee.com/blog/how-regionalization-based-recommendations-can-improve-your-operations) for more. ![](https://www.recombee.com/img/blog/modern-recommender-systems-part-3-objectives/2.png) ### Business Objectives Reflecting economic and strategic goals of online platforms, these metrics typically justify the investment in recommender systems and align recommendation strategy with revenue generation and operational efficiency. * **Profitability**: E-commerce upsell and cross-sell through business rules and logics like ecommerce:cross-sell; subscription-based services balance engagement-only content for free users with premium content promotions to drive conversions through strategic boosters; increased page views generate more ad impressions through optimized infinite scroll scenarios; affiliate and outbound click optimization through targeted boosting of monetizable content. * **Cost Efficiency**: Recombee runs a private cloud with almost thousand servers across the globe. All algorithms and data storage systems are implemented in an extremely efficient way to reduce operational overhead while providing enterprise-grade performance and reliability. * **Customer Retention and Loyalty**: Personalized experiences through recombee:personal and news:daily-news foster loyalty through niche content discovery and habit formation. Diversification models prevent filter bubbles and maintain long-term engagement by introducing variety that keeps users returning over extended periods. ### Product Objectives These objectives ensure that the recommender system contributes positively to the overall product experience, reputation, and ethical considerations while maintaining technical excellence and regulatory compliance. * **Speed and Responsiveness**: Automatic performance optimization ensuring low-latency recommendations across all scenarios, with sub-100ms response times that don't degrade user experience. Efficient massively parallelized algorithms and data pipelines maintain performance even under high load. * **Brand Consistency**: Logics like video:editors-picks combined with filters, boosters, and constraints enable curated content that aligns with brand values and editorial standards. Custom filters ensure only brand-appropriate content appears in recommendations, while boosters can promote content that reinforces brand identity. * **Fairness and Ethics**: Built-in algorithmic fairness measures and bias mitigation strategies deployed automatically across all recommendation scenarios. Diversity constraints prevent discrimination, while balanced exposure algorithms ensure equitable treatment of content creators and fair representation across demographic groups. * **Compliance**: Automatic adherence to data protection regulations (GDPR, CCPA) and industry standards without requiring manual configuration. Privacy-by-design architecture ensures user data is handled securely, while audit trails provide transparency for regulatory review. * **Security**: Automatic security measures protecting against malicious attacks (recommendation poisoning, data breaches) and ensuring system integrity. Rate limiting, input validation, and secure data handling protect both the platform and its users from potential threats. Beyond the configurable logics, Recombee inherently manages several critical aspects to ensure a high-quality, reliable service. The system's architecture is built from the ground up for speed and scalability, consistently delivering recommendations with low latency, even under high-demand scenarios. Security and user privacy are foundational pillars, handled at the platform level in adherence with best practices and regulatory requirements, without necessitating direct user configuration. Furthermore, many core recommendation approaches, such as those powering homepages or email campaigns, incorporate built-in mechanisms to promote diversity and fair exposure of items. This proactive approach helps prevent users from being confined to filter bubbles and ensures a broader range of content gets a fair opportunity to be discovered. Recombee's modular Logics and Scenarios provide a flexible, robust way to optimize for a broad spectrum of recommendation objectives—many of which are handled "out of the box" by the system, freeing teams to focus on high-level strategy rather than low-level tuning. In summary, defining objectives in recommender systems is an iterative, stakeholder-driven process that balances competing user, content, business, and product goals. Modern systems must navigate complex trade-offs between engagement and diversity, relevance and coverage, profitability and fairness; all while maintaining technical performance and ethical standards. Key lessons from industry practice show that successful objective definition requires measurable metrics that can be tracked over time, clear prioritization when objectives conflict, and regular reassessment as business priorities evolve. The most effective systems establish objective hierarchies where primary goals (like user engagement) are supported by secondary objectives (like content diversity) that prevent long-term degradation. In the next Chapter, we will examine how these diverse objectives translate into specific machine learning tasks—the technical foundation that enables modern recommender systems to optimize for multiple goals simultaneously. Recommendation Engine Personalization ## Next Articles [![](https://www.recombee.com/img/blog/recombee-partners-with-the-telegraph-to-deliver-ai-driven-personalisation-to-millions-of-readers.png)](https://www.recombee.com/blog/recombee-partners-with-the-telegraph-to-deliver-ai-driven-personalisation-to-millions-of-readers) ### [Recombee Partners with The Telegraph to Deliver AI-Driven Personalisation to Millions of Readers](https://www.recombee.com/blog/recombee-partners-with-the-telegraph-to-deliver-ai-driven-personalisation-to-millions-of-readers) Prague, 11th September 2025 – Recombee, a leading AI-powered recommendation platform, has announced a strategic partnership with The Telegraph... ![](https://www.recombee.com/img/blog/authors/recombee.png) Recombee Oct 02, 2025 Company News [![](https://www.recombee.com/img/blog/the-building-blocks-of-privacy-friendly-personalization.png)](https://www.recombee.com/blog/the-building-blocks-of-privacy-friendly-personalization) ### [The Building Blocks of Privacy-Friendly Personalization](https://www.recombee.com/blog/the-building-blocks-of-privacy-friendly-personalization) Personalization can be achieved without compromising user privacy. While many personalization systems have historically relied on practices now considered intrusive, like... ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Aug 07, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/no-code-search-widget-personalized-powerful-effortless.png)](https://www.recombee.com/blog/no-code-search-widget-personalized-powerful-effortless) ### [No-Code Search Widget: Personalized, Powerful, Effortless](https://www.recombee.com/blog/no-code-search-widget-personalized-powerful-effortless) At Recombee, we don't just excel at recommendations – we provide powerful full-text search capabilities too. Our Quick, No-Code Search Widget exemplifies this, offering a seamless... ![](https://www.recombee.com/img/blog/authors/ondrejfiedler.png) Ondrej Fiedler Jul 25, 2025 Recommendation Engine Personalization --- # AI Personalization Through Content Recommendations | Video > Source: https://www.recombee.com/blog/real-time-personalization-of-content-with-ai-powered-recommendations > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # Real-Time Personalization of Content With AI-Powered Recommendations ![](https://www.recombee.com/img/blog/authors/karenharazimova.png) Karen Harazimova Sep 16, 2022 Do you manage a publishing company, online gaming platform, or a streaming site with a content-heavy catalog and are thinking about how to improve the user experience? You can help your users find what they are looking for and engage them with content using AI-powered personalization. How? With 100+ machine learning algorithms, Recombee can analyze content and user behavior with AI and provide each user with tailored content throughout their experience on the given site. **You can see how AI recommendations can personalize content in the short video below.** The video is recorded in an anonymous window, where the selected scenarios recommend content for returning viewers as well as first-time visitors. As the particular user has not made any interactions yet, the homepage scenarios **“Trending Now”, “New Releases”** or **“Popular” show a mix of content that is popular among the users.** Once the user makes the first interaction, for example, clicks on a cartoon movie, the journey continues at the detail page where new recommendations are updated in real-time. In the instance of cartoon movies, detail page scenarios such as **“Related Content”** or **“Watch Next”** show content from the same category based on the content attributes similarities. Once the user returns to the homepage, the content offering is personalized based on the user’s on-site behavior. All Recombee’s [content recommendations](https://www.recombee.com/content-recommendations) are AI-powered, real-time, and relevant to each individual. With every new click or view, the recommendations are seamlessly updated and consider the changes in user preferences as they browse. Alongside to the VOD example, content recommendations are applicable to other domains, such as media companies, online gaming, music, or podcasts. In Recombee’s recent [case studies](https://www.recombee.com/case-studies), content recommendations have led to **over a 40% increase in CTR and reported an improvement in user experience and engagement.** For personalization-seeking sites, Recombee provides a free recommendations audit that analyzes the current state of personalization and consults the best fit scenarios. You may start by reviewing commonly-used scenarios in our [Scenario Setup Guide for Content Recommendations](https://www.recombee.com/handbook/download-content-recommendations-35kl362g0946239g0er) and arrange for an audit call at [business@recombee.com](mailto:business@recombee.com) with one of our specialists. ## Let’s connect! Interested in a custom personalization roadmap for your business? Meet our team and let’s talk about recommendations. **Our specialist’s at your disposal** ![](https://www.recombee.com/img/team/petr-popov.png) For a recommendation audit contact Petr [petr.popov@recombee.com](mailto:petr.popov@recombee.com) ![](https://www.recombee.com/img/team/filip-hanus.png) For integration inquiries contact Filip [filip.hanus@recombee.com](mailto:filip.hanus@recombee.com) ![](https://www.recombee.com/img/team/karen-harazimova.png) For partnership collaboration contact Karen [karen.harazimova@recombee.com](mailto:karen.harazimova@recombee.com) Personalization ## Next Articles [![](https://www.recombee.com/img/blog/repsys-opensource-library-for-interactive-evaluation-of-recommendation-systems.png)](https://www.recombee.com/blog/repsys-opensource-library-for-interactive-evaluation-of-recommendation-systems) ### [Visual and Interactive Evaluation of Recommender Systems](https://www.recombee.com/blog/repsys-opensource-library-for-interactive-evaluation-of-recommendation-systems) When building modern real-world artificial intelligence systems, it is increasingly important to validate that the system works correctly. This is however not an easy task. Existing tools for machine learning practitioners... ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Sep 19, 2022 Recommendation Engine [![](https://www.recombee.com/img/blog/ai-powered-content-recommendations-with-a-headless-cms.png)](https://www.luminary.com/blog/ai-content-recommendations-headless-cms-recombee) ### [AI-Powered Content Recommendations With a Headless CMS](https://www.luminary.com/blog/ai-content-recommendations-headless-cms-recombee) Thanks to its API-first nature, it is quite straightforward to integrate your headless CMS with the most powerful AI-powered content recommendations available on the market. Luminary just did that with their own website, Kontent.ai and Recombee. ![](https://www.recombee.com/img/blog/authors/andythompson.png) Andy Thompson (Luminary) Aug 31, 2022 Partnerships [![](https://www.recombee.com/img/blog/making-linear-autoencoders-work-for-large-scale-recommendation-systems.png)](https://www.recombee.com/blog/making-linear-autoencoders-work-for-large-scale-recommendation-systems) ### [Making Linear Autoencoders Work for Large Scale Recommendation Systems](https://www.recombee.com/blog/making-linear-autoencoders-work-for-large-scale-recommendation-systems) Linear autoencoders for collaborative filtering in recommender systems are simple and surprisingly accurate as we explained in our blogpost on how linear methods work. The critical disadvantage of methods like EASE is that they are not applicable to real-world problems... ![](https://www.recombee.com/img/blog/authors/vojtechvancura.png) Vojtech Vancura Aug 29, 2022 Recommendation Engine --- # AI Adoption in the Media Industry | Recombee Personalization > Source: https://www.recombee.com/blog/the-ai-revolution-in-the-media-industry > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # The AI (R)Evolution in the Media Industry ![](https://www.recombee.com/img/blog/authors/annapetr.png) Anna Dolezelova & Petr Popov Oct 23, 2023 ![](https://www.recombee.com/img/blog/the-ai-revolution-in-the-media-industry/main.png) ## Introduction In today's digital age, personalization has become the cornerstone of the media industry. Whether it's tailoring content recommendations, refining marketing strategies, or enhancing user experiences, the power of personalization is revolutionizing how media is consumed. At the heart of this transformation lies artificial intelligence (AI), a technology that has seamlessly integrated into the media sector, reshaping the way we engage with content. In this article, we delve into the insights and experiences of media professionals to uncover the evolving landscape of AI in the media industry. To gain a comprehensive perspective on AI adoption in the sector, we turn to industry experts who have applied AI across an array of use cases. ## Industry Overview AI adoption in the media industry is not an isolated phenomenon. According to a [recent report by PwC](https://www.pwc.com/gx/en/issues/data-and-analytics/publications/artificial-intelligence-study.html), AI is poised to transform the entire media and entertainment landscape. It predicts that by 2030, AI will play a crucial role in content creation, distribution, and personalization, resulting in a more immersive and engaging user experience. Across the industry spectrum, there’s a mosaic of AI use cases. On the positive side, AI has brought about efficiency and automation in production, cost savings, advanced CGI effects, personalized marketing, and innovations in virtual production. It has transformed everything from camera recording to content creation in movies. First and foremost, AI-driven personalization is a defining trend in the media landscape. It's revolutionizing content delivery, pricing, and marketing, offering users tailored experiences. AI-driven cameras are revolutionizing the way sporting events are captured and analyzed, providing viewers with a more immersive and dynamic experience. For example, during live broadcasts of football matches, AI-powered cameras can automatically track players and provide close-up shots of key moments, enhancing viewer engagement. Fraud detection mechanisms, another AI application, are critical in maintaining trust and integrity in the media industry. As the consumption of digital content continues to rise, so does the risk of fraud, including ad fraud and copyright infringement. AI algorithms can analyze vast amounts of data in real-time to detect suspicious activities and protect both content creators and consumers. Churn prevention, a vital aspect of user retention, relies on AI-powered analytics to understand customer behavior and preferences. By identifying patterns that may indicate a user is considering canceling a subscription or discontinuing engagement, media companies can take proactive measures to retain their audience. It's essential to recognize that while AI holds immense potential, it is also a potentially dangerous technology that requires careful consideration. Therefore, regulations may be necessary to ensure responsible and ethical AI usage in the media industry. ## Integration of AI Into Media Workflows ![](https://www.recombee.com/img/blog/the-ai-revolution-in-the-media-industry/01.png) Implementing AI solutions often presents unforeseen challenges. Considering the newness of AI, extensive research is essential to navigate this transformative landscape. We need to underscore the need for transparency in AI adoption and the importance of explaining how AI benefits users. The vision extends to a future where all content is personalized and dynamically generated. AI implementation often requires a cultural shift within media organizations. It involves educating and upskilling employees to work alongside AI systems effectively. This change management process can be challenging, but it is crucial for the successful integration of AI into media workflows. ## Latest Trends Looking ahead, media professionals foresee a host of trends that will continue to shape the industry. These trends include enhanced personalization, content generation automation, and the integration of AI with emerging technologies such as augmented reality. **AI-driven personalization** is set to reach new heights. With the growing volume of digital content available, users expect tailored recommendations that align with their preferences. AI algorithms will increasingly leverage user behavior data, historical interactions, and even biometric signals to curate content that caters to individual tastes. **Content generation automation** is also on the horizon. AI-powered tools can assist in creating written articles, video scripts, and even music compositions. This trend is expected to streamline content production processes and enable media companies to produce a wider variety of content at a faster pace. ![](https://www.recombee.com/img/blog/the-ai-revolution-in-the-media-industry/02.png) **The fusion of AI with emerging technologies** like augmented reality (AR) promises to deliver immersive and interactive media experiences. AR applications will enable users to engage with content in new ways, such as virtually trying on clothing or exploring 3D visualizations of historical events. ## AI-Powered Personalization Personalization, enabled by AI, is at the core of these trends. AI is driving strategies that tailor content delivery, pricing models, and marketing campaigns to individuals' preferences, thereby elevating user experiences. This shift towards personalized content consumption is reshaping the media landscape. User data privacy and security are paramount in the era of AI-driven personalization. Media companies should adopt robust data protection measures and ensure compliance with privacy regulations, starting with (and aiming far beyond) the notorious GDPR. Balancing personalization with data security is essential to build and maintain user trust. ## The Role of Recommender Systems ![](https://www.recombee.com/img/blog/the-ai-revolution-in-the-media-industry/03.png) Recommender systems, a subset of AI applications, have become instrumental in shaping content consumption habits. Media professionals with experience in these systems provide valuable insights into their benefits and challenges. Recommender systems rely on AI algorithms to analyze user preferences and behaviors, making content discovery more efficient and tailored to individual tastes. [In our previous article](https://www.recombee.com/blog/breaking-the-news-the-role-of-ai-in-modern-journalism), we discuss the transparency and responsibility in recommendation systems for public media. These systems have significantly influenced how viewers discover and engage with content across various media platforms. ### Technical and Ethical Concerns Adapting to AI involves not only initial implementation but also ongoing maintenance and continuous software coding. Professionals in the field are acutely aware of the ethical dilemmas posed by AI, especially in the context of achieving true consciousness and ethical considerations in post-production processes. Some of the pertaining concerns are: 1. Building and maintaining effective recommender systems are complex endeavors. Addressing issues like algorithm bias, data quality, and the balance between popular content and diverse recommendations poses significant challenges. 2. Algorithm bias is a concern that media professionals are actively working to mitigate. AI algorithms, if not carefully designed and monitored, can unintentionally reinforce biases present in the training data. To combat this, professionals emphasize the importance of diverse and representative training datasets and ongoing algorithm auditing. 3. Ensuring data quality is another hurdle in the world of recommender systems. AI relies heavily on data, and inaccurate or incomplete data can lead to suboptimal recommendations. Media companies are investing in data collection and preprocessing techniques to enhance the quality and relevance of their recommendations. 4. Balancing recommendations between popular content and diverse options is a delicate art. While AI algorithms excel at predicting user preferences based on historical data, they must also introduce viewers to new and unexpected content. Achieving this equilibrium requires constant fine-tuning of algorithms and feedback loops from users. In one of our publications from late last year, we [explain how at Recombee we frequently interpret the environment in which the recommender systems are exposed](https://www.recombee.com/blog/bandit-models-exploiting-popularity-and-curiosity-to-recommend-trending-content) to effectively respond to changes in popularity. 5. Last but not least, having insights into how much and what kind of content is being consumed, what kind of content to push or buy, and suggestions on what to do with that information is what many of industry professionals are missing. So tools providing such insights will become invaluable. ### Bursting Bubbles Filter bubbles, which limit exposure to diverse viewpoints, are a growing concern. Experts suggest strategies for overcoming these bubbles, including diversifying content sources and leveraging AI to broaden perspectives. Diversity in content is critical to combating filter bubbles. Media companies can actively curate and recommend a wide range of content, introducing users to different perspectives, genres, and cultures. AI can play a pivotal role in this process by identifying and promoting content that challenges preexisting biases and preferences. At Recombee, we strive for innovation that brings diversity naturally. Our solution offers a wide variety of customizable [business rules](https://pavelkordik.substack.com/p/explaining-recommender-systems-to) for delivering tailored content selection. With business rules, our customers can directly influence what content is recommended. [Filters](https://docs.recombee.com/reql_filtering_and_boosting#filtering) can be used to select candidates for recommendations, such as showing only movies released in the last 7 days or in a particular language. [Boosters](https://docs.recombee.com/reql_filtering_and_boosting#boosting) help create a bias or preference for a selected criteria, such as boosting videos selected by the editorial team or boosting videos from a specific category. We are also continually developing new features to automate the diversification of recommendations, such as the latest release of [Constraints](https://www.youtube.com/watch?v=Mq2HORnrEIc). This feature enables you to provide a homepage with a diverse range of categories, offering something for every customer from multiple segments. In the case of a video-on-demand platform with various channels, Constraints can ensure that the platform’s recommendations span across different genres like crime, comedy, news, and more, catering to a broader audience. ### The Future of Personalization So how does the trajectory of personalization in the industry look like? Media experts share predictions on how AI-driven personalization will continue to evolve to meet ever-changing consumer expectations. AI algorithms are increasingly leveraging user behavior data, historical interactions, and even biometric signals to curate content that caters to individual tastes. This enhanced personalization is reshaping the media landscape, delivering content that deeply resonates with audiences. These experts also foresee a future where personalization becomes more than just content recommendations. AI will enable platforms to offer tailored pricing models, dynamic content generation, and individualized marketing campaigns that resonate deeply with each user. The line between content creator and consumer will blur as AI empowers users to co-create content, personalizing their media experiences like never before. ## Conclusion The integration of AI into the media industry is not only inevitable but transformative. It empowers media professionals to deliver content that resonates deeply with individual viewers, enhancing user experiences and engagement. While challenges persist, the potential for meaningful personalization that respects user privacy remains within reach. The future of the media industry is undoubtedly personalized, and AI is the driving force propelling us into this new era. As we navigate this landscape, the balance between personalization, data security, and diversity will shape the media industry's future in profound ways. ### Key Takeaways * The rising use of AI poses questions about the future of human involvement in creative roles. * The shift towards personalized content consumption is reshaping the media landscape. * By 2030, AI will play a crucial role in content creation, distribution, and personalization. * AI implementation often requires a cultural shift within media organizations. * Regulations may be necessary to ensure responsible and ethical AI usage in the media industry. * The line between content creator and consumer will blur as AI empowers users to co-create content. Recommendation Engine Personalization ## Next Articles [![](https://www.recombee.com/img/blog/ai-assistants-know-your-preferences-even-better-than-you-do.png)](https://www.recombee.com/blog/ai-assistants-know-your-preferences-even-better-than-you-do) ### [AI Assistants Know Your Preferences, Even Better Than You Do](https://www.recombee.com/blog/ai-assistants-know-your-preferences-even-better-than-you-do) Recommender systems and ethical controversies ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Nov 23, 2023 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/modern-recommender-systems-part-1-introduction.png)](https://www.recombee.com/blog/modern-recommender-systems-part-1-introduction) ### [Modern Recommender Systems - Part 1: Introduction](https://www.recombee.com/blog/modern-recommender-systems-part-1-introduction) How machine learning methods simplify item discovery and search. ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Apr 17, 2023 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/explaining-recommender-systems-to-product-owners.png)](https://pavelkordik.substack.com/p/explaining-recommender-systems-to) ### [Explaining Recommender Systems to Product Owners](https://pavelkordik.substack.com/p/explaining-recommender-systems-to) In my presentation at the Data Technology Seminar organized by the European Broadcasting Union, I have focused on demonstrating that recommender systems can actually help public media organizations to better fulfill their role in society and reduce content distribution biases. ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Mar 27, 2023 Personalization Recommendation Engine --- # Matrix Completion | Recombee AI Recommender > Source: https://www.recombee.com/blog/inductive-matrix-completion-how-to-improve-recommendations-for-cold-start-users-and-items-by-incorporating-their-attributes > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # Inductive Matrix Completion: How to Improve Recommendations for Cold Start Users and Items by Incorporating Their Attributes ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Mar 20, 2023 Matrix completion (MC), the problem of recovering the missing entries of a partially observed matrix, has found use in a wide range of domains. Still, its potentially most successful application is as a collaborative filtering technique for recommender systems (RSs). MC techniques are an accurate and efficient approach for recovering a high number of unobserved entries from a small number of observed data. To recover missing entries of a matrix, it is first necessary to make an assumption on the structure of the ground truth matrix. The most common assumption is that the matrix is of low rank; that is, the matrix can be factorized in a space smaller than the number of users and items. In recommender systems, we model each user as a row, each item as a column, and each user's rating for an item as an entry in that matrix. However, we typically don't observe all the ratings in the matrix - in fact, we might only have a tiny fraction of them. The low-rank assumption is a way to deal with this sparsity. Essentially, it suggests that there are underlying factors that are shared by groups of users and items - for example, if a bunch of users all rate action movies highly, there might be an underlying "action movie" factor that they all respond to. Similarly, if a bunch of movies all have car chases and explosions, there might be an underlying "action movie" factor that they all share. Therefore, this assumption lets us fill in the missing entries in the matrix more accurately, because we can use the known ratings to infer the underlying factors and then use those factors to make predictions for the missing entries. For the sake of simplicity, we will be using movie recommendations as a running example. Hence the data will consist of user and movie ratings. Depending on the application, MC can deal with type1-type2 pairings, such as user-book, user-product, user-user, product-product, etc. Despite their pervasiveness, theoretical foundations, high accuracy, and efficiency, traditional matrix completion methods unfortunately fail to address the so-called cold-start problem. One of the most fundamental problems in RSs, the cold-start problem consists of profiling movies or users for whom the system has never observed a single interaction. This is a realistic scenario for commercial recommendation systems, where new users and new movies are introduced regularly. To help you understand this, the figure below shows a typical matrix completion method. However, this method would not be able to handle a scenario where there is no prior information or interactions available for new users or movies. ![](https://www.recombee.com/img/blog/inductive-matrix-completion-how-to-improve-recommendations-for-cold-start-users-and-items-by-incorporating-their-attributes/01.svg) This issue arises because standard MC methods are user- and movie-agnostic: users' and movies' embeddings are learned only based on interactions and do not account for any additional information. As the data lacks some user-i (resp item-j) interactions, the method can't build the i-th (resp. j-th) vector of the user's (resp items') embeddings. However, it is not uncommon for Recommender Systems to have access to side information about users (e.g., browser the user uses, device type, etc.) and movies (e.g., director, genres, etc.). To address this issue, one can employ Inductive Matrix Completion (IMC) and recommend movies and users without observed interactions. Unlike traditional matrix completion methods, IMC uses side information matrices to induce the learning process in a collaborative filtering approach. Basically, it learns a proxy combining users and item attributes in a low-rank space. ![](https://www.recombee.com/img/blog/inductive-matrix-completion-how-to-improve-recommendations-for-cold-start-users-and-items-by-incorporating-their-attributes/02.svg) The figure above demonstrates how inductive matrix completion works. Throughout the learning process, the IMC algorithm uses side information from all users and items that so far have interacted with the system. In our toy example, the technique is aware of the users' browser and device type; for the items, we consider that the RS knows if they belong to the action or comedy and the duration of the movie. Then, the method utilizes the side information to learn the matrix in the middle of the figure that serves as a proxy for the side information of users and items. With that learning, IMC can use this matrix to recommend to users that their browser and device type are known but have yet to interact with the system (highlighted in green). Note that IMC can also be used with more complex side information, such as item figures or textual descriptions. In that case, they must be converted to vector representations (e.g., by convolutional neural autoencoders). ## Is Inductive Matrix Completion an Accurate Method? Well... as in any machine learning model, it will highly depend on the side information quality. It's hard to predict someone's favourite movie genre based on their favourite pizza topping. Here is the same: if the side information is related to the interactions, IMC can produce excellent practical results. We clarify that such methods can also be used in a hybrid approach, where traditional matrix factorization is incorporated together with IMC. For reference, see the link. There is a lot of rigorous theoretical analysis behind IMC, and recently, I was part of a team of researchers from five different institutions who investigated a specific model of IMC. Our work was approved for oral presentation at the AAAI 2023 conference, which is a very prestigious machine learning conference. You can read our paper if you're interested in the details of our approach. We showed through the proof of generalisation bounds that IMC can give excellent results when the side information we use is strongly related to the interactions we observe. In other words, if we have good side information that tells us about the relationships between users and items, we can use that to make very accurate predictions about missing entries in the matrix. This category of research is crucial for comprehending how AI approaches mathematical function and for providing statistical data on the quality of outcomes produced by machine learning techniques. IMC is a powerful tool that can help solve the problem of cold-start recommendations, where we have limited information about interactions of new users or items. By using additional side information, IMC can enhance the key performance indicators (KPIs) of recommender systems significantly. IMC is just one of several toolkits offered by Recombee that deliver improved content recommendations in cold-start scenarios. If you found our work interesting and helpful, we would greatly appreciate it if you would cite us in your own research, provide feedback or comments, or get in touch with our research team for more technical details. ## References Ledent, Antoine; Alves, Rodrigo; Lei, Yunwen; Guermeur,Yann;Kloft, Marius. Generalization Bounds for Inductive Matrix Completion in Low-noise Settings. In: Proceedings of the AAAI Conference on Artificial Intelligence. 2023 (To appear) Ledent, Antoine; Alves, Rodrigo; Kloft, Marius. Orthogonal Inductive Matrix Completion. IEEE Transactions on Neural Networks and Learning Systems, 2021 Personalization Recommendation Engine ## Next Articles [![](https://www.recombee.com/img/blog/explaining-recommender-systems-to-product-owners.png)](https://pavelkordik.substack.com/p/explaining-recommender-systems-to) ### [Explaining Recommender Systems to Product Owners](https://pavelkordik.substack.com/p/explaining-recommender-systems-to) In my presentation at the Data Technology Seminar organized by the European Broadcasting Union, I have focused on demonstrating that recommender systems can actually help public media organizations to better fulfill their role in society and reduce content distribution biases. ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Mar 27, 2023 Personalization Recommendation Engine [![](https://www.recombee.com/img/blog/breaking-the-news-the-role-of-ai-in-modern-journalism.png)](https://www.recombee.com/blog/breaking-the-news-the-role-of-ai-in-modern-journalism) ### [Breaking the News: The Role of AI in Modern Journalism](https://www.recombee.com/blog/breaking-the-news-the-role-of-ai-in-modern-journalism) Artificial Intelligence (AI) has rapidly transformed the media industry in recent years. From automated news production to trend analysis and personalized content recommendations, AI has brought significant changes to the way media is created, distributed, and consumed. ![](https://www.recombee.com/img/blog/authors/tanalancova.png) Tana Lancova Mar 14, 2023 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/innovative-personalization-features-for-2023.png)](https://www.recombee.com/blog/innovative-personalization-features-for-2023) ### [Innovative Personalization Features for 2023](https://www.recombee.com/blog/innovative-personalization-features-for-2023) The digital world is changing; users' expectations for personalization are increasing, and our Recombee features are continuously improving. One of our focuses is to support our clients in providing the best possible user experiences... ![](https://www.recombee.com/img/blog/authors/gabrielatakacova.png) Gabriela Takacova Feb 17, 2023 New Features Recommendation Engine --- # Recombee Item Segmentations | Recombee Recommender > Source: https://www.recombee.com/blog/recombee-item-segmentations > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # Recombee Item Segmentations ![](https://www.recombee.com/img/blog/authors/tomasrehorek.png) Tomas Rehorek Jan 11, 2023 ![](https://www.recombee.com/img/blog/recombee-item-segmentations/main.png) Item Segmentations is Recombee's original and elegant solution to various advanced tasks related to hierarchical and relational data. The feature provides a flexible way to group Items (products or pieces of content in your catalog) into Segments based on shared conditions. Items can be segmented by authors, genres, vendors, categories, but also by their combinations and advanced conditions such as year of production or price ranges. Compared to traditional approaches based on so-called item affinities, Recombee Item Segments are separate objects that can: 1. be **Recommended and Searched** in novel ways, 2. be **Flexibly and Dynamically** defined and modified on top of the existing catalog. ## Recommending and Searching Since they are independent entities, Item Segments can be recommended and searched for. _"Top Categories for you", "Similar Artists",_ or _"Searching for a Category",_ are all natural use cases for recommendation and searching Item Segments. Recombee models actively work with the relations between Items and Item Segments, ensuring the information is properly propagated. Interacting with Items from a particular Item Segment yields indirect information about a user's interest in a particular Segment. ## Flexibility Item Segmentations are defined using the existing data in the Items catalog. Therefore, no additional implementation work is typically needed to create the Segmentations in your Recombee Database. Multiple Segmentations can co-exist in the same Recombee database on top of the same Item catalog. And each Segmentation can focus on different aspects and characteristics (e.g. one Segmentation based on category and another based on the vendor of the product). In the simplest case, an Item Segmentation is based on a single property (column) of the Items catalog (e.g. category). However, we also offer more advanced ways for defining the Item Segmentation that make use of the ReQL (Recombee Query Language). This comes in handy e.g. in the case of the personalized re-ordering of the rows on the homepage (each row can be defined by a ReQL expression that describes items that are available within the row). Item Segments automatically appear and disappear on the fly along with changes in the Item catalog. Additionally, a single Item may (or may not) be a part of multiple Segments within the same Segmentation (e.g. movie with multiple genres). ## How Does This Look in Practice? Let's demonstrate the power of this feature with a real-life example on a VoD platform. Consider the following data about the Items. ![](https://www.recombee.com/img/blog/recombee-item-segmentations/01.png) You can easily create a Segmentation based on the genres property in the Recombee Admin UI. ![](https://www.recombee.com/img/blog/recombee-item-segmentations/02.png) You will get a preview of the Segments yielded by the created Segmentation. Now you can use the _Recommend Item Segments To User_ API Endpoint to recommend the top genres for each user! ![](https://www.recombee.com/img/blog/recombee-item-segmentations/03.png) Do you want to know more? [**Explore Item Segmentations Docs**](https://docs.recombee.com/segmentations) ## Let's Connect Are you keen on trying Recombee Item Segmentations? We are happy to provide integration support at [support@recombee.com](mailto:support@recombee.com) or assist with any general inquiries at [business@recombee.com](mailto:business@recombee.com). If you just want to learn more about how personalization can be applied to your use case, get inspired in our [Case Study](https://www.recombee.com/case-studies) section, and explore the application of AI recommendations in various domains. New Features Recommendation Engine ## Next Articles [![](https://www.recombee.com/img/blog/innovative-personalization-features-for-2023.png)](https://www.recombee.com/blog/innovative-personalization-features-for-2023) ### [Innovative Personalization Features for 2023](https://www.recombee.com/blog/innovative-personalization-features-for-2023) The digital world is changing; users' expectations for personalization are increasing, and our Recombee features are continuously improving. One of our focuses is to support our clients in providing the best possible user experiences... ![](https://www.recombee.com/img/blog/authors/gabrielatakacova.png) Gabriela Takacova Feb 17, 2023 New Features Recommendation Engine [![](https://www.recombee.com/img/blog/bandit-models-exploiting-popularity-and-curiosity-to-recommend-trending-content.png)](https://www.recombee.com/blog/bandit-models-exploiting-popularity-and-curiosity-to-recommend-trending-content) ### [Bandit Models: Exploiting Popularity and Curiosity to Recommend Trending Content](https://www.recombee.com/blog/bandit-models-exploiting-popularity-and-curiosity-to-recommend-trending-content) Humans are inherently curious. In fact, curiosity is linked to the evolution of humankind. For instance, according to famous historian Yuval Noah Harari in his bestseller book "Sapiens", our language skills evolved as a way of gossiping... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Oct 16, 2022 Recommendation Engine [![](https://www.recombee.com/img/blog/keeping-up-with-digital-media-convergence.png)](https://www.recombee.com/blog/keeping-up-with-digital-media-convergence) ### [Keeping Up With Digital Media Convergence](https://www.recombee.com/blog/keeping-up-with-digital-media-convergence) At Recombee, we felt the transition within the media industry accelerated by the pandemic. OTT and CTV consumption ballooned at a significant rate. ![](https://www.recombee.com/img/blog/authors/annadolezelova.png) Anna Dolezelova Oct 8, 2022 Personalization --- # Recommendation Engine | Blog > Source: https://www.recombee.com/blog/recommendation-engine > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # Blog ## Recommendation Engine [All](https://www.recombee.com/blog) [Recommendation Engine](https://www.recombee.com/blog/recommendation-engine) [Personalization](https://www.recombee.com/blog/personalization) [Integrations](https://www.recombee.com/blog/integrations) [New Features](https://www.recombee.com/blog/new-features) [Partnerships](https://www.recombee.com/blog/partnerships) [Company News](https://www.recombee.com/blog/company-news) [![](https://www.recombee.com/img/blog/new-feature-ab-testing.png)](https://www.recombee.com/blog/new-feature-ab-testing) ### [New Feature: A/B Testing](https://www.recombee.com/blog/new-feature-ab-testing) Personalization is never finished. Every change to your recommendation strategy raises new questions. Should you boost fresh content? Change used Logic? Introduce new filters? ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Aug 14, 2026 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/mid-year-roundup-2026-so-far.png)](https://www.recombee.com/blog/mid-year-roundup-2026-so-far) ### [Mid-Year Roundup: 2026 So Far](https://www.recombee.com/blog/mid-year-roundup-2026-so-far) The digital world continues to change at an incredible pace. As technology evolves and user expectations keep rising, creating experiences that feel relevant has never been more important. ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Jul 20, 2026 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/a-2025-research-retrospective.png)](https://www.recombee.com/blog/a-2025-research-retrospective) ### [A 2025 Research Retrospective](https://www.recombee.com/blog/a-2025-research-retrospective) In 2025, I became Head of Research at Recombee. I knew it would be a challenge: and I wasn’t wrong. Recombee is a leader in recommendation-as-a-service, with thousands of clients... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Jan 29, 2026 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/looking-back-at-2025.png)](https://www.recombee.com/blog/looking-back-at-2025) ### [Looking Back at 2025](https://www.recombee.com/blog/looking-back-at-2025) 2025 marked 10 years of Recombee. A decade of building personalization from first principles, shaped by research, real-world deployments, and close collaboration with partners across industries. ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Jan 20, 2026 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/product-highlights-from-2025.png)](https://www.recombee.com/blog/product-highlights-from-2025) ### [Product Highlights from 2025](https://www.recombee.com/blog/product-highlights-from-2025) In 2025, we focused on making advanced personalization easier to implement, scale, and maintain across products and platforms. Here’s a look at the key product updates we released last year. ![](https://www.recombee.com/img/blog/authors/ondrejfiedler.png) Ondrej Fiedler Jan 08, 2026 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/ai-news-and-outlook-for-2026.png)](https://www.recombee.com/blog/ai-news-and-outlook-for-2026) ### [AI News and Outlook for 2026](https://www.recombee.com/blog/ai-news-and-outlook-for-2026) Here’s what caught my attention in AI research lately, and where things might be heading in 2026\. After 25+ years in this field, the pace has gotten hard to keep up with. I’m trying to make sense of... ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Dec 17, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/modern-recommender-systems-part-3-objectives.png)](https://www.recombee.com/blog/modern-recommender-systems-part-3-objectives) ### [Modern Recommender Systems - Part 3: Objectives](https://www.recombee.com/blog/modern-recommender-systems-part-3-objectives) Learning objectives of recommender systems and personalized search. ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Sep 03, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/the-building-blocks-of-privacy-friendly-personalization.png)](https://www.recombee.com/blog/the-building-blocks-of-privacy-friendly-personalization) ### [The Building Blocks of Privacy-Friendly Personalization](https://www.recombee.com/blog/the-building-blocks-of-privacy-friendly-personalization) Personalization can be achieved without compromising user privacy. While many personalization systems have historically relied on practices now considered intrusive, like... ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Aug 07, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/no-code-search-widget-personalized-powerful-effortless.png)](https://www.recombee.com/blog/no-code-search-widget-personalized-powerful-effortless) ### [No-Code Search Widget: Personalized, Powerful, Effortless](https://www.recombee.com/blog/no-code-search-widget-personalized-powerful-effortless) At Recombee, we don't just excel at recommendations – we provide powerful full-text search capabilities too. Our Quick, No-Code Search Widget exemplifies this, offering a seamless... ![](https://www.recombee.com/img/blog/authors/ondrejfiedler.png) Ondrej Fiedler Jul 25, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/how-regionalization-based-recommendations-can-improve-your-operations.png)](https://www.recombee.com/blog/how-regionalization-based-recommendations-can-improve-your-operations) ### [How Regionalization-Based Recommendations Can Improve Your Operations](https://www.recombee.com/blog/how-regionalization-based-recommendations-can-improve-your-operations) From ancient trade routes to modern urban planning, geography has consistently shaped human decisions and opportunities. Today, in the world of online business and personalized recommendations... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Jul 18, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/shield-the-universal-framework-making-ai-search-safer-for-everyone.png)](https://www.recombee.com/blog/shield-the-universal-framework-making-ai-search-safer-for-everyone) ### [SHIELD: The Universal Framework Making AI Search Safer for Everyone](https://www.recombee.com/blog/shield-the-universal-framework-making-ai-search-safer-for-everyone) Imagine searching for "glass tubing" and getting recommendations for drug manufacturing equipment. As AI-powered search becomes ubiquitous — from online marketplaces to social networks... ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Jun 24, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/making-recommendations-fairer-a-new-way-to-guarantee-exposure-for-all.png)](https://www.recombee.com/blog/making-recommendations-fairer-a-new-way-to-guarantee-exposure-for-all) ### [Making Recommendations Fairer: A New Way to Guarantee Exposure for All](https://www.recombee.com/blog/making-recommendations-fairer-a-new-way-to-guarantee-exposure-for-all) As recommender systems become more widespread across digital platforms, concerns around fairness are coming to the forefront. Standard relevance-based ranking techniques, while effective... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Apr 29, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/2025-sneak-peek.png)](https://www.recombee.com/blog/2025-sneak-peek) ### [2025 Sneak Peek](https://www.recombee.com/blog/2025-sneak-peek) This year is already off to an exciting start, and we’re rolling out new tools to improve efficiency and optimize recommendations. Here’s what’s available and what’s coming next. ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Mar 19, 2025 Recommendation Engine New Features [![](https://www.recombee.com/img/blog/build-vs-buy-deciding-the-best-approach-for-your-recommender-system.png)](https://www.recombee.com/blog/build-vs-buy-deciding-the-best-approach-for-your-recommender-system) ### [Build vs. Buy: Deciding the Best Approach for Your Recommender System](https://www.recombee.com/blog/build-vs-buy-deciding-the-best-approach-for-your-recommender-system) When it comes to deciding between buying a recommender system and building one from scratch, the choice isn’t always straightforward. Both options come with their own set of pros and cons... ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova & Ondrej Fiedler Mar 14, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/are-you-here-to-stay-unraveling-the-dynamics-of-stable-and-curious-audiences-in-web-systems.png)](https://www.recombee.com/blog/are-you-here-to-stay-unraveling-the-dynamics-of-stable-and-curious-audiences-in-web-systems) ### [Are You Here to Stay? Unraveling the Dynamics of Stable and Curious Audiences in Web Systems](https://www.recombee.com/blog/are-you-here-to-stay-unraveling-the-dynamics-of-stable-and-curious-audiences-in-web-systems) Why do influencers frequently request their subscribers to enable all notifications for their channels? This practice stems from their awareness that not all subscribers are regular... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Mar 03, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/recombee-research-2024.png)](https://www.recombee.com/blog/recombee-research-2024) ### [Recombee Research 2024](https://www.recombee.com/blog/recombee-research-2024) Recombee has always been deeply connected to academia, with four of our six co-founders holding Ph.D. degrees. Over the years, our investment in research has grown alongside our company and the increasing demand for advanced features in the market. ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Feb 23, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/introducing-beeformer-a-framework-for-training-foundational-models-for-recommender-systems.png)](https://www.recombee.com/blog/introducing-beeformer-a-framework-for-training-foundational-models-for-recommender-systems) ### [Introducing beeFormer: A Framework for Training Foundational Models for Recommender Systems](https://www.recombee.com/blog/introducing-beeformer-a-framework-for-training-foundational-models-for-recommender-systems) In the fast-evolving world of recommender systems, understanding both how users interact with content and the actual content itself is crucial. Many existing recommender systems struggle to balance these two aspects... ![](https://www.recombee.com/img/blog/authors/vojtechvancura.png) Vojtech Vancura Oct 15, 2024 New Features Recommendation Engine [![](https://www.recombee.com/img/blog/insights-the-next-level-of-analytics-in-recombee-ui.png)](https://www.recombee.com/blog/insights-the-next-level-of-analytics-in-recombee-ui) ### [Insights: The Next Level of Analytics in Recombee UI](https://www.recombee.com/blog/insights-the-next-level-of-analytics-in-recombee-ui) Insights, the analytics section of our Admin UI, offers various predefined and fully customizable reports to track recommended items and how users interact with these recommendations. ![](https://www.recombee.com/img/blog/authors/ondrejfiedler.png) Ondrej Fiedler May 09, 2024 New Features Recommendation Engine [![](https://www.recombee.com/img/blog/elevate-your-personalization-strategy-with-recombees-innovative-features.png)](https://www.recombee.com/blog/elevate-your-personalization-strategy-with-recombees-innovative-features) ### [Elevate Your Personalization Strategy with Recombee's Innovative Features](https://www.recombee.com/blog/elevate-your-personalization-strategy-with-recombees-innovative-features) The digital landscape and customer preferences and behavior are changing faster than ever now. To help our clients stay on top of the game, our team has focused on developing innovative features... ![](https://www.recombee.com/img/blog/authors/janvaluch.png) Jan Valuch Mar 13, 2024 New Features Recommendation Engine [![](https://www.recombee.com/img/blog/modern-recommender-systems-part-2-data.png)](https://www.recombee.com/blog/modern-recommender-systems-part-2-data) ### [Modern Recommender Systems - Part 2: Data](https://www.recombee.com/blog/modern-recommender-systems-part-2-data) Data used by modern recommenders and how we can measure progress towards goals. ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Mar 07, 2024 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/is-this-comment-useful-enhancing-personalized-recommendations-by-considering-user-rating-uncertainty.png)](https://www.recombee.com/blog/is-this-comment-useful-enhancing-personalized-recommendations-by-considering-user-rating-uncertainty) ### [Is This Comment Useful? Enhancing Personalized Recommendations by Considering User Rating Uncertainty](https://www.recombee.com/blog/is-this-comment-useful-enhancing-personalized-recommendations-by-considering-user-rating-uncertainty) Picture this: you're on the hunt for the perfect new smartphone, browsing through your favourite online electronics store. The online store’s recommendation engine pops up with what it thinks could be your possible next gadget love... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Mar 01, 2024 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/recombeelabs-2023-research-publications.png)](https://www.recombee.com/blog/recombeelabs-2023-research-publications) ### [Recombeelab's 2023 Research Publications](https://www.recombee.com/blog/recombeelabs-2023-research-publications) Recombeelab, a joint research laboratory of Recombee and the Faculty of Information Technology at the Czech Technical University in Prague, experienced a highly productive year in 2023, publishing a series of insightful and impactful papers in the field of recommendation systems. ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Jan 19, 2024 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/ai-news-and-outlook-for-2024.png)](https://www.recombee.com/blog/ai-news-and-outlook-for-2024) ### [AI News and Outlook for 2024](https://www.recombee.com/blog/ai-news-and-outlook-for-2024) We look at the most interesting research directions and assess the state of knowledge in key areas of AI. We'll also estimate future developments in 2024 so you know what to prepare for. ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Jan 16, 2024 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/ai-assistants-know-your-preferences-even-better-than-you-do.png)](https://www.recombee.com/blog/ai-assistants-know-your-preferences-even-better-than-you-do) ### [AI Assistants Know Your Preferences, Even Better Than You Do](https://www.recombee.com/blog/ai-assistants-know-your-preferences-even-better-than-you-do) Recommender systems and ethical controversies ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Nov 23, 2023 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/the-ai-revolution-in-the-media-industry.png)](https://www.recombee.com/blog/the-ai-revolution-in-the-media-industry) ### [The AI (R)Evolution in the Media Industry](https://www.recombee.com/blog/the-ai-revolution-in-the-media-industry) In today's digital age, personalization has become the cornerstone of the media industry. Whether it's tailoring content recommendations, refining marketing strategies, or enhancing user experiences... ![](https://www.recombee.com/img/blog/authors/annapetr.png) Anna Dolezelova & Petr Popov Oct 23, 2023 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/modern-recommender-systems-part-1-introduction.png)](https://www.recombee.com/blog/modern-recommender-systems-part-1-introduction) ### [Modern Recommender Systems - Part 1: Introduction](https://www.recombee.com/blog/modern-recommender-systems-part-1-introduction) How machine learning methods simplify item discovery and search. ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Apr 17, 2023 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/explaining-recommender-systems-to-product-owners.png)](https://pavelkordik.substack.com/p/explaining-recommender-systems-to) ### [Explaining Recommender Systems to Product Owners](https://pavelkordik.substack.com/p/explaining-recommender-systems-to) In my presentation at the Data Technology Seminar organized by the European Broadcasting Union, I have focused on demonstrating that recommender systems can actually help public media organizations to better fulfill their role in society and reduce content distribution biases. ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Mar 27, 2023 Personalization Recommendation Engine [![](https://www.recombee.com/img/blog/inductive-matrix-completion-how-to-improve-recommendations-for-cold-start-users-and-items-by-incorporating-their-attributes.png)](https://www.recombee.com/blog/inductive-matrix-completion-how-to-improve-recommendations-for-cold-start-users-and-items-by-incorporating-their-attributes) ### [Inductive Matrix Completion: How to Improve Recommendations for Cold Start Users and Items by Incorporating Their Attributes](https://www.recombee.com/blog/inductive-matrix-completion-how-to-improve-recommendations-for-cold-start-users-and-items-by-incorporating-their-attributes) Matrix completion (MC), the problem of recovering the missing entries of a partially observed matrix, has found use in a wide range of domains. Still, its potentially most successful application is as a collaborative filtering technique for recommender systems (RSs)... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Mar 20, 2023 Personalization Recommendation Engine [![](https://www.recombee.com/img/blog/breaking-the-news-the-role-of-ai-in-modern-journalism.png)](https://www.recombee.com/blog/breaking-the-news-the-role-of-ai-in-modern-journalism) ### [Breaking the News: The Role of AI in Modern Journalism](https://www.recombee.com/blog/breaking-the-news-the-role-of-ai-in-modern-journalism) Artificial Intelligence (AI) has rapidly transformed the media industry in recent years. From automated news production to trend analysis and personalized content recommendations, AI has brought significant changes to the way media is created, distributed, and consumed. ![](https://www.recombee.com/img/blog/authors/tanalancova.png) Tana Lancova Mar 14, 2023 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/innovative-personalization-features-for-2023.png)](https://www.recombee.com/blog/innovative-personalization-features-for-2023) ### [Innovative Personalization Features for 2023](https://www.recombee.com/blog/innovative-personalization-features-for-2023) The digital world is changing; users' expectations for personalization are increasing, and our Recombee features are continuously improving. One of our focuses is to support our clients in providing the best possible user experiences... ![](https://www.recombee.com/img/blog/authors/gabrielatakacova.png) Gabriela Takacova Feb 17, 2023 New Features Recommendation Engine [![](https://www.recombee.com/img/blog/recombee-item-segmentations.png)](https://www.recombee.com/blog/recombee-item-segmentations) ### [Recombee Item Segmentations](https://www.recombee.com/blog/recombee-item-segmentations) Item Segmentations are Recombee's original and elegant solution to various advanced tasks related to hierarchical and relational data. The feature provides a flexible way to group items (products or pieces of content) into segments... ![](https://www.recombee.com/img/blog/authors/tomasrehorek.png) Tomas Rehorek Jan 11, 2023 New Features Recommendation Engine [![](https://www.recombee.com/img/blog/bandit-models-exploiting-popularity-and-curiosity-to-recommend-trending-content.png)](https://www.recombee.com/blog/bandit-models-exploiting-popularity-and-curiosity-to-recommend-trending-content) ### [Bandit Models: Exploiting Popularity and Curiosity to Recommend Trending Content](https://www.recombee.com/blog/bandit-models-exploiting-popularity-and-curiosity-to-recommend-trending-content) Humans are inherently curious. In fact, curiosity is linked to the evolution of humankind. For instance, according to famous historian Yuval Noah Harari in his bestseller book "Sapiens", our language skills evolved as a way of gossiping... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Oct 16, 2022 Recommendation Engine [![](https://www.recombee.com/img/blog/repsys-opensource-library-for-interactive-evaluation-of-recommendation-systems.png)](https://www.recombee.com/blog/repsys-opensource-library-for-interactive-evaluation-of-recommendation-systems) ### [Visual and Interactive Evaluation of Recommender Systems](https://www.recombee.com/blog/repsys-opensource-library-for-interactive-evaluation-of-recommendation-systems) When building modern real-world artificial intelligence systems, it is increasingly important to validate that the system works correctly. This is however not an easy task. Existing tools for machine learning practitioners... ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Sep 19, 2022 Recommendation Engine [![](https://www.recombee.com/img/blog/making-linear-autoencoders-work-for-large-scale-recommendation-systems.png)](https://www.recombee.com/blog/making-linear-autoencoders-work-for-large-scale-recommendation-systems) ### [Making Linear Autoencoders Work for Large Scale Recommendation Systems](https://www.recombee.com/blog/making-linear-autoencoders-work-for-large-scale-recommendation-systems) Linear autoencoders for collaborative filtering in recommender systems are simple and surprisingly accurate as we explained in our blogpost on how linear methods work. The critical disadvantage of methods like EASE is that they are not applicable to real-world problems... ![](https://www.recombee.com/img/blog/authors/vojtechvancura.png) Vojtech Vancura Aug 29, 2022 Recommendation Engine [![](https://www.recombee.com/img/blog/linear-methods-and-autoencoders-in-recommender-systems.png)](https://www.recombee.com/blog/linear-methods-and-autoencoders-in-recommender-systems) ### [Linear Methods and Autoencoders in Recommender Systems](https://www.recombee.com/blog/linear-methods-and-autoencoders-in-recommender-systems) Linear regression is probably the simplest and surprisingly efficient machine learning method. It should be the method of your first choice, according to the famous KISS principle. Also, it often works better than sophisticated methods, because it is... ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Sep 7, 2021 Recommendation Engine [![](https://www.recombee.com/img/blog/recombee-xperience-kentico.png)](https://xperience.io/discover/blog/recombee-kentico-xperience-1-on-1-personalization) ### [Recombee and Kentico Xperience: Guide to One-On-One Personalization](https://xperience.io/discover/blog/recombee-kentico-xperience-1-on-1-personalization) Recombee expanded its integration options - and now is available at the Kentico Xperience platform! Analyzing different types of personalization, we look into why Kentiko chose our AI-powered recommendation engine over manual segmentation. ![](https://www.recombee.com/img/blog/authors/gabrielatakacova.png) Gabriela Takacova May 6, 2021 Personalization Recommendation Engine [![](https://www.recombee.com/img/blog/23d22w82u3d52as.png)](https://medium.com/recombee-blog/deep-learning-for-recommender-systems-next-basket-prediction-and-sequential-product-recommendation-796228b34dee) ### [Deep Learning for Recommender Systems: Next Basket Prediction and Sequential Product Recommendation](https://medium.com/recombee-blog/deep-learning-for-recommender-systems-next-basket-prediction-and-sequential-product-recommendation-796228b34dee) Accurate “next basket prediction” will be enabling next generation e-commerce — predictive shopping and logistics. In this blogpost, we will discuss the deep learning technology behind next basket... ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Nov 10, 2020 Personalization Recommendation Engine [![](https://cdn-images-1.medium.com/max/1000/0*IhtrJIi4PvdZ2SAz)](https://medium.com/recombee-blog/check-out-our-new-client-side-integration-support-and-deploy-personalized-recommendations-faster-7dd7bf5b6241) ### [Check out Our New Client-Side Integration Support and Deploy Personalized Recommendations Faster](https://medium.com/recombee-blog/check-out-our-new-client-side-integration-support-and-deploy-personalized-recommendations-faster-7dd7bf5b6241) We knew we had to bring something new to the table, when participating as a Beta Startup at Web Summit, the largest technology conference… ![](https://www.recombee.com/img/blog/authors/ondrejfiedler.png) Ondrej Fiedler Jan 7, 2019 Recommendation Engine [![](https://cdn-images-1.medium.com/max/1000/0*To3_sKOFKyDE-v0k)](https://medium.com/recombee-blog/machine-learning-for-recommender-systems-part-2-deep-recommendation-sequence-prediction-automl-f134bc79d66b) ### [Machine Learning for Recommender Systems — Part 2 (Deep Recommendation, Sequence Prediction, AutoML…](https://medium.com/recombee-blog/machine-learning-for-recommender-systems-part-2-deep-recommendation-sequence-prediction-automl-f134bc79d66b) In the first part of our talk, we discussed basic algorithms, their evaluation and cold start problem. Below we show how deep learning… ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Jun 7, 2018 Recommendation Engine [![](https://cdn-images-1.medium.com/max/1000/0*0we0HYG_xQOlAvc5)](https://medium.com/recombee-blog/machine-learning-for-recommender-systems-part-1-algorithms-evaluation-and-cold-start-6f696683d0ed) ### [Machine Learning for Recommender Systems — Part 1 (Algorithms, Evaluation and Cold Start)](https://medium.com/recombee-blog/machine-learning-for-recommender-systems-part-1-algorithms-evaluation-and-cold-start-6f696683d0ed) Recommender systems are one of the most successful and widespread application of machine learning technologies in business. There were many… ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Jun 3, 2018 Recommendation Engine [![](https://cdn-images-1.medium.com/max/1000/1*hyT8WbrDzFNuyMcS4ppB4A.png)](https://medium.com/recombee-blog/migrating-to-recombee-from-microsoft-cognitive-services-recommendations-9e0c0a9b34a5) ### [Migrating to Recombee From Microsoft Cognitive Services Recommendations](https://medium.com/recombee-blog/migrating-to-recombee-from-microsoft-cognitive-services-recommendations-9e0c0a9b34a5) Microsoft has recently discontinued the Recommendations within the Azure Cognitive Services (MCSR). If you used this service, you are… ![](https://www.recombee.com/img/blog/authors/ondrejfiedler.png) Ondrej Fiedler Mar 22, 2018 Recommendation Engine [![](https://cdn-images-1.medium.com/max/1000/1*C1FQaVx1ztTq-y0H0YsQ0g.png)](https://medium.com/recombee-blog/personalized-push-notifications-enabled-by-artificial-intelligence-8ac057bc97ba) ### [Personalized Push Notifications Enabled by Artificial Intelligence](https://medium.com/recombee-blog/personalized-push-notifications-enabled-by-artificial-intelligence-8ac057bc97ba) Recent progress in artificial intelligence enables us to design proactive AI systems. Whereas traditional recommender systems produce… ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Mar 4, 2018 Recommendation Engine [![](https://cdn-images-1.medium.com/max/1000/1*d8rJ5EWZTOgfEp-lKYBJeA.jpeg)](https://medium.com/recombee-blog/evaluating-recommender-systems-choosing-the-best-one-for-your-business-c688ab781a35) ### [Evaluating Recommender Systems: Choosing the Best One for Your Business](https://medium.com/recombee-blog/evaluating-recommender-systems-choosing-the-best-one-for-your-business-c688ab781a35) Together with the endless expansion of E-commerce and online media in the last years, there are more and more Software-as-a-Service (SaaS)… ![](https://www.recombee.com/img/blog/authors/tomasrehorek.png) Tomas Rehorek Dec 18, 2016 Personalization Recommendation Engine [![](https://cdn-images-1.medium.com/max/1000/1*iKboSkp-zHi8ZNH8ChtP1w.png)](https://medium.com/recombee-blog/the-value-of-personalized-recommendations-for-your-business-6b2e81ce0a4d) ### [The Value of Personalized Recommendations for Your Business](https://medium.com/recombee-blog/the-value-of-personalized-recommendations-for-your-business-6b2e81ce0a4d) The e-commerce boom makes online environment more competitive. Internet retailers seek competitive advantages and a personalized experience… ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Aug 23, 2016 Personalization Recommendation Engine [![](https://miro.medium.com/max/1400/1*NARxz9O7ZX3Un_HMacPcWw.png)](https://medium.com/recombee-blog/recommender-systems-explained-d98e8221f468) ### [Recommender Systems Explained](https://medium.com/recombee-blog/recommender-systems-explained-d98e8221f468) In this article, I overview broad area of recommender systems, explain how individual algorithms work. ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Jul 12, 2016 Recommendation Engine [![](https://miro.medium.com/max/266/1*HUs4oA40zHOJkEo4ihLAHA.png)](https://medium.com/recombee-blog/personalized-recommendations-in-ruby-e3fcaa5de6be) ### [Personalized Recommendations in Ruby](https://medium.com/recombee-blog/personalized-recommendations-in-ruby-e3fcaa5de6be) For those of you, who develop in Ruby, we prepared a simple client application enabling you to benefit from our personalized recommendations. ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik May 16, 2016 Recommendation Engine [![](https://miro.medium.com/max/1400/1*NARxz9O7ZX3Un_HMacPcWw.png)](https://medium.com/recombee-blog/artificial-intelligence-in-the-cloud-310da5e0325e) ### [Artificial Intelligence in the Cloud](https://medium.com/recombee-blog/artificial-intelligence-in-the-cloud-310da5e0325e) At Recombee, we “think big”, and prefer making big leaps in technology over taking small steps. Our team has been involved in data science and artificial intelligence research for many years. Beginning in 2012, we began to capitalize our knowledge and experience, developing products which… ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Mar 30, 2016 Personalization Recommendation Engine --- # Blog > Source: https://www.recombee.com/blog > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # Blog ## We Develop Global Recommendation Service and Share Our Insights Here [All](https://www.recombee.com/blog) [Recommendation Engine](https://www.recombee.com/blog/recommendation-engine) [Personalization](https://www.recombee.com/blog/personalization) [Integrations](https://www.recombee.com/blog/integrations) [New Features](https://www.recombee.com/blog/new-features) [Partnerships](https://www.recombee.com/blog/partnerships) [Company News](https://www.recombee.com/blog/company-news) [![](https://www.recombee.com/img/blog/new-feature-ab-testing.png)](https://www.recombee.com/blog/new-feature-ab-testing) ### [New Feature: A/B Testing](https://www.recombee.com/blog/new-feature-ab-testing) Personalization is never finished. Every change to your recommendation strategy raises new questions. Should you boost fresh content? Change used Logic? Introduce new filters? ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Aug 14, 2026 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/mid-year-roundup-2026-so-far.png)](https://www.recombee.com/blog/mid-year-roundup-2026-so-far) ### [Mid-Year Roundup: 2026 So Far](https://www.recombee.com/blog/mid-year-roundup-2026-so-far) The digital world continues to change at an incredible pace. As technology evolves and user expectations keep rising, creating experiences that feel relevant has never been more important. ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Jul 20, 2026 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/a-2025-research-retrospective.png)](https://www.recombee.com/blog/a-2025-research-retrospective) ### [A 2025 Research Retrospective](https://www.recombee.com/blog/a-2025-research-retrospective) In 2025, I became Head of Research at Recombee. I knew it would be a challenge: and I wasn’t wrong. Recombee is a leader in recommendation-as-a-service, with thousands of clients... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Jan 29, 2026 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/looking-back-at-2025.png)](https://www.recombee.com/blog/looking-back-at-2025) ### [Looking Back at 2025](https://www.recombee.com/blog/looking-back-at-2025) 2025 marked 10 years of Recombee. A decade of building personalization from first principles, shaped by research, real-world deployments, and close collaboration with partners across industries. ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Jan 20, 2026 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/product-highlights-from-2025.png)](https://www.recombee.com/blog/product-highlights-from-2025) ### [Product Highlights from 2025](https://www.recombee.com/blog/product-highlights-from-2025) In 2025, we focused on making advanced personalization easier to implement, scale, and maintain across products and platforms. Here’s a look at the key product updates we released last year. ![](https://www.recombee.com/img/blog/authors/ondrejfiedler.png) Ondrej Fiedler Jan 08, 2026 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/ai-news-and-outlook-for-2026.png)](https://www.recombee.com/blog/ai-news-and-outlook-for-2026) ### [AI News and Outlook for 2026](https://www.recombee.com/blog/ai-news-and-outlook-for-2026) Here’s what caught my attention in AI research lately, and where things might be heading in 2026\. After 25+ years in this field, the pace has gotten hard to keep up with. I’m trying to make sense of... ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Dec 17, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/recombee-partners-with-the-telegraph-to-deliver-ai-driven-personalisation-to-millions-of-readers.png)](https://www.recombee.com/blog/recombee-partners-with-the-telegraph-to-deliver-ai-driven-personalisation-to-millions-of-readers) ### [Recombee Partners with The Telegraph to Deliver AI-Driven Personalisation to Millions of Readers](https://www.recombee.com/blog/recombee-partners-with-the-telegraph-to-deliver-ai-driven-personalisation-to-millions-of-readers) Prague, 11th September 2025 – Recombee, a leading AI-powered recommendation platform, has announced a strategic partnership with The Telegraph... ![](https://www.recombee.com/img/blog/authors/recombee.png) Recombee Oct 02, 2025 Company News [![](https://www.recombee.com/img/blog/modern-recommender-systems-part-3-objectives.png)](https://www.recombee.com/blog/modern-recommender-systems-part-3-objectives) ### [Modern Recommender Systems - Part 3: Objectives](https://www.recombee.com/blog/modern-recommender-systems-part-3-objectives) Learning objectives of recommender systems and personalized search. ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Sep 03, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/the-building-blocks-of-privacy-friendly-personalization.png)](https://www.recombee.com/blog/the-building-blocks-of-privacy-friendly-personalization) ### [The Building Blocks of Privacy-Friendly Personalization](https://www.recombee.com/blog/the-building-blocks-of-privacy-friendly-personalization) Personalization can be achieved without compromising user privacy. While many personalization systems have historically relied on practices now considered intrusive, like... ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Aug 07, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/no-code-search-widget-personalized-powerful-effortless.png)](https://www.recombee.com/blog/no-code-search-widget-personalized-powerful-effortless) ### [No-Code Search Widget: Personalized, Powerful, Effortless](https://www.recombee.com/blog/no-code-search-widget-personalized-powerful-effortless) At Recombee, we don't just excel at recommendations – we provide powerful full-text search capabilities too. Our Quick, No-Code Search Widget exemplifies this, offering a seamless... ![](https://www.recombee.com/img/blog/authors/ondrejfiedler.png) Ondrej Fiedler Jul 25, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/how-regionalization-based-recommendations-can-improve-your-operations.png)](https://www.recombee.com/blog/how-regionalization-based-recommendations-can-improve-your-operations) ### [How Regionalization-Based Recommendations Can Improve Your Operations](https://www.recombee.com/blog/how-regionalization-based-recommendations-can-improve-your-operations) From ancient trade routes to modern urban planning, geography has consistently shaped human decisions and opportunities. Today, in the world of online business and personalized recommendations... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Jul 18, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/shield-the-universal-framework-making-ai-search-safer-for-everyone.png)](https://www.recombee.com/blog/shield-the-universal-framework-making-ai-search-safer-for-everyone) ### [SHIELD: The Universal Framework Making AI Search Safer for Everyone](https://www.recombee.com/blog/shield-the-universal-framework-making-ai-search-safer-for-everyone) Imagine searching for "glass tubing" and getting recommendations for drug manufacturing equipment. As AI-powered search becomes ubiquitous — from online marketplaces to social networks... ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Jun 24, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/making-recommendations-fairer-a-new-way-to-guarantee-exposure-for-all.png)](https://www.recombee.com/blog/making-recommendations-fairer-a-new-way-to-guarantee-exposure-for-all) ### [Making Recommendations Fairer: A New Way to Guarantee Exposure for All](https://www.recombee.com/blog/making-recommendations-fairer-a-new-way-to-guarantee-exposure-for-all) As recommender systems become more widespread across digital platforms, concerns around fairness are coming to the forefront. Standard relevance-based ranking techniques, while effective... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Apr 29, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/2025-sneak-peek.png)](https://www.recombee.com/blog/2025-sneak-peek) ### [2025 Sneak Peek](https://www.recombee.com/blog/2025-sneak-peek) This year is already off to an exciting start, and we’re rolling out new tools to improve efficiency and optimize recommendations. Here’s what’s available and what’s coming next. ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Mar 19, 2025 Recommendation Engine New Features [![](https://www.recombee.com/img/blog/build-vs-buy-deciding-the-best-approach-for-your-recommender-system.png)](https://www.recombee.com/blog/build-vs-buy-deciding-the-best-approach-for-your-recommender-system) ### [Build vs. Buy: Deciding the Best Approach for Your Recommender System](https://www.recombee.com/blog/build-vs-buy-deciding-the-best-approach-for-your-recommender-system) When it comes to deciding between buying a recommender system and building one from scratch, the choice isn’t always straightforward. Both options come with their own set of pros and cons... ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova & Ondrej Fiedler Mar 14, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/are-you-here-to-stay-unraveling-the-dynamics-of-stable-and-curious-audiences-in-web-systems.png)](https://www.recombee.com/blog/are-you-here-to-stay-unraveling-the-dynamics-of-stable-and-curious-audiences-in-web-systems) ### [Are You Here to Stay? Unraveling the Dynamics of Stable and Curious Audiences in Web Systems](https://www.recombee.com/blog/are-you-here-to-stay-unraveling-the-dynamics-of-stable-and-curious-audiences-in-web-systems) Why do influencers frequently request their subscribers to enable all notifications for their channels? This practice stems from their awareness that not all subscribers are regular... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Mar 03, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/recombee-research-2024.png)](https://www.recombee.com/blog/recombee-research-2024) ### [Recombee Research 2024](https://www.recombee.com/blog/recombee-research-2024) Recombee has always been deeply connected to academia, with four of our six co-founders holding Ph.D. degrees. Over the years, our investment in research has grown alongside our company and the increasing demand for advanced features in the market. ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Feb 23, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/key-trends-in-news-and-media-for-2025-and-the-role-of-ai-powered-solutions.png)](https://www.recombee.com/blog/key-trends-in-news-and-media-for-2025-and-the-role-of-ai-powered-solutions) ### [Key Trends in News & Media for 2025 and the Role of AI-Powered Solutions](https://www.recombee.com/blog/key-trends-in-news-and-media-for-2025-and-the-role-of-ai-powered-solutions) The way news is produced, delivered, and consumed keeps shifting, and as we move through 2025, a few big changes have become impossible to ignore. ![](https://www.recombee.com/img/blog/authors/annapetr.png) Gabriela Takacova & Violeta Milarova Feb 20, 2025 Personalization [![](https://www.recombee.com/img/blog/2024-wrap-up.png)](https://www.recombee.com/blog/2024-wrap-up) ### [2024 Wrap-Up](https://www.recombee.com/blog/2024-wrap-up) As we wrap up 2024, let’s take a moment to celebrate the milestones we’ve achieved together. Your collaboration and feedback have been at the heart of our progress. Here’s a quick recap of last year’s highlights. ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Jan 27, 2025 New Features Personalization [![](https://www.recombee.com/img/blog/celestial-tiger-entertainment-launches-new-chinese-movie-app-cmgo-with-diagnal.png)](https://www.diagnal.com/cmgo/) ### [Celestial Tiger Entertainment launches new Chinese Movie app, CMGO, with DIAGNAL](https://www.diagnal.com/cmgo/) With Recombee’s AI-powered recommendation engine working with DIAGNAL Enhance, CMGO serves up personalised experiences for each viewer, driving engagement for the service. ![](https://www.recombee.com/img/blog/authors/diagnal.png) Diagnal Nov 14, 2024 Personalization Partnerships [![](https://www.recombee.com/img/blog/introducing-beeformer-a-framework-for-training-foundational-models-for-recommender-systems.png)](https://www.recombee.com/blog/introducing-beeformer-a-framework-for-training-foundational-models-for-recommender-systems) ### [Introducing beeFormer: A Framework for Training Foundational Models for Recommender Systems](https://www.recombee.com/blog/introducing-beeformer-a-framework-for-training-foundational-models-for-recommender-systems) In the fast-evolving world of recommender systems, understanding both how users interact with content and the actual content itself is crucial. Many existing recommender systems struggle to balance these two aspects... ![](https://www.recombee.com/img/blog/authors/vojtechvancura.png) Vojtech Vancura Oct 15, 2024 New Features Recommendation Engine [![](https://www.recombee.com/img/blog/video-recommendations-made-easy-integrating-axinom-mosaic-with-recombee.png)](https://www.axinom.com/webinar/video-backends-with-recommendations) ### [Video Recommendations Made Easy: Integrating Axinom Mosaic with Recombee](https://www.axinom.com/webinar/video-backends-with-recommendations) In this webinar we look into building a data-driven video backend geared towards personalized video recommendations, integration with Axinom Mosaic, and how to transform user experiences on streaming platforms. ![](https://www.recombee.com/img/blog/authors/grigorygrin.png) Grigory Grin (Axinom) Sep 09, 2024 Personalization Partnerships [![](https://www.recombee.com/img/blog/insights-the-next-level-of-analytics-in-recombee-ui.png)](https://www.recombee.com/blog/insights-the-next-level-of-analytics-in-recombee-ui) ### [Insights: The Next Level of Analytics in Recombee UI](https://www.recombee.com/blog/insights-the-next-level-of-analytics-in-recombee-ui) Insights, the analytics section of our Admin UI, offers various predefined and fully customizable reports to track recommended items and how users interact with these recommendations. ![](https://www.recombee.com/img/blog/authors/ondrejfiedler.png) Ondrej Fiedler May 09, 2024 New Features Recommendation Engine [![](https://www.recombee.com/img/blog/recombee-partners-with-axinom-to-enhance-video-streaming-experiences.png)](https://www.recombee.com/blog/recombee-partners-with-axinom-to-enhance-video-streaming-experiences) ### [Recombee Partners With Axinom to Enhance Video Streaming Experiences](https://www.recombee.com/blog/recombee-partners-with-axinom-to-enhance-video-streaming-experiences) This collaboration is set to introduce a new era of personalized and engaging digital user experiences. ![](https://www.recombee.com/img/blog/authors/janvaluch.png) Jan Valuch Mar 20, 2024 Partnerships Personalization [![](https://www.recombee.com/img/blog/elevate-your-personalization-strategy-with-recombees-innovative-features.png)](https://www.recombee.com/blog/elevate-your-personalization-strategy-with-recombees-innovative-features) ### [Elevate Your Personalization Strategy with Recombee's Innovative Features](https://www.recombee.com/blog/elevate-your-personalization-strategy-with-recombees-innovative-features) The digital landscape and customer preferences and behavior are changing faster than ever now. To help our clients stay on top of the game, our team has focused on developing innovative features... ![](https://www.recombee.com/img/blog/authors/janvaluch.png) Jan Valuch Mar 13, 2024 New Features Recommendation Engine [![](https://www.recombee.com/img/blog/modern-recommender-systems-part-2-data.png)](https://www.recombee.com/blog/modern-recommender-systems-part-2-data) ### [Modern Recommender Systems - Part 2: Data](https://www.recombee.com/blog/modern-recommender-systems-part-2-data) Data used by modern recommenders and how we can measure progress towards goals. ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Mar 07, 2024 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/recombee-real-time-ai-recommendations-as-the-new-destination-in-segment.png)](https://www.recombee.com/blog/recombee-real-time-ai-recommendations-as-the-new-destination-in-segment) ### [Recombee Real-Time AI Recommendations as the New Destination in Segment](https://www.recombee.com/blog/recombee-real-time-ai-recommendations-as-the-new-destination-in-segment) Segment has enabled its users to enjoy Recombee personalization services without the need to leave their platform and with minimum coding involved. With a few simple clicks, domains using Segment can upgrade their services to maximize the digital experience for their customers. ![](https://www.recombee.com/img/blog/authors/adelasloupenska.png) Adela Sloupenska Mar 05, 2024 Personalization Integrations Partnerships [![](https://www.recombee.com/img/blog/is-this-comment-useful-enhancing-personalized-recommendations-by-considering-user-rating-uncertainty.png)](https://www.recombee.com/blog/is-this-comment-useful-enhancing-personalized-recommendations-by-considering-user-rating-uncertainty) ### [Is This Comment Useful? Enhancing Personalized Recommendations by Considering User Rating Uncertainty](https://www.recombee.com/blog/is-this-comment-useful-enhancing-personalized-recommendations-by-considering-user-rating-uncertainty) Picture this: you're on the hunt for the perfect new smartphone, browsing through your favourite online electronics store. The online store’s recommendation engine pops up with what it thinks could be your possible next gadget love... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Mar 01, 2024 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/recombeelabs-2023-research-publications.png)](https://www.recombee.com/blog/recombeelabs-2023-research-publications) ### [Recombeelab's 2023 Research Publications](https://www.recombee.com/blog/recombeelabs-2023-research-publications) Recombeelab, a joint research laboratory of Recombee and the Faculty of Information Technology at the Czech Technical University in Prague, experienced a highly productive year in 2023, publishing a series of insightful and impactful papers in the field of recommendation systems. ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Jan 19, 2024 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/ai-news-and-outlook-for-2024.png)](https://www.recombee.com/blog/ai-news-and-outlook-for-2024) ### [AI News and Outlook for 2024](https://www.recombee.com/blog/ai-news-and-outlook-for-2024) We look at the most interesting research directions and assess the state of knowledge in key areas of AI. We'll also estimate future developments in 2024 so you know what to prepare for. ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Jan 16, 2024 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/ai-assistants-know-your-preferences-even-better-than-you-do.png)](https://www.recombee.com/blog/ai-assistants-know-your-preferences-even-better-than-you-do) ### [AI Assistants Know Your Preferences, Even Better Than You Do](https://www.recombee.com/blog/ai-assistants-know-your-preferences-even-better-than-you-do) Recommender systems and ethical controversies ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Nov 23, 2023 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/the-ai-revolution-in-the-media-industry.png)](https://www.recombee.com/blog/the-ai-revolution-in-the-media-industry) ### [The AI (R)Evolution in the Media Industry](https://www.recombee.com/blog/the-ai-revolution-in-the-media-industry) In today's digital age, personalization has become the cornerstone of the media industry. Whether it's tailoring content recommendations, refining marketing strategies, or enhancing user experiences... ![](https://www.recombee.com/img/blog/authors/annapetr.png) Anna Dolezelova & Petr Popov Oct 23, 2023 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/modern-recommender-systems-part-1-introduction.png)](https://www.recombee.com/blog/modern-recommender-systems-part-1-introduction) ### [Modern Recommender Systems - Part 1: Introduction](https://www.recombee.com/blog/modern-recommender-systems-part-1-introduction) How machine learning methods simplify item discovery and search. ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Apr 17, 2023 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/explaining-recommender-systems-to-product-owners.png)](https://pavelkordik.substack.com/p/explaining-recommender-systems-to) ### [Explaining Recommender Systems to Product Owners](https://pavelkordik.substack.com/p/explaining-recommender-systems-to) In my presentation at the Data Technology Seminar organized by the European Broadcasting Union, I have focused on demonstrating that recommender systems can actually help public media organizations to better fulfill their role in society and reduce content distribution biases. ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Mar 27, 2023 Personalization Recommendation Engine [![](https://www.recombee.com/img/blog/inductive-matrix-completion-how-to-improve-recommendations-for-cold-start-users-and-items-by-incorporating-their-attributes.png)](https://www.recombee.com/blog/inductive-matrix-completion-how-to-improve-recommendations-for-cold-start-users-and-items-by-incorporating-their-attributes) ### [Inductive Matrix Completion: How to Improve Recommendations for Cold Start Users and Items by Incorporating Their Attributes](https://www.recombee.com/blog/inductive-matrix-completion-how-to-improve-recommendations-for-cold-start-users-and-items-by-incorporating-their-attributes) Matrix completion (MC), the problem of recovering the missing entries of a partially observed matrix, has found use in a wide range of domains. Still, its potentially most successful application is as a collaborative filtering technique for recommender systems (RSs)... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Mar 20, 2023 Personalization Recommendation Engine [![](https://www.recombee.com/img/blog/breaking-the-news-the-role-of-ai-in-modern-journalism.png)](https://www.recombee.com/blog/breaking-the-news-the-role-of-ai-in-modern-journalism) ### [Breaking the News: The Role of AI in Modern Journalism](https://www.recombee.com/blog/breaking-the-news-the-role-of-ai-in-modern-journalism) Artificial Intelligence (AI) has rapidly transformed the media industry in recent years. From automated news production to trend analysis and personalized content recommendations, AI has brought significant changes to the way media is created, distributed, and consumed. ![](https://www.recombee.com/img/blog/authors/tanalancova.png) Tana Lancova Mar 14, 2023 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/innovative-personalization-features-for-2023.png)](https://www.recombee.com/blog/innovative-personalization-features-for-2023) ### [Innovative Personalization Features for 2023](https://www.recombee.com/blog/innovative-personalization-features-for-2023) The digital world is changing; users' expectations for personalization are increasing, and our Recombee features are continuously improving. One of our focuses is to support our clients in providing the best possible user experiences... ![](https://www.recombee.com/img/blog/authors/gabrielatakacova.png) Gabriela Takacova Feb 17, 2023 New Features Recommendation Engine [![](https://www.recombee.com/img/blog/recombee-item-segmentations.png)](https://www.recombee.com/blog/recombee-item-segmentations) ### [Recombee Item Segmentations](https://www.recombee.com/blog/recombee-item-segmentations) Item Segmentations are Recombee's original and elegant solution to various advanced tasks related to hierarchical and relational data. The feature provides a flexible way to group items (products or pieces of content) into segments... ![](https://www.recombee.com/img/blog/authors/tomasrehorek.png) Tomas Rehorek Jan 11, 2023 New Features Recommendation Engine [![](https://www.recombee.com/img/blog/bandit-models-exploiting-popularity-and-curiosity-to-recommend-trending-content.png)](https://www.recombee.com/blog/bandit-models-exploiting-popularity-and-curiosity-to-recommend-trending-content) ### [Bandit Models: Exploiting Popularity and Curiosity to Recommend Trending Content](https://www.recombee.com/blog/bandit-models-exploiting-popularity-and-curiosity-to-recommend-trending-content) Humans are inherently curious. In fact, curiosity is linked to the evolution of humankind. For instance, according to famous historian Yuval Noah Harari in his bestseller book "Sapiens", our language skills evolved as a way of gossiping... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Oct 16, 2022 Recommendation Engine [![](https://www.recombee.com/img/blog/keeping-up-with-digital-media-convergence.png)](https://www.recombee.com/blog/keeping-up-with-digital-media-convergence) ### [Keeping Up With Digital Media Convergence](https://www.recombee.com/blog/keeping-up-with-digital-media-convergence) At Recombee, we felt the transition within the media industry accelerated by the pandemic. OTT and CTV consumption ballooned at a significant rate. ![](https://www.recombee.com/img/blog/authors/annadolezelova.png) Anna Dolezelova Oct 8, 2022 Personalization [![](https://www.recombee.com/img/blog/recombee-in-e-mail-marketing-a-partner-success-story-with-ryzeo.png)](https://www.recombee.com/blog/recombee-in-e-mail-marketing-a-partner-success-story-with-ryzeo) ### [Recombee in E-mail Marketing: A Partner Success Story with Ryzeo](https://www.recombee.com/blog/recombee-in-e-mail-marketing-a-partner-success-story-with-ryzeo) Do you feel there is a potential to increase your success with customers through an efficient recommender engine? You're highly likely right. Adding a recommender service to your emailing campaigns gives each client tailored product recommendations in all of their emails. ![](https://www.recombee.com/img/blog/authors/russellmiller.png) Russell Miller (Ryzeo) Oct 5, 2022 Partnerships [![](https://www.recombee.com/img/blog/how-we-are-using-ai-to-power-content-recommendations.png)](https://revium.com.au/blog/how-we-are-using-ai-to-power-content-recommendations) ### [How We Are Using AI to Power Content Recommendations](https://revium.com.au/blog/how-we-are-using-ai-to-power-content-recommendations) In this article we walk you through how we are using the AI recommendation engine Recombee embedded in our headless CMS StoryBlok to drive content recommendations throughout our own website. ![](https://www.recombee.com/img/blog/authors/revium.png) Revium Sep 30, 2022 Partnerships [![](https://www.recombee.com/img/blog/repsys-opensource-library-for-interactive-evaluation-of-recommendation-systems.png)](https://www.recombee.com/blog/repsys-opensource-library-for-interactive-evaluation-of-recommendation-systems) ### [Visual and Interactive Evaluation of Recommender Systems](https://www.recombee.com/blog/repsys-opensource-library-for-interactive-evaluation-of-recommendation-systems) When building modern real-world artificial intelligence systems, it is increasingly important to validate that the system works correctly. This is however not an easy task. Existing tools for machine learning practitioners... ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Sep 19, 2022 Recommendation Engine [![](https://www.recombee.com/img/blog/real-time-personalization-of-content-with-ai-powered-recommendations.png)](https://www.recombee.com/blog/real-time-personalization-of-content-with-ai-powered-recommendations) ### [Real-Time Personalization of Content With AI-Powered Recommendations](https://www.recombee.com/blog/real-time-personalization-of-content-with-ai-powered-recommendations) Do you manage a publishing company, online gaming platform, or a streaming site with a content-heavy catalog and are thinking about how to improve the user experience? ![](https://www.recombee.com/img/blog/authors/karenharazimova.png) Karen Harazimova Sep 16, 2022 Personalization [![](https://www.recombee.com/img/blog/ai-powered-content-recommendations-with-a-headless-cms.png)](https://www.luminary.com/blog/ai-content-recommendations-headless-cms-recombee) ### [AI-Powered Content Recommendations With a Headless CMS](https://www.luminary.com/blog/ai-content-recommendations-headless-cms-recombee) Thanks to its API-first nature, it is quite straightforward to integrate your headless CMS with the most powerful AI-powered content recommendations available on the market. Luminary just did that with their own website, Kontent.ai and Recombee. ![](https://www.recombee.com/img/blog/authors/andythompson.png) Andy Thompson (Luminary) Aug 31, 2022 Partnerships [![](https://www.recombee.com/img/blog/making-linear-autoencoders-work-for-large-scale-recommendation-systems.png)](https://www.recombee.com/blog/making-linear-autoencoders-work-for-large-scale-recommendation-systems) ### [Making Linear Autoencoders Work for Large Scale Recommendation Systems](https://www.recombee.com/blog/making-linear-autoencoders-work-for-large-scale-recommendation-systems) Linear autoencoders for collaborative filtering in recommender systems are simple and surprisingly accurate as we explained in our blogpost on how linear methods work. The critical disadvantage of methods like EASE is that they are not applicable to real-world problems... ![](https://www.recombee.com/img/blog/authors/vojtechvancura.png) Vojtech Vancura Aug 29, 2022 Recommendation Engine [![](https://www.recombee.com/img/blog/new-features-for-a-better-personalization-experience.png)](https://www.recombee.com/blog/new-features-for-a-better-personalization-experience) ### [New Features for a Better Personalization Experience](https://www.recombee.com/blog/new-features-for-a-better-personalization-experience) Like most of the world, the majority of 2021 was spent on home office or in isolation - which left us with all the time to be invested in work (and Netflix :) ) and improving UX for our clients. We are now happy to share new features we can offer to reach new levels of personalization. ![](https://www.recombee.com/img/blog/authors/ondrejfiedler.png) Ondrej Fiedler Mar 07, 2022 New Features [![](https://www.recombee.com/img/blog/advancing-your-career-in-artificial-intelligence-with-prg-ai-and-recombee.png)](https://www.recombee.com/blog/advancing-your-career-in-artificial-intelligence-with-prg-ai-and-recombee) ### [Advancing Your Career in Artificial Intelligence with prg.ai and Recombee](https://www.recombee.com/blog/advancing-your-career-in-artificial-intelligence-with-prg-ai-and-recombee) At Recombee, we have always collaborated with academia — after all, five of our co-founders graduated from the Czech Technical University in Prague, one of the largest and oldest technical universities in Europe, and most of them hold a Ph.D. degree. ![](https://www.recombee.com/img/blog/authors/karenharazimova.png) Karen Harazimova Nov 21, 2021 [![](https://www.recombee.com/img/blog/linear-methods-and-autoencoders-in-recommender-systems.png)](https://www.recombee.com/blog/linear-methods-and-autoencoders-in-recommender-systems) ### [Linear Methods and Autoencoders in Recommender Systems](https://www.recombee.com/blog/linear-methods-and-autoencoders-in-recommender-systems) Linear regression is probably the simplest and surprisingly efficient machine learning method. It should be the method of your first choice, according to the famous KISS principle. Also, it often works better than sophisticated methods, because it is... ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Sep 7, 2021 Recommendation Engine [![](https://www.recombee.com/img/blog/recombee-xperience-kentico.png)](https://xperience.io/discover/blog/recombee-kentico-xperience-1-on-1-personalization) ### [Recombee and Kentico Xperience: Guide to One-On-One Personalization](https://xperience.io/discover/blog/recombee-kentico-xperience-1-on-1-personalization) Recombee expanded its integration options - and now is available at the Kentico Xperience platform! Analyzing different types of personalization, we look into why Kentiko chose our AI-powered recommendation engine over manual segmentation. ![](https://www.recombee.com/img/blog/authors/gabrielatakacova.png) Gabriela Takacova May 6, 2021 Personalization Recommendation Engine [![](https://www.recombee.com/img/blog/recombee-in-2020/main.png)](https://www.recombee.com/blog/recombee-in-2020) ### [Recombee in 2020: New Features and Improvements](https://www.recombee.com/blog/recombee-in-2020) We know that this year has been quite challenging for many people, including ourselves. However, today we want to focus entirely on the positive (no pun included) side of the year and the stuff we are the proudest of. ![](https://www.recombee.com/img/blog/authors/ondrejfiedler.png) Ondrej Fiedler Dec 30, 2020 New Features [![](https://www.recombee.com/img/blog/23d22w82u3d52as.png)](https://medium.com/recombee-blog/deep-learning-for-recommender-systems-next-basket-prediction-and-sequential-product-recommendation-796228b34dee) ### [Deep Learning for Recommender Systems: Next Basket Prediction and Sequential Product Recommendation](https://medium.com/recombee-blog/deep-learning-for-recommender-systems-next-basket-prediction-and-sequential-product-recommendation-796228b34dee) Accurate “next basket prediction” will be enabling next generation e-commerce — predictive shopping and logistics. In this blogpost, we will discuss the deep learning technology behind next basket... ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Nov 10, 2020 Personalization Recommendation Engine [![](https://www.recombee.com/img/blog/how-interdisciplinary-collaboration-can-accelerate-ai-innovation.png)](https://www.recombee.com/blog/how-interdisciplinary-collaboration-can-accelerate-ai-innovation) ### [How Interdisciplinary Collaboration Can Accelerate AI Innovation](https://www.recombee.com/blog/how-interdisciplinary-collaboration-can-accelerate-ai-innovation) In a world where innovation is the new standard, Recombee uses the power of interdisciplinary collaboration to stay at the cutting edge of innovation. Partnering up with the leading player in the food industry (Bofrost) and academia (FIT CTU), allowed Recombee to hold a student competition to create AI which can shape the future of the food industry. ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Sep 1, 2020 [![](https://cdn-images-1.medium.com/max/1000/1*dSVF4ZxwQPIaIdnmPCsmPQ.png)](https://medium.com/recombee-blog/introduction-to-personalized-search-2b70eb5fa5ae) ### [Introduction to Personalized Search](https://medium.com/recombee-blog/introduction-to-personalized-search-2b70eb5fa5ae) Personalized search should take into account user preferences and interactions of similar users. We combined search engine and recommender. ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Mar 24, 2020 [![](https://cdn-images-1.medium.com/max/1000/1*YTIhIUMgAequmZ7cmghKTA@2x.png)](https://medium.com/recombee-blog/recombee-in-2019-new-features-and-improvements-bcbfc85acb80) ### [Recombee in 2019: New Features and Improvements](https://medium.com/recombee-blog/recombee-in-2019-new-features-and-improvements-bcbfc85acb80) This year was really huge for us. We worked on new features so hard that we almost forgot to write a blog post about them :) ![](https://www.recombee.com/img/blog/authors/tomasrehorek.png) Tomas Rehorek Dec 15, 2019 New Features [![](https://cdn-images-1.medium.com/max/1000/0*IhtrJIi4PvdZ2SAz)](https://medium.com/recombee-blog/check-out-our-new-client-side-integration-support-and-deploy-personalized-recommendations-faster-7dd7bf5b6241) ### [Check out Our New Client-Side Integration Support and Deploy Personalized Recommendations Faster](https://medium.com/recombee-blog/check-out-our-new-client-side-integration-support-and-deploy-personalized-recommendations-faster-7dd7bf5b6241) We knew we had to bring something new to the table, when participating as a Beta Startup at Web Summit, the largest technology conference… ![](https://www.recombee.com/img/blog/authors/ondrejfiedler.png) Ondrej Fiedler Jan 7, 2019 Recommendation Engine [![](https://cdn-images-1.medium.com/max/1000/0*To3_sKOFKyDE-v0k)](https://medium.com/recombee-blog/machine-learning-for-recommender-systems-part-2-deep-recommendation-sequence-prediction-automl-f134bc79d66b) ### [Machine Learning for Recommender Systems — Part 2 (Deep Recommendation, Sequence Prediction, AutoML…](https://medium.com/recombee-blog/machine-learning-for-recommender-systems-part-2-deep-recommendation-sequence-prediction-automl-f134bc79d66b) In the first part of our talk, we discussed basic algorithms, their evaluation and cold start problem. Below we show how deep learning… ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Jun 7, 2018 Recommendation Engine [![](https://cdn-images-1.medium.com/max/1000/0*0we0HYG_xQOlAvc5)](https://medium.com/recombee-blog/machine-learning-for-recommender-systems-part-1-algorithms-evaluation-and-cold-start-6f696683d0ed) ### [Machine Learning for Recommender Systems — Part 1 (Algorithms, Evaluation and Cold Start)](https://medium.com/recombee-blog/machine-learning-for-recommender-systems-part-1-algorithms-evaluation-and-cold-start-6f696683d0ed) Recommender systems are one of the most successful and widespread application of machine learning technologies in business. There were many… ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Jun 3, 2018 Recommendation Engine [![](https://cdn-images-1.medium.com/max/1000/1*hyT8WbrDzFNuyMcS4ppB4A.png)](https://medium.com/recombee-blog/migrating-to-recombee-from-microsoft-cognitive-services-recommendations-9e0c0a9b34a5) ### [Migrating to Recombee From Microsoft Cognitive Services Recommendations](https://medium.com/recombee-blog/migrating-to-recombee-from-microsoft-cognitive-services-recommendations-9e0c0a9b34a5) Microsoft has recently discontinued the Recommendations within the Azure Cognitive Services (MCSR). If you used this service, you are… ![](https://www.recombee.com/img/blog/authors/ondrejfiedler.png) Ondrej Fiedler Mar 22, 2018 Recommendation Engine [![](https://cdn-images-1.medium.com/max/1000/1*C1FQaVx1ztTq-y0H0YsQ0g.png)](https://medium.com/recombee-blog/personalized-push-notifications-enabled-by-artificial-intelligence-8ac057bc97ba) ### [Personalized Push Notifications Enabled by Artificial Intelligence](https://medium.com/recombee-blog/personalized-push-notifications-enabled-by-artificial-intelligence-8ac057bc97ba) Recent progress in artificial intelligence enables us to design proactive AI systems. Whereas traditional recommender systems produce… ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Mar 4, 2018 Recommendation Engine [![](https://cdn-images-1.medium.com/max/1000/1*E9kKJlAERzHEi4V556Wj4Q.png)](https://medium.com/recombee-blog/personalized-recommendations-in-10-minutes-bcbea144974b) ### [Personalized Recommendations in 10 Minutes](https://medium.com/recombee-blog/personalized-recommendations-in-10-minutes-bcbea144974b) We have just released a video tutorial that will guide you through the integration of the Recombee recommendation service to your… ![](https://www.recombee.com/img/blog/authors/ondrejfiedler.png) Ondrej Fiedler Jan 11, 2017 [![](https://cdn-images-1.medium.com/max/1000/1*d8rJ5EWZTOgfEp-lKYBJeA.jpeg)](https://medium.com/recombee-blog/evaluating-recommender-systems-choosing-the-best-one-for-your-business-c688ab781a35) ### [Evaluating Recommender Systems: Choosing the Best One for Your Business](https://medium.com/recombee-blog/evaluating-recommender-systems-choosing-the-best-one-for-your-business-c688ab781a35) Together with the endless expansion of E-commerce and online media in the last years, there are more and more Software-as-a-Service (SaaS)… ![](https://www.recombee.com/img/blog/authors/tomasrehorek.png) Tomas Rehorek Dec 18, 2016 Personalization Recommendation Engine [![](https://cdn-images-1.medium.com/max/1000/1*iKboSkp-zHi8ZNH8ChtP1w.png)](https://medium.com/recombee-blog/the-value-of-personalized-recommendations-for-your-business-6b2e81ce0a4d) ### [The Value of Personalized Recommendations for Your Business](https://medium.com/recombee-blog/the-value-of-personalized-recommendations-for-your-business-6b2e81ce0a4d) The e-commerce boom makes online environment more competitive. Internet retailers seek competitive advantages and a personalized experience… ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Aug 23, 2016 Personalization Recommendation Engine [![](https://miro.medium.com/max/1400/1*NARxz9O7ZX3Un_HMacPcWw.png)](https://medium.com/recombee-blog/recommender-systems-explained-d98e8221f468) ### [Recommender Systems Explained](https://medium.com/recombee-blog/recommender-systems-explained-d98e8221f468) In this article, I overview broad area of recommender systems, explain how individual algorithms work. ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Jul 12, 2016 Recommendation Engine [![](https://miro.medium.com/max/1102/1*fAWNYkIxk2hyLw1JiEi_9g.png)](https://medium.com/recombee-blog/generating-client-libraries-for-recombee-recommendation-api-75a76f915c47) ### [Generating Client Libraries for Recombee Recommendation API](https://medium.com/recombee-blog/generating-client-libraries-for-recombee-recommendation-api-75a76f915c47) Client libraries help programmers to integrate an API into their systems in a faster, easier and more readable way. We have recently published clients for Ruby and PHP, and we wish to provide clients for other major programming… ![](https://www.recombee.com/img/blog/authors/ondrejfiedler.png) Ondrej Fiedler May 17, 2016 [![](https://miro.medium.com/max/266/1*HUs4oA40zHOJkEo4ihLAHA.png)](https://medium.com/recombee-blog/personalized-recommendations-in-ruby-e3fcaa5de6be) ### [Personalized Recommendations in Ruby](https://medium.com/recombee-blog/personalized-recommendations-in-ruby-e3fcaa5de6be) For those of you, who develop in Ruby, we prepared a simple client application enabling you to benefit from our personalized recommendations. ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik May 16, 2016 Recommendation Engine [![](https://miro.medium.com/max/1400/1*NARxz9O7ZX3Un_HMacPcWw.png)](https://medium.com/recombee-blog/artificial-intelligence-in-the-cloud-310da5e0325e) ### [Artificial Intelligence in the Cloud](https://medium.com/recombee-blog/artificial-intelligence-in-the-cloud-310da5e0325e) At Recombee, we “think big”, and prefer making big leaps in technology over taking small steps. Our team has been involved in data science and artificial intelligence research for many years. Beginning in 2012, we began to capitalize our knowledge and experience, developing products which… ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Mar 30, 2016 Personalization Recommendation Engine --- # 2025 Sneak Peek > Source: https://www.recombee.com/blog/2025-sneak-peek > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # 2025 Sneak Peek ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Mar 19, 2025 ![](https://www.recombee.com/img/blog/2025-sneak-peek/main.png) This year is already off to an exciting start, and we’re rolling out new tools to improve efficiency and optimize recommendations. Here’s what’s available and what’s coming next. ## Copy Settings Between Recombee Databases - (Now Available!) ![](https://www.recombee.com/img/blog/2025-sneak-peek/01.png) Managing multiple Recombee databases is now more efficient than ever. With this newly released feature, you can effortlessly copy settings such as defined properties, business rules, scenarios, and widgets between databases. Configure and fine-tune your setup in a development environment, test it thoroughly, and then easily transfer everything to your production database. Say goodbye to manual duplication and hello to streamlined workflows, improved consistency, and reduced errors. This feature empowers you to focus on innovation while ensuring a smooth transition from testing to production. ## Optimize with A/B Testing for Scenarios ![](https://www.recombee.com/img/blog/2025-sneak-peek/02.png) Coming in 2025, Recombee will introduce built-in A/B Testing for Scenarios, giving you the power to fine-tune your recommendations like never before. This feature will allow you to test different configurations such as varying Boosters or Logics within the same Scenario. By running controlled experiments and evaluating performance metrics, you can determine which settings deliver the best results for your users. Whether you're optimizing for engagement, conversions, or other KPIs, A/B Testing will provide valuable insights to help you make data-driven decisions and maximize the impact of recommendations. ## Recombee Widget SDK ![](https://www.recombee.com/img/blog/2025-sneak-peek/03.png) In 2025, Recombee will launch the Widget SDK, a powerful new tool designed to make creating customized recommendation widgets a whole lot easier. While our JavaScript SDK offers full flexibility to build visualizations from scratch, and the No-Code Widget Editor in our Admin UI enables widget creation without any programming, the Widget SDK strikes the perfect balance. It empowers developers to create highly customized widgets without starting from scratch, simplifying the process while retaining creative control. ## Algorithmic Improvements ![](https://www.recombee.com/img/blog/2025-sneak-peek/04.png) This year, we will continue improving our LLM-based semantic search, making it available to more customers. To further enhance the performance of our recommender system, we are developing advanced collaborative filtering algorithms using sparse autoencoders, along with multimodal transformer-based models that process both images and text. The deployment of the multimodal beeFormer will significantly improve the accuracy of our recommendations, particularly in cold-start scenarios and beyond. We’re excited for what this year holds and can’t wait to bring you even more tools to drive your success! 🤝 Recommendation Engine New Features ## Next Articles [![](https://www.recombee.com/img/blog/making-recommendations-fairer-a-new-way-to-guarantee-exposure-for-all.png)](https://www.recombee.com/blog/making-recommendations-fairer-a-new-way-to-guarantee-exposure-for-all) ### [Making Recommendations Fairer: A New Way to Guarantee Exposure for All](https://www.recombee.com/blog/making-recommendations-fairer-a-new-way-to-guarantee-exposure-for-all) As recommender systems become more widespread across digital platforms, concerns around fairness are coming to the forefront. Standard relevance-based ranking techniques, while effective... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Apr 29, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/build-vs-buy-deciding-the-best-approach-for-your-recommender-system.png)](https://www.recombee.com/blog/build-vs-buy-deciding-the-best-approach-for-your-recommender-system) ### [Build vs. Buy: Deciding the Best Approach for Your Recommender System](https://www.recombee.com/blog/build-vs-buy-deciding-the-best-approach-for-your-recommender-system) When it comes to deciding between buying a recommender system and building one from scratch, the choice isn’t always straightforward. Both options come with their own set of pros and cons... ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova & Ondrej Fiedler Mar 14, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/are-you-here-to-stay-unraveling-the-dynamics-of-stable-and-curious-audiences-in-web-systems.png)](https://www.recombee.com/blog/are-you-here-to-stay-unraveling-the-dynamics-of-stable-and-curious-audiences-in-web-systems) ### [Are You Here to Stay? Unraveling the Dynamics of Stable and Curious Audiences in Web Systems](https://www.recombee.com/blog/are-you-here-to-stay-unraveling-the-dynamics-of-stable-and-curious-audiences-in-web-systems) Why do influencers frequently request their subscribers to enable all notifications for their channels? This practice stems from their awareness that not all subscribers are regular... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Mar 03, 2025 Recommendation Engine Personalization --- # A 2025 Research Retrospective > Source: https://www.recombee.com/blog/a-2025-research-retrospective > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # A 2025 Research Retrospective ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Jan 29, 2026 ![](https://www.recombee.com/img/blog/a-2025-research-retrospective/main.png) In 2025, I became Head of Research at Recombee. I knew it would be a challenge: and I wasn’t wrong. Recombee is a leader in recommendation-as-a-service, with thousands of clients across a wide range of domains. Keeping research moving forward while staying close to real product constraints (latency, scale, reliability, feedback loops…) is exciting, but it’s definitely not easy. Toward the end of the year, I did something I rarely manage in the middle of ongoing projects: I stepped back and looked at the recommender-systems field as a whole. More importantly, I looked at where our industrial research team has actually moved the needle. Recommendation is evolving fast right now. In industry, that speed does not show up as abstract trends; it shows up as a constant stream of concrete decisions: tuning models, shipping features, meeting strict latency budgets, and ensuring we are not accidentally creating harmful feedback loops. Just as importantly, any new modeling improvement must be weighed against how difficult it is to productionalize, monitor, and maintain as part of a larger product. This post is my attempt to connect those day-to-day decisions to the underlying ideas. I want to share what we worked on this year and offer an industrial researcher’s view of what “modern recommendation” looks like in practice, where scale, uncertainty, sequential dynamics, and engineering constraints can matter just as much as the choice of model family. ## A Strong Year for Recombee Research It is challenging, but also very rewarding. I am very happy (and proud) to share that, in total, our researchers published 15 papers across five conferences, with the vast majority ranked A or A\*, and six journals, mostly high-impact Q1\. There is no way I can cover everything in one post, so I will focus on a handful of contributions that best show two things: how we have been contributing to the broader community, and how that work translates into real value for our clients. ### The Future is Sparse One recurring bottleneck in real systems is retrieval and serving cost. For instance, dense embeddings are a convenient default, but at scale they become one of the most expensive parts of the stack: memory, bandwidth, and approximate nearest-neighbor search all start to dominate. Several of our 2025 results orbit the idea that sparsity is not merely a compression trick applied at the end, but a representational choice that can change the trade-offs you can achieve. If your pipeline depends on retrieving candidates quickly and repeatedly, then being able to store and search representations efficiently is not a “nice to have”: it’s the condition under which more sophisticated ranking is even feasible. For our clients, this means faster retrieval, which unlocks larger candidate pools and leaves more budget for stronger ranking models. In short, efficiency at retrieval directly translates into better recommendations at production scale. ### Scalability is always in the room Another theme we kept coming back to in 2025 was being disciplined about baselines. In industry, you quickly relearn a humbling lesson: “simple” models can stay surprisingly competitive when they are implemented well and evaluated properly at scale. That is not an argument against deep learning. We use plenty of deep models in our ML stack. It is an argument for being honest about what added complexity actually buys you, when it is worth it, and where it belongs in the recommendation stack. That is why we spent time evaluating linear models and shallow autoencoder-style recommenders on large datasets. We wanted to see what truly breaks as the interaction matrix grows, separate algorithmic approximations from real objective changes, and keep a set of scalable baselines that new methods have to beat under realistic conditions. For companies, this is practical, not philosophical. The best model is not always the one with the fanciest architecture. It is the one that stays stable under continuous updates, can be monitored and explained, and delivers the best trade-off between quality, latency, and cost. ### Beyond “Next Click” With thousands of clients and billions of interactions, top k or next click is definitely not our only scenario. We also spent time on sequential and dynamic settings where “predict the next click” is not the right way to think about the problem. Baskets, sessions, and recurring behaviors have real structure, and static models often miss it. At the same time, jumping straight to heavyweight sequence encoders can turn the system into something that is hard to control and even harder to debug. That is why we explored sequence approaches that stay interpretable, with temporal windows and dependency operators made explicit. Even when a model is not simple computationally, it can still be simple scientifically: you can ask what it learned, which dependencies matter most, and how those dependencies change across cohorts or seasons. In applied work, that kind of interpretability is not a nice extra. It is what lets you connect model behavior to product hypotheses and operational constraints. This means better recommendations in session and repeat interaction scenarios, without turning the system into a black box. The added interpretability makes it easier to debug issues, run safer experiments, and translate model behavior into clear product and business decisions. ### Trust, Safety, and Responsible Recommendations We also kept a parallel track on trust-related properties that companies increasingly need in their recommender stack: explanation, safety, fairness, and diversity. The point here is not to attach slogans to systems, but to treat these as technical objects, with datasets, constraints, and measurable properties that can be tested. Explanations matter because they improve debugging and human oversight. Diversity and serendipity matter because they reduce the risk of systems collapsing into popularity loops, especially in cold start settings. Fairness matters because group and marketplace scenarios can amplify distributional imbalances. Safety matters because stronger semantic search and richer representations can surface harmful or sensitive content unless you explicitly align for it and evaluate it. ### Generative AI in Recommendation Systems Generative AI also became a real part of our 2025 agenda, not as a replacement for recommender systems, but as an additional set of tools that changes how we represent items, interpret intent, and communicate decisions. In our collaboration with The Telegraph, we explored how LLMs can support editorial work by making segment level patterns easier to inspect and act on, while keeping the recommender backbone responsible for ranking and evaluation. In parallel, we started treating LLMs as first-class components in responsible recommendation: if language models are used to create or refine representations, they also shape fairness outcomes, so they need explicit constraints and auditing rather than implicit trust. This is shaping how we integrate LLMs into recommenders in practice, where they sit in the pipeline, what they are allowed to influence, and how their impact can be measured under real product constraints. Finally, 2025 also had a community dimension that matters to us as researchers. **RecSys 2025** was held in Prague (Recombee home), and several of us were involved not only as authors but also in organizing roles (including general chair, industry chair, and local chairs). For a field that sits between academia and product, conferences are not just places where papers are presented: they are part of the infrastructure that shapes standards for evaluation, reproducibility, and the quality of dialogue between research and practice. Contributing to that infrastructure is, in a very literal sense, part of advancing the science. ## …A Stronger 2026! ![](https://www.recombee.com/img/blog/a-2025-research-retrospective/01.png) Looking into 2026, we’re already planning to be at **The ACM Web Conference 2026 in Dubai** (April 13-17, 2026), which feels like a natural venue for the kind of “web-scale” questions that increasingly shape recommender systems in practice. The product-facing challenges aren’t getting smaller: interfaces are becoming conversational, feedback is messier and more implicit, and the boundary between “retrieval,” “recommendation,” and “assistance” is blurring. From a research perspective, a lot of our attention is therefore shifting toward **GenAI in conversational settings** (where the system must both understand intent and respond under uncertainty) and toward **agentic recommendation**, which are systems that don’t only rank items, but plan and adapt sequences of actions while staying controllable, evaluable, and safe. If you’d like to collaborate with us (whether you’re tackling recommendation problems in a product setting, working on the underlying theory, or exploring new interfaces like conversational and agentic systems) feel free to reach out. We’re always happy to discuss concrete problems, exchange ideas, and learn from other perspectives. And if our work is useful for your own research or engineering efforts, we’d appreciate it if you read it, cite it where relevant. ## Published Peer-Reviewed Publications \[1\] Kasalický, P., Spišák, M., Vančura, V., Bohuněk, D., Alves, R., & Kordík, P. (2025). _The Future is Sparse: Embedding Compression for Scalable Retrieval in Recommender Systems._ In **Proceedings of the 19th ACM Conference on Recommender Systems (RecSys 2025)** (pp. 1099–1103). \[2\] Spišák, M., Alves, R., Kelleher, T., Sheppard, J., Fiedler, O., Kosovrasti, E., Vančura, V., Kasalický, P., & Kordík, P. (2025). _Segment-Aware Analytics for Real-Time Editorial Support in Media Groups: Lessons from The Telegraph._ In **INRA 2025: 13th International Workshop on News Recommendation and Analytics** (CEUR Workshop Proceedings, Vol. 4056). \[3\] Vančura, V., Kasalický, P., Alves, R., & Kordík, P. (2025). _Evaluating Linear Shallow Autoencoders on Large Scale Datasets._ **ACM Transactions on Recommender Systems.** \[4\] Žid, Č., Alves, R., & Kordík, P. (2025). _Active Recommendation for Email Outreach Dynamics._ In **Proceedings of the 34th ACM International Conference on Information and Knowledge Management (CIKM 2025)** (pp. 5540–5544). \[5\] Zmeškalová, T., Ledent, A., Spišák, M., Kordík, P., & Alves, R. (2025). _Recurrent Autoregressive Linear Model for Next-Basket Recommendation._ In **Proceedings of the 19th ACM Conference on Recommender Systems (RecSys 2025)** (pp. 1273–1278). \[6\] Koštejn, V., Peška, L., & Spišák, M. (2025). _SAGEA: Sparse Autoencoder-based Group Embeddings Aggregation for Fairness-Preserving Group Recommendations._ In **Proceedings of the 19th ACM Conference on Recommender Systems (RecSys 2025)** (pp. 1290–1295). \[7\] Poernomo, J., Tan, N. G. L., Alves, R., & Ledent, A. (2025). _Probabilistic Modeling, Learnability and Uncertainty Estimation for Interaction Prediction in Movie Rating Datasets._ In **Proceedings of the 19th ACM Conference on Recommender Systems (RecSys 2025)** (pp. 1261–1266). \[8\] Ledent, A., Kasalický, P., Alves, R., & Lauw, H. W. (2025). _Conv4Rec: A 1-by-1 Convolutional Autoencoder for User Profiling Through Joint Analysis of Implicit and Explicit Feedback._ **IEEE Transactions on Neural Networks and Learning Systems.** \[9\] Ledent, A., Alves, R., & Lei, Y. (2025). _Generalization Bounds for Rank-sparse Neural Networks._ **Neurips 2025.** \[10\] Spacek, F., Vancura, V., & Kordik, P. (2025). _Mitigating Risks in Marketplace Semantic Search: A Dataset for Harmful and Sensitive Query Alignment._ In **Proceedings of the 33rd ACM Conference on User Modeling, Adaptation and Personalization (UMAP 2025)** (pp. 329–334). \[11\] Kuznetsov, S., & Kordík, P. (2025). _Improving recommendation diversity and serendipity with an ontology-based algorithm for cold start environments._ **International Journal of Data Science and Analytics**, 20(2), 431–443. \[12\] Alves, R. (2025). _SCORE: A convolutional approach for football event forecasting._ **International Journal of Forecasting.** \[13\] Cahlik, V., Alves, R., & Kordík, P. (2025). _Reasoning-grounded natural language explanations for language models._ In **Proceedings of the World Conference on Explainable Artificial Intelligence** (pp. 3–18). \[14\] Hänsch, S., Sajdoková, A., Rabau, A., Rybář, V., Alves, R., & Kordík, P. (2025). _Data-driven closure model selection for multiphase CFD via matrix completion._ **AI Thermal Fluids.** \[15\] Stambrouski, T., & Alves, R. (2025). _Multitask learning for cognitive sciences triplet analysis._ **Expert Systems with Applications**, 267, 126187. Recommendation Engine Personalization ## Next Articles [![](https://www.recombee.com/img/blog/mid-year-roundup-2026-so-far.png)](https://www.recombee.com/blog/mid-year-roundup-2026-so-far) ### [Mid-Year Roundup: 2026 So Far](https://www.recombee.com/blog/mid-year-roundup-2026-so-far) The digital world continues to change at an incredible pace. As technology evolves and user expectations keep rising, creating experiences that feel relevant has never been more important. ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Jul 20, 2026 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/looking-back-at-2025.png)](https://www.recombee.com/blog/looking-back-at-2025) ### [Looking Back at 2025](https://www.recombee.com/blog/looking-back-at-2025) 2025 marked 10 years of Recombee. A decade of building personalization from first principles, shaped by research, real-world deployments, and close collaboration with partners across industries. ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Jan 20, 2026 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/product-highlights-from-2025.png)](https://www.recombee.com/blog/product-highlights-from-2025) ### [Product Highlights from 2025](https://www.recombee.com/blog/product-highlights-from-2025) In 2025, we focused on making advanced personalization easier to implement, scale, and maintain across products and platforms. Here’s a look at the key product updates we released last year. ![](https://www.recombee.com/img/blog/authors/ondrejfiedler.png) Ondrej Fiedler Jan 08, 2026 Recommendation Engine Personalization --- # AI Assistants Know Your Preferences, Even Better Than You Do > Source: https://www.recombee.com/blog/ai-assistants-know-your-preferences-even-better-than-you-do > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # AI Assistants Know Your Preferences, Even Better Than You Do ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Nov 23, 2023 ![](https://www.recombee.com/img/blog/ai-assistants-know-your-preferences-even-better-than-you-do/main.png) The internet and online services have changed significantly in the last decade. They are smarter. They can better understand what you might like. And that's thanks to the inconspicuous algorithms - recommender systems. These are currently the most widely used group of algorithms in practice, along with conversational intelligence. It is good to know about them and be aware of how they work. Before the advent of the internet, scientists were working on how to help people navigate a large catalog of items. [Information retrieval algorithms](https://nlp.stanford.edu/IR-book/) solved how to select the right documents from many documents, such as items in a public library, for visitors. The original algorithms were based on the visitor entering what they were looking for, and the system returned the documents that best matched the query. These algorithms were not personalized. Everyone got the same results for the same query. Classic algorithms did not use machine learning and only solved the similarity of the query and the documents. The similarity of the query and the document was determined, for example, as the cosine distance of long sparse vectors counting the words that occur in these texts. In the vectors, in addition to the frequency of words, their importance was also taken into account. For example, the [tf-idf method](https://en.wikipedia.org/wiki/Tf%E2%80%93idf) gives less weight to words that occur in most documents. With the advent of the internet and online services, the size of various catalogs of items has exploded. In addition, it turned out that visitors often do not know what they are looking for. They would rather have something recommended to them, if the recommendation is relevant for them, which can be a real problem given the number of options. And here is where intelligent assistants in the form of recommender systems come in. Users connect to online services through personal computers and mobile phones, which merchants can uniquely identify using so-called cookies. They are also often logged in to the service using their personal account. Thanks to this, their visits are not anonymous and service providers can track their behavior. The most interesting are, of course, various interactions with the items in the catalog, whether their viewing, purchase, or explicit evaluation (e.g., adding to favorite items). ![](https://www.recombee.com/img/blog/ai-assistants-know-your-preferences-even-better-than-you-do/01.png) _Recommender systems enable online services to be personalized for individual users. Each user gets a different website containing items that were predicted relevant by AI algorithms powering the recommender system._ One of the first recommender systems was [GroupLens](https://en.wikipedia.org/wiki/GroupLens_Research), which started development in 1992 and recommended relevant articles to users based on their explicit historical interactions. However, it turned out that users are not very willing to rate items explicitly (stars, thumbs up or down), and therefore their preferences must be inferred from so-called implicit interactions (displaying an item, reading it, purchasing, etc.). Modern recommender systems learn on the implicit interactions of visitors with items to offer what to whom when. And that in all possible areas. You can recommend songs, books, videos, goods, real estate, jobs, news, etc. Although different areas have their own specifics, recommender algorithms are very similar. And while information retrieval algorithms solved how to help the user find an item as quickly and accurately as possible, the goal of modern recommender systems is to help users discover relevant items that they may not know about yet and generally improve personalization and the associated user experience when using online services. This leads to the fact that users are more loyal, return to the service more often and for a longer period of time, and thus their [value](https://www.subbly.co/blog/what-subscription-businesses-need-to-know-about-customer-lifetime-value-clv/) for the service provider increases overall. The most common method used for this is called [collaborative filtering](https://en.wikipedia.org/wiki/Collaborative_filtering), where we look for visitors with similar behavior and recommend items that the visitor has not yet seen. These algorithms help people discover new things on the given platform. The opposite are algorithms that remind visitors of items that they have already interacted with. Even that in itself is not a simple task, and algorithms must take into account the visitor's historical behavior as well as the periodicity of the items (how often visitors typically return to the item). Another class of algorithms are popularity algorithms (for example, contextual bandits), which recommend items trending in a group of users. [Bandit algorithms](https://www.recombee.com/blog/bandit-models-exploiting-popularity-and-curiosity-to-recommend-trending-content) try to balance appropriately between exploration and exploitation. Exploration offers users items that could become popular for that particular group. Exploitation then maximizes the use of already proven items that have been shown to work well for users. The system is rewarded, for example, if it recommends an item that the user is satisfied with. It gradually improves so that it maximizes the collected rewards (see [reinforcement learning](https://en.wikipedia.org/wiki/Reinforcement_learning)). Most modern algorithms also use the available information about the items (such as metadata such as text description, categories, images, etc.). These metadata are processed by modern neural networks and thus form their representation of the catalog of items. Thanks to this, they are able to recognize similar items, even if they are brand new and do not have any interactions yet. These algorithms are then often combined in a suitable way to create the final recommendation for the user. Algorithms must also be very fast (the user will not wait) and adaptive to various changes (for example, changes in the catalog of items). Modern recommender systems are also able to help users even if they have a more specific idea of what they are looking for. The user then communicates this idea to the system through various interactions (text search, filtering, etc.). This has somewhat unified recommender algorithms with personalized search and ranking algorithms, which were historically separate research fields (also known as [learning to rank](https://en.wikipedia.org/wiki/Learning_to_rank)). Currently, there is a public debate on how to ensure that visitors to online services have control over what the systems recommend to them and what data they use for this. The right to privacy (I don't want to share any data or interactions about myself) is in conflict with the usability of the service (I want the recommender system to have as much information about me as possible and be able to offer me the most relevant items from the catalog immediately). In my opinion, the user should be able to work with their [user profile](https://link.springer.com/chapter/10.1007/978-3-319-53676-7_2), similar to how we already know it from web browsers, where we can, for example, switch between multiple user profiles. Unfortunately, it is not yet entirely clear how to enable users to work effectively with their profile and how to [represent](https://dl.acm.org/doi/pdf/10.1145/3477495.3531873) their profile in a way that is understandable. Smaller online service providers may be the ones who suffer the most from [any regulation](https://www.aiact-info.eu/), as the development of a user interface for configuring user profiles will be expensive. However, without the ability to use modern recommender algorithms, their online services and products will not be competitive, as all larger online services already use recommender systems, and personalization is to a certain extent already standard for users. **This text was originally written for the [Czech edition of Wired journal](https://www.wired.cz/clanky/nenapadni-ai-pomocnici-vedi-co-hledate-lepe-nez-vy-sami) in collaboration with [prg.ai](https://prg.ai/en/).** Recommendation Engine Personalization ## Next Articles [![](https://www.recombee.com/img/blog/ai-news-and-outlook-for-2024.png)](https://www.recombee.com/blog/ai-news-and-outlook-for-2024) ### [AI News and Outlook for 2024](https://www.recombee.com/blog/ai-news-and-outlook-for-2024) We look at the most interesting research directions and assess the state of knowledge in key areas of AI. We'll also estimate future developments in 2024 so you know what to prepare for. ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Jan 16, 2024 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/the-ai-revolution-in-the-media-industry.png)](https://www.recombee.com/blog/the-ai-revolution-in-the-media-industry) ### [The AI (R)Evolution in the Media Industry](https://www.recombee.com/blog/the-ai-revolution-in-the-media-industry) In today's digital age, personalization has become the cornerstone of the media industry. Whether it's tailoring content recommendations, refining marketing strategies, or enhancing user experiences... ![](https://www.recombee.com/img/blog/authors/annapetr.png) Anna Dolezelova & Petr Popov Oct 23, 2023 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/modern-recommender-systems-part-1-introduction.png)](https://www.recombee.com/blog/modern-recommender-systems-part-1-introduction) ### [Modern Recommender Systems - Part 1: Introduction](https://www.recombee.com/blog/modern-recommender-systems-part-1-introduction) How machine learning methods simplify item discovery and search. ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Apr 17, 2023 Recommendation Engine Personalization --- # AI News and Outlook for 2026 > Source: https://www.recombee.com/blog/ai-news-and-outlook-for-2026 > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # AI News and Outlook for 2026 ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Dec 17, 2025 ![](https://www.recombee.com/img/blog/ai-news-and-outlook-for-2026/main.png) Here’s what caught my attention in AI research lately, and where things might be heading in 2026. After 25+ years in this field, the pace has gotten hard to keep up with. I’m trying to make sense of what actually matters from the flood of papers and demos. One thing that’s become obvious: **AI can now handle far more complex questions than before**. What used to be simple search has turned into systems that actually understand what you’re asking and figure out whether they need to look something up, reason through it, solve a problem, or spin up a whole agent workflow. The shift is dramatic. We’ve gone from systems that could barely handle simple Q&A to agents that can solve complex multi-step problems, navigate open-ended environments, and even play sophisticated games while continuously learning and adapting. This progress spans multiple fronts: better reasoning architectures, improved memory systems at all levels (from working memory to long-term episodic storage), and world models that now incorporate actions — not just passive observation, but active interaction with environments and prediction of consequences. This convergence is particularly driving robotics forward. When you combine better reasoning, persistent memory, and world models that understand cause and effect through actions, you get systems that can actually operate in the real world rather than just simulate it. **Note**: I’m planning a separate blog post on what’s new in recommender systems — there were too many great contributions at [RecSys 2025](https://recsys25.recombee.com/) alone, and I need more time to properly explore all the papers. ## Measuring Intelligence Large language models are everywhere now, which raises an obvious question: **how do we actually measure intelligence?** Most benchmarks test specific skills, but real intelligence means generalizing, reasoning, and adapting to situations you haven’t seen before. The **Abstraction and Reasoning Corpus (ARC-AGI)** ([arcprize.org](https://arcprize.org/)) tests whether AI can actually generalize beyond what it was trained on. It started as visual pattern puzzles but also applies to language tasks. The idea is _fluid intelligence_ — can you learn a simple rule from a few examples and apply it to new cases? ([Chollet, 2019](https://arxiv.org/abs/1911.01547)) ![](https://www.recombee.com/img/blog/ai-news-and-outlook-for-2026/1.png) Example ARC-AGI-2 task demonstrating compositional generalization. Tasks require combining multiple rules, multi-step reasoning, and contextual rule application. ARC-AGI tasks are easy for humans (often solvable in one or two tries) but extremely challenging for LLMs. Over 2025, model performance finally exceeded 90% on ARC-AGI-1\. Other, less well-known benchmarks are still unresolved — for example [CritPT](https://artificialanalysis.ai/evaluations/critpt), where even frontier LLMs perform poorly despite excelling on standard benchmarks. ## ARC-AGI-2 and Beyond [ARC-AGI-2](https://arxiv.org/pdf/2505.11831) (Chollet et al., 2025) raises the bar. It resists brute-force search and requires compositional generalization — combining rules in new ways across larger grids. Humans solve these tasks in about 2.7 minutes on average, but AI systems still struggle with rule composition. [ARC-AGI-3](https://three.arcprize.org/) is the next step: interactive environments where agents have to perceive, decide, and act without being told what to do. Can AI infer the goal by exploring? ![](https://www.recombee.com/img/blog/ai-news-and-outlook-for-2026/2.png) ARC-AGI-3: Interactive games requiring exploration, memory, and on-the-fly learning. The pattern is clear: AI is great at pattern matching but still weak at robust reasoning and open-ended learning of the kind humans do naturally. ## Refinement Loops The winning teams of the [2025 ARC Prize](https://arcprize.org/blog/arc-prize-2025-results-analysis) relied on outer-loop refinement, test-time fine-tuning, and synthetic data techniques rather than pure scaling. A refinement loop is simple: propose a solution, evaluate it, refine based on feedback, repeat until it works. This approach proved to be very beneficial. Teams like MindsAI (3rd place) used test-time training, while ARChitects(2nd place) applied recursive self-refinement with 2D diffusion. The [NVARC solution](https://developer.nvidia.com/blog/nvidia-kaggle-grandmasters-win-artificial-general-intelligence-competition/) involved extensive data engineering, enabling the training of a highly efficient 4B reasoning model, combined with novel recursive architectures (see below). Larger closed-source models such as [GPT-5.2](https://openai.com/index/introducing-gpt-5-2/) still outperform these smaller, more elegant systems by a large margin, but the resource asymmetry is significant. ## Recursive Transformers A recurrent transformer reuses some or all of its blocks across multiple iterations. Instead of stacking many different blocks, a smaller set is applied repeatedly with shared weights. ([TRM paper](https://arxiv.org/abs/2510.04871)) ### Hierarchical Reasoning Model (HRM) HRM ([paper](https://arxiv.org/abs/2506.21734), [GitHub](https://github.com/sapientinc/HRM)) is a nested recurrent transformer with two time scales: a fast loop for refinement and a slow loop that updates global state. ![](https://www.recombee.com/img/blog/ai-news-and-outlook-for-2026/3.png) Visualization of intermediate predictions by HRM on benchmark tasks. Top: Maze Hard. Middle: Sudoku-Extreme. Bottom: ARC-AGI-2 Task. HRM does this with just 27M parameters and 1,000 training samples, yet outperforms much larger models on reasoning benchmarks. The [TRM paper](https://arxiv.org/abs/2510.04871) shows similar results — a 7M recursive model can outperform models with 100× more parameters by encoding multi-step reasoning through iteration instead of learning it implicitly. ## Open-Ended Learning: The SIMA 2 Agent Open-ended learning refers to systems that keep acquiring new skills and goals without being limited to a fixed set. Unlike standard multi-task reinforcement learning with predetermined tasks, an open-ended learner generates its own objectives and learning signals. ### SIMA 2: Scalable, Instructable, Multiworld Agent [SIMA 2](https://deepmind.google/technologies/sima/) from DeepMind is an embodied agent that plays 3D games like a human. It was trained on 20+ environments with hundreds of skills and generalizes to new games like ASKA and MineDojo without extra training. It improves by proposing its own tasks and evaluating success autonomously. ![](https://www.recombee.com/img/blog/ai-news-and-outlook-for-2026/4.png) SIMA 2 Agent solving two different tasks: The agent demonstrates its ability to generalize across diverse game environments and tasks. SIMA 2 operates in a loop: perceive (visual + text), reason and plan, act, observe feedback, update policy, repeat. It uses internal reasoning blocks for chain-of-thought planning, has around 4k token context windows with short-term episodic memory, and replay buffers, and lets the LLM both propose tasks and evaluate success. The results are great: a raw [Gemini](https://deepmind.google/technologies/gemini/) model achieves 3-7% success, while SIMA 2 reaches near-human performance. Vision-action integration and structured learning matter far more than model size alone. ## Scaling Agent Systems As AI systems improve, a practical question emerges: when does it make sense to use multiple agents instead of a single one? Multi-agent systems work best on parallelizable tasks. [Finance-Agent’s](https://arxiv.org/pdf/2512.08296) centralized multi-agent system achieved an 81% improvement over a single agent by using specialized roles such as researcher, analyst, and executor. However, on tightly coupled sequential tasks, multi-agent setups can perform worse due to coordination overhead. ![](https://www.recombee.com/img/blog/ai-news-and-outlook-for-2026/5.png) Scaling Agent Systems: Understanding when multiple agents help vs. hurt. The lesson is that task decomposability matters a lot. Multi-agent approaches work when tasks can be cleanly split and parallelized Coordination costs grow super-linearly, and capability saturation is real: if a single agent is already strong, adding more agents can yield diminishing or even negative returns (Scaling Multi-Agent Systems). Specialization consistently beats redundancy and diversity in agentic systems is as important as in model ensembles. ## Memory Architectures: Beyond Context Windows As AI tackles increasingly complex, long-term tasks, the limitations of current memory systems become clear. Unlike humans, who can recall specific conversations, learn incrementally without forgetting, and build rich episodic memories, LLMs suffer from catastrophic forgetting, hallucinate facts, and lack persistent personal memory across sessions. For AI to truly collaborate with humans over long periods, it needs human-like memory: the ability to form lasting relationships, remember individual preferences and context, learn continuously without losing old knowledge, and distinguish between what it knows with certainty and what it is uncertain about. This drives the need for new memory architectures that separate reliable facts from uncertain inferences, maintain persistent identity across interactions, and support lifelong learning. Modern AI memory can be classified along three dimensions: object (personal vs. system), form (parametric vs. non-parametric), and time (short-term vs. long-term). This results in eight memory quadrants, each with distinct roles, analogous to human memory types. ([Memory Taxonomy Survey](https://arxiv.org/pdf/2504.15965)) ![](https://www.recombee.com/img/blog/ai-news-and-outlook-for-2026/6.png) Memory taxonomy in AI systems showing the three-dimensional classification: Object (Personal vs. System), Form (Parametric vs. Non-parametric), and Time (Short-term vs. Long-term). ### Nested Learning: Continuous Memory [Nested Learning](https://abehrouz.github.io/files/NL.pdf) from Google Research (NeurIPS 2025) represents an interesting direction in memory architecture design. Traditional AI systems separate model parameters from learning algorithms, whereas nested learning creates a unified memory hierarchy in which different layers learn at different timescales and retain information at multiple levels of abstraction. ![](https://www.recombee.com/img/blog/ai-news-and-outlook-for-2026/7.png) Nested Learning Paradigm that represents a machine learning model and its training procedure as a set of nested optimization problems. This memory architecture enables continuous adaptation without catastrophic forgetting: inner loops handle fast adaptation to immediate tasks, while outer loops preserve long-term knowledge. The result is a system that naturally balances stability and plasticity, making it well suited for lifelong learning scenarios in which AI agents must retain past experiences while adapting to new environments. ### MemoryOS and A-MEM [MemoryOS](https://arxiv.org/abs/2506.06326) is a memory operating system for AI agents, inspired by operating system memory management. It has three levels: Short-Term Memory (hot, similar to a cache for recent dialogue), Mid-Term Memory (warm, storing summaries and important segments), and Long-Term Memory (cold, large-scale persistent storage). [A-MEM](https://arxiv.org/abs/2506.06326) (Adaptive Memory for Embodied Multimodal Agents) introduces a memory architecture with three integral components in memory storage. During note construction, the system processes new interaction memories and stores them as notes with multiple attributes. The link generation process retrieves the most relevant historical memories and uses an LLM to determine whether connections should be established between them. The concept of a 'box' describes how related memories become interconnected through similar contextual descriptions. Individual memories can exist simultaneously in multiple boxes. During retrieval, the system extracts query embeddings using a text encoding model and searches the memory database for relevant matches. When a memory is retrieved, related memories linked within the same box are automatically accessed. ![](https://www.recombee.com/img/blog/ai-news-and-outlook-for-2026/8.png) A-MEM: Adaptive Memory architecture for embodied multimodal agents with interconnected memory boxes. These memory systems are becoming essential as AI tackles increasingly complex, long-term tasks that require reliable retention and retrieval. ## World Models: From Video Generation to Robotics As AI systems increasingly need to plan and generalize to solve new tasks, world models are becoming more important. These systems predict future environment states and are evolving from simple predictors into systems that generate realistic video, capture physical dynamics, and control robots. ### Video Generation as World Modeling Models like [Sora](https://openai.com/index/video-generation-models-as-world-simulators/) from OpenAI, [Veo](https://deepmind.google/models/veo/) from Google DeepMind, and HunyuanVideo from Tencent show that video generation models can act as world simulators (see [World Models paper](https://arxiv.org/abs/1803.10122)). They learn physics, object permanence, and causal relationships from large-scale video data, which changes how we think about AI interaction with the physical world. ![](https://www.recombee.com/img/blog/ai-news-and-outlook-for-2026/9.png) Veo: Advanced video generation model functioning as a world simulator. ### Vision-Language-Action (VLA) Models for Robotics A VLA model is a foundation model that takes visual input, processes language commands, and outputs actions to control an agent. Vision + language + action equals powerful embodied AI. [DeepMind’s Gemini Robotics](https://arxiv.org/abs/2503.20020) (2025) and [NVIDIA GR00T N1](https://arxiv.org/pdf/2503.14734) demonstrate what VLA models can achieve in real-world settings. ![](https://www.recombee.com/img/blog/ai-news-and-outlook-for-2026/10.png) GR00T N1: A Vision-Language-Action model for humanoid robots with dual-system design. These systems can perform tasks that seemed impossible just a few years ago. They handle open-vocabulary instructions (not just fixed commands), remain robust to novel objects and unstructured environments, adapt to new tasks from roughly 100 demos, and transfer capabilities from two-armed systems to humanoid robots. Robotics is the ultimate test because it requires AI to handle real-world complexity — not just generating pixels, but manipulating physical objects under real constraints such as geometry, friction, and randomness. The model’s internal representations must align with physical reality. ## What Lies Ahead in 2026? Predicting AI is getting harder, but here’s what I’m seeing based on current trends. Looking ahead to 2026, several key trends are likely to shape AI progress: * **Reasoning benchmarks** will diversify beyond current limitations. [ARC-AGI-3](https://three.arcprize.org/) will introduce interactive elements, while new benchmarks emerge in physics simulations, advanced mathematics, and agentic collaboration involving multiple AI systems working together. * **Open-ended learning** will expand into multi-agent environments where general agents compete with specialized models, increasing the importance of collaboration, competition, and communication as we move toward AI societies. * **Memory architectures** will advance through hierarchical memory stacks (hot-warm-cold), nested learning, and continuous learning without catastrophic forgetting, alongside more realistic benchmarks with practical constraints. * **Robotics** will see growing competition between general-purpose humanoid agents and specialized hardware, leading to a foundational model zoo for on-device AI that includes smaller efficient models, specialized hardware, and hybrid systems. * **Cloud AI ecosystems** will explode with reasoning-as-a-service, memory-as-a-service, and world-model-as-a-service components that can be composed like building blocks. * **Process innovation** will continue to outpace raw scale, with more test-time training, refinement loops, evolutionary problem-solving, and smarter inference-time compute. * Finally, a **benchmark revolution** will bring more realistic evaluation beyond static puzzles, including interactive and dynamic assessments, collaboration benchmarks, and cross-domain testing across physics, math, and logic. ## Beyond 2026 With all these advances, we are heading toward an era of proactive digital assistants. They won’t be passive anymore — they’ll advocate, plan, and act on our behalf. People who invest in better assistants are likely to become more efficient and more competitive, as their systems operate for them in the digital world. These assistants will be connected to the internet, improving themselves with real-time information and iteratively refining their outputs and actions. Access to validated information and high-quality data will increasingly become a decisive advantage. The convergence of abstract reasoning (ARC), world modeling (SIMA), memory mechanisms, and physical embodiment points toward AGI systems that can understand instructions, maintain world models, plan long-term, learn continuously, and act in the real world. We also need to identify the right interfaces between humans and AI. Not everyone wants neural implants, and people are still hesitant to communicate with machines purely through natural language — perhaps because machines have only recently begun to understand us. Either way, this transition will take time. Careful experimentation will be needed to ensure these technologies genuinely improve efficiency and quality of life. Here’s to 2026 — potentially the beginning of a full-scale technological AI revolution. If handled well, it is likely to bring more good than harm, and we all have a role to play in shaping it. ### Further Reading * [ARC Prize](https://arcprize.org/) — The competition and benchmarks * [ARC-AGI-2 Paper](https://arxiv.org/pdf/2505.11831) — Chollet et al., 2025 * [HRM Paper](https://arxiv.org/abs/2506.21734) — Hierarchical Reasoning Model * [SIMA 2](https://deepmind.google/technologies/sima/) — DeepMind’s Scalable Instructable Multiworld Agent * [Scaling Agent Systems](https://arxiv.org/pdf/2512.08296) — When multi-agent helps vs. hurts * [Memory Taxonomy Survey](https://arxiv.org/pdf/2504.15965) — 3D-8Q taxonomy * [Nested Learning](https://abehrouz.github.io/files/NL.pdf) — Google Research, NeurIPS 2025 * [Gemini Robotics](https://arxiv.org/abs/2503.20020) — Vision-Language-Action for robots * [GR00T N1](https://arxiv.org/pdf/2503.14734) — NVIDIA’s humanoid robot model * [CritPT Benchmark](https://artificialanalysis.ai/evaluations/critpt) — Critical thinking evaluation * [GPT-5.2](https://openai.com/index/introducing-gpt-5-2/) — OpenAI’s latest model * [TRM Paper](https://arxiv.org/abs/2510.04871) — Recursive reasoning with tiny networks Recommendation Engine Personalization ## Next Articles [![](https://www.recombee.com/img/blog/product-highlights-from-2025.png)](https://www.recombee.com/blog/product-highlights-from-2025) ### [Product Highlights from 2025](https://www.recombee.com/blog/product-highlights-from-2025) In 2025, we focused on making advanced personalization easier to implement, scale, and maintain across products and platforms. Here’s a look at the key product updates we released last year. ![](https://www.recombee.com/img/blog/authors/ondrejfiedler.png) Ondrej Fiedler Jan 08, 2026 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/recombee-partners-with-the-telegraph-to-deliver-ai-driven-personalisation-to-millions-of-readers.png)](https://www.recombee.com/blog/recombee-partners-with-the-telegraph-to-deliver-ai-driven-personalisation-to-millions-of-readers) ### [Recombee Partners with The Telegraph to Deliver AI-Driven Personalisation to Millions of Readers](https://www.recombee.com/blog/recombee-partners-with-the-telegraph-to-deliver-ai-driven-personalisation-to-millions-of-readers) Prague, 11th September 2025 – Recombee, a leading AI-powered recommendation platform, has announced a strategic partnership with The Telegraph... ![](https://www.recombee.com/img/blog/authors/recombee.png) Recombee Oct 02, 2025 Company News [![](https://www.recombee.com/img/blog/modern-recommender-systems-part-3-objectives.png)](https://www.recombee.com/blog/modern-recommender-systems-part-3-objectives) ### [Modern Recommender Systems - Part 3: Objectives](https://www.recombee.com/blog/modern-recommender-systems-part-3-objectives) Learning objectives of recommender systems and personalized search. ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Sep 03, 2025 Recommendation Engine Personalization --- # Are You Here to Stay? Unraveling the Dynamics of Stable and Curious Audiences in Web Systems > Source: https://www.recombee.com/blog/are-you-here-to-stay-unraveling-the-dynamics-of-stable-and-curious-audiences-in-web-systems > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # Are You Here to Stay? Unraveling the Dynamics of Stable and Curious Audiences in Web Systems ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Mar 03, 2025 Why do influencers frequently request their subscribers to enable all notifications for their channels? This practice stems from their awareness that not all subscribers are regular viewers of their content. While the number of subscriptions is important, the true determinant of sustained interest lies in more than just subscriber count. The popularity dynamics of online content are influenced by various internal and external factors, including content quality, metadata, age, **recommendation algorithms,** keyword rankings, promotions, and social network effects. For instance, some videos maintain significant viewership for hundreds of weeks after their initial posting, while others experience the majority of their views within the first few hours. This indicates a concentrated viewership during the early stages. ![](https://www.recombee.com/img/blog/are-you-here-to-stay-unraveling-the-dynamics-of-stable-and-curious-audiences-in-web-systems/main.png) Understanding the popularity of items in online systems, especially in recommendation scenarios, is crucial for the success of most of the online businesses. However, accurately predicting the enduring appeal of online content is challenging due to the distinct patterns exhibited in the popularity dynamics of online items. But why? These dynamics often involve one or more peaks of popularity bursts that intermingle with the regular and stable audience of the content. To address this challenge, one must focus on differentiating two types of audiences: the curious and the loyal. The curious audience is attracted by external and viral events, such as gossip, while the loyal audience represents stable viewership. Empirical evidence reveals that content like keyword-discovered videos, popular TV episodes, and music videos maintain steady popularity over time, dominated by loyal audiences. In contrast, news, sports, and movie content often undergo rapid popularity surges followed by quick declines, mostly due to temporally limited events (e.g., breaking news). In these cases, the curious audience tends to prevail over the loyal audience. To better understand how loyal and curious audiences can help to improve recommendation systems key performance indicators (KPIs), we recommend our previous blog post [here](https://www.recombee.com/blog/bandit-models-exploiting-popularity-and-curiosity-to-recommend-trending-content). The primary challenge in distinguishing between stable and curious audiences arises from the lack of individual labels that differentiate these viewer types. Instead, we typically only have the total number of viewers for an observed series of events, which is a combination of both audience types. Disentangling these two audiences poses difficulties, particularly because during viral events, curious users tend to dominate the channel's activity, leading to a burst in the overall series of events. These bursts can become so prominent that they obscure the presence of stable users during these periods. ![](https://www.recombee.com/img/blog/are-you-here-to-stay-unraveling-the-dynamics-of-stable-and-curious-audiences-in-web-systems/01.png) Another significant challenge arises from the potential for the stable audience to modify its typical behavior in response to bursts or external events. For instance, the unexpected death of Michael Jackson in June 2009 triggered a surge in media and web activity, leading to increased music sales, video views, and posts discussing him. During this period, both existing and new fans engaged with Jackson's work, transforming him into an enduring musical icon. This behavior is illustrated in the left-hand side of the figure above, where the blue line represents the cumulative web activity associated with Michael Jackson's YouTube searches by an American audience. Initially, it displayed a relatively constant growth rate until his death (vertical black line), followed by a sudden spike in interest. Over time, the blue line returned to a consistent growth rate. We will illustrate how to distinguish between two types of web activities during such events: (1) regular stable activity (yellow line) and (2) activity driven by unexpected events generated by the curious audience (green line). Notably, the yellow curve changed its slope after Jackson's unexpected death, marking a significant transition event that not only led to a short-term burst of activity but also consistently altered the stable audience. The revival of his songs, tribute notes, and the younger generation's discovery of Jackson's work contributed to a sustained increase in interaction. Conversely, the end of Barack Obama's presidential term (right-hand side of the figure) resulted in reduced political activity and mentions. Thus, understanding the change in popularity can be beneficial for recommender systems that can anticipate and respond to the current situation of an item. By accurately distinguishing between stable and curious audiences, these systems can better tailor recommendations, predict future trends, and optimize content delivery to maintain engagement and maximize viewer satisfaction. Fast reaction is critical, for instance, in the news industry, where audiences can quickly turn to competitors' websites for information, and losing the opportunity to fully leverage trending content could be detrimental. However, many events occur frequently and may only interest niche audiences (e.g., the injury of a volleyball player). Thus, the solution is not to classify every such event as "breaking news" and broadcast it to everyone, as this can negatively impact the loyalty of users who are not interested. Therefore, it is important to identify "curious" users who might be interested in these niche events, allowing for more personalized and relevant content delivery. But how do we detect the intensity of both audiences if these systems are often only aware of the temporal dynamics of the items throughout the observed interactions? Notably, events like Michael Jackson's death and the end of Obama's term are observed through interactions, and if the system is not aware that such events can happen, it cannot harness the potential of a burst. This could impact revenue, retention, and other KPIs for recommenders. To address this and other questions, our research proposes a novel methodology called BPoP (Burst-induced Poisson Process). BPoP is a mix of point processes that form a statistical framework to learn and infer about multiple series of events. Our research was accepted for publication in ACM The Web Conference, one of the most prestigious data mining/machine learning conferences and the premier event for web research. I had the pleasure of contributing to this research with top-class researchers from prestigious institutions located in four countries across three continents. ![](https://www.recombee.com/img/blog/are-you-here-to-stay-unraveling-the-dynamics-of-stable-and-curious-audiences-in-web-systems/02.png) More formally, our model is able to flexibly incorporate dependencies between two hidden and underlying point processes involving the stable fanbase and the curious audience. Therefore, BPoP is capable of disentangling the observed audience (interactions, represented by blue dots) into two different stochastic point processes: one representing the curious audience (green) and another one the stable audience (yellow dots). The transitions of the stable audience (white dots), which we never observe, are influenced by the bursts observed in the curious audience. For that, we assume a generative process where we can define the parameters of the models and treat the labels of the observed points as latent variables, enabling recovery of the parameters with the traditional EM (Expectation Maximization) Algorithm. For detailed derivation, we refer to our paper, which may be more interesting to a technical audience. We show that BPoP mimics the bursts of events seen in real data and is also able to efficiently capture the time-varying background rates that realistically represent the fan base. If you like our research, are interested in understanding more, have questions, or see opportunities to collaborate with us, please feel free to contact us. We look forward to discussing our research with you. You can also cite our paper as follows if any of this content is useful. Understanding, explaining, and measuring audiences of items - broadly known as item popularity - is key to producing more accurate recommendation algorithms. ## References Rodrigo Alves, Antoine Ledent, Renato Assunção, Pedro Vaz-De-Melo, and Marius Kloft. "Unraveling the Dynamics of Stable and Curious Audiences in Web Systems." In Proceedings of the ACM on Web Conference 2024, pp. 2464-2475\. 2024. Recommendation Engine Personalization ## Next Articles [![](https://www.recombee.com/img/blog/build-vs-buy-deciding-the-best-approach-for-your-recommender-system.png)](https://www.recombee.com/blog/build-vs-buy-deciding-the-best-approach-for-your-recommender-system) ### [Build vs. Buy: Deciding the Best Approach for Your Recommender System](https://www.recombee.com/blog/build-vs-buy-deciding-the-best-approach-for-your-recommender-system) When it comes to deciding between buying a recommender system and building one from scratch, the choice isn’t always straightforward. Both options come with their own set of pros and cons... ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova & Ondrej Fiedler Mar 14, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/recombee-research-2024.png)](https://www.recombee.com/blog/recombee-research-2024) ### [Recombee Research 2024](https://www.recombee.com/blog/recombee-research-2024) Recombee has always been deeply connected to academia, with four of our six co-founders holding Ph.D. degrees. Over the years, our investment in research has grown alongside our company and the increasing demand for advanced features in the market. ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Feb 23, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/key-trends-in-news-and-media-for-2025-and-the-role-of-ai-powered-solutions.png)](https://www.recombee.com/blog/key-trends-in-news-and-media-for-2025-and-the-role-of-ai-powered-solutions) ### [Key Trends in News & Media for 2025 and the Role of AI-Powered Solutions](https://www.recombee.com/blog/key-trends-in-news-and-media-for-2025-and-the-role-of-ai-powered-solutions) The way news is produced, delivered, and consumed keeps shifting, and as we move through 2025, a few big changes have become impossible to ignore. ![](https://www.recombee.com/img/blog/authors/annapetr.png) Gabriela Takacova & Violeta Milarova Feb 20, 2025 Personalization --- # Non-stationary Multi-Armed Bandits | Recombee Recommender > Source: https://www.recombee.com/blog/bandit-models-exploiting-popularity-and-curiosity-to-recommend-trending-content > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # Bandit Models: Exploiting Popularity and Curiosity to Recommend Trending Content ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Oct 16, 2022 ![](https://www.recombee.com/img/blog/bandit-models-exploiting-popularity-and-curiosity-to-recommend-trending-content/main.png) Humans are inherently curious. In fact, curiosity is linked to the evolution of humankind. For instance, according to famous historian Yuval Noah Harari in his bestseller book "Sapiens", our language skills evolved as a way of gossiping: "It is not enough for individual men and women to know the whereabouts of lions and bison. It's much more important for them to know who in their band hates whom, who is sleeping with whom, who is honest, and who is a cheat". In this regard, one might wonder how this trait we all possess can help us develop more accurate recommendations? And how do curiosity and popularity interact in the context of recommender systems? In this blog post, we will answer these and other questions and explain how we at Recombee frequently interpret the environment in which the recommender systems are exposed to effectively respond to changes in popularity. Let us start with an example of one of the most fundamental recommender system problems: cold-start recommendation (CSR), where we aim to profile and recommend items to **new users.** This problem is especially challenging due to the lack of information about the preferences and behavior of the users. For example, when a new user visits a video streamer website, an RS needs to choose a video solely based on user-agnostic information (e.g., the average times previous users watched a video). After the recommendation, the RS observes whether the new user watches the recommended video. The website aims to catch the users' attention and maximize the total number of interacted videos. Therefore, providing effective CSR here requires identifying the \`trending' items most popular among the website's audience. ## Multi-Armed Bandit Algorithms for Recommender Systems ![](https://www.recombee.com/img/blog/bandit-models-exploiting-popularity-and-curiosity-to-recommend-trending-content/1.png) A popular option to solve this problem is to model CSR as a multi-armed bandit (MAB) problem. In MABs, at each trial, a gambler selects an arm to pull and observes the reward. Throughout the trial history, the gambler improves his policy to maximize the reward at the end. ![](https://www.recombee.com/img/blog/bandit-models-exploiting-popularity-and-curiosity-to-recommend-trending-content/2.png) Recommender systems can be modeled as MABs. In our video streamer example, at each recommendation requisition, the RS selects an item (arm) to show to the user and observes if he watches or not the item (observes the reward). Throughout the recommendation history, an algorithm improves the strategy to maximize the number of watched videos. ![](https://www.recombee.com/img/blog/bandit-models-exploiting-popularity-and-curiosity-to-recommend-trending-content/3.png) Although model RSs as MABs can be efficient and accurate, their standard setting assumes that there exists a fixed most popular item overall time. However, this is inadequate in practice: assuming that the items' popularity is not changing over time is highly unrealistic while considering practical applications. We, therefore, can model CSR as a non-stationary MAB problem, where the most popular item changes over time. Interestingly, we continuously face the classic exploration/exploitation dilemma known from reinforcement learning: the RS must maintain a balance between recommending a classic popular item and recommending the object of the current viral fad. To illustrate this dilemma, let us consider the following real-world example. Suppose that an RS must select among videos of two artists: the South Korean singer Psy and the British singer David Bowie. ![](https://www.recombee.com/img/blog/bandit-models-exploiting-popularity-and-curiosity-to-recommend-trending-content/4.png) The grey curves in the above graphs show the cumulative percentage of youtube activity ([only USA audience](https://trends.google.com/trends/)) associated with both artists from 2008 to 2020 (for details regarding data collection and preprocessing, see the referenced article). Most of the time, the rate of growth of the system activity is approximately constant. However, this linearity is sometimes broken by sudden bursts of events highlighted by the vertical red and blue lines. The first spike (vertical red line) matches with the _Gangnam Style_ release. The hit had an unprecedented explosion of popularity and its music video “broke” the YouTube view counter's limit. The second burst of events (vertical blue line) coincides with David Bowie's unexpected death. This unfortunate exogenous event triggered the audience's curiosity and substantially increased his video views and reactions in posts and comments. In this frenzy, old fans recalled his talent by listening to his songs again, while the new generation discovered a treasure and quickly transformed into a musical and show business icon. We then explain the variations in the users' activity by the existence of two types of audiences (disentangled in the above image, third graph): the **loyal audience** and the **curious audience.** The loyal audience (green curve, third graph) is constituted by fans who assiduously follow the topic. In contrast, the curious audience (yellow curve, third graph) only turned their attention to the topic due to an extraordinary event. Thus, the environmental context in which the RS must make decisions alternates between calm periods (where the users' behavior is driven by the loyal audience), and disruptive or _bursty_ periods (where the curious audience is driving sudden bursts of interest in certain topics). In our real-life example, during stable periods, the ratio between the singers' popularity is stable, and Bowie is consistently more popular than Psy (second graph). On the other hand, during the disruptive period dominated by _curious_ behavior, the relative popularity between Psy and Bowie changed drastically. Note that, during the recommendation event, the RS only observes the grey curve: it does not know to whose audience it belongs, nor which artist's content the user wishes to interact with. Hence, such information must be inferred. ## Curiosity-Aware MABs In our example, we can clearly see how user curiosity can affect the item's popularity. The next question is how can we recognize the alternation between audiences in order to anticipate and identify new trends? For that, we will present the BMAB (Burst-induced MAB) algorithm. The core principle is to use two instances of the classic [Thompson sampling (TS) algorithm](https://web.stanford.edu/~bvr/pubs/TS_Tutorial.pdf): one for loyal behavior and the other for curious behavior. To support our explanation we will illustrate our algorithm execution for the time series displayed in our Psy/Bowie example. In the beginning, we (1) initialize all entries of TS priors as uniform distributions. This indicates that the algorithm does not initially know which item (Psy or Bowie) is the most popular and must learn through user interactions. In order to maximize the reward, the BMAB algorithm aims to learn (by updating its priors) the reward distributions of _both states_ with enough confidence to select the best arm at the event time. Therefore, at each recommendation event, the algorithm needs (2) to detect the state of the system. In this blog post, we assume that an oracle is available to provide an estimate of the state at each recommendation requisition. In practice, the role of the oracle can be assumed by [our realistic state detector.](https://dl.acm.org/doi/10.1145/3460231.3474250) In the next step, we (3) sample an item according to the prior distribution corresponding to the **current active state.** Finally, we (4) observe the reward of the selected arm and update the priors of the active state. In the video below, we compare the execution of our algorithm with the standard MAB model. Observe that, after the release of the single "Gangnam Style", our model activates the curious mode and promptly identifies Psy as the most popular current item. In contrast, the standard MAB model takes a considerable amount of time to detect such a shift. Moreover, in this instance, our model assumes that the transition from one state to another (loyal to curious or curious to loyal) comes with its own reward distribution. Accordingly, we aim to treat each state segment as a separate MAB problem, resetting the Thompson priors at the beginning of each segment whilst keeping a global count for the periods where the loyal audience dominates. However, due to the uncertainty inherent in the state prediction method, we engineer a soft transition procedure: whenever a state appears to be ending, the priors corresponding to the state are gradually forgotten rather than discarded immediately. This behavior can be observed in the graphs at the bottom of the video. As you can see, monitoring the audience to quickly recognize changes in popularity may be crucial to identifying trending items with the potential to significantly enhance recommendation KPIs. Curiosity-aware MABs are one of the several Recombee toolkits that deliver improved content recommendations. If you enjoyed our work, please cite us, provide us with comments, or contact our research team for further technical details. ### Reference This blog article is adapted from the contribution provided below. For details of data, please check the paper. Please include appropriate citations for any use of this content, whether in whole or in part. Rodrigo Alves, Antoine Ledent, and Marius Kloft. 2021. [Burst-induced Multi-Armed Bandit for Learning Recommendation.](https://dl.acm.org/doi/10.1145/3460231.3474250) In Fifteenth ACM Conference on Recommender Systems (RecSys '21). Association for Computing Machinery, New York, NY, USA, 292–301\. https://doi.org/10.1145/3460231.3474250 Recommendation Engine ## Next Articles [![](https://www.recombee.com/img/blog/recombee-item-segmentations.png)](https://www.recombee.com/blog/recombee-item-segmentations) ### [Recombee Item Segmentations](https://www.recombee.com/blog/recombee-item-segmentations) Item Segmentations are Recombee's original and elegant solution to various advanced tasks related to hierarchical and relational data. The feature provides a flexible way to group items (products or pieces of content) into segments... ![](https://www.recombee.com/img/blog/authors/tomasrehorek.png) Tomas Rehorek Jan 11, 2023 New Features Recommendation Engine [![](https://www.recombee.com/img/blog/keeping-up-with-digital-media-convergence.png)](https://www.recombee.com/blog/keeping-up-with-digital-media-convergence) ### [Keeping Up With Digital Media Convergence](https://www.recombee.com/blog/keeping-up-with-digital-media-convergence) At Recombee, we felt the transition within the media industry accelerated by the pandemic. OTT and CTV consumption ballooned at a significant rate. ![](https://www.recombee.com/img/blog/authors/annadolezelova.png) Anna Dolezelova Oct 8, 2022 Personalization [![](https://www.recombee.com/img/blog/recombee-in-e-mail-marketing-a-partner-success-story-with-ryzeo.png)](https://www.recombee.com/blog/recombee-in-e-mail-marketing-a-partner-success-story-with-ryzeo) ### [Recombee in E-mail Marketing: A Partner Success Story with Ryzeo](https://www.recombee.com/blog/recombee-in-e-mail-marketing-a-partner-success-story-with-ryzeo) Do you feel there is a potential to increase your success with customers through an efficient recommender engine? You're highly likely right. Adding a recommender service to your emailing campaigns gives each client tailored product recommendations in all of their emails. ![](https://www.recombee.com/img/blog/authors/russellmiller.png) Russell Miller (Ryzeo) Oct 5, 2022 Partnerships --- # Build vs. Buy: Deciding the Best Approach for Your Recommender System > Source: https://www.recombee.com/blog/build-vs-buy-deciding-the-best-approach-for-your-recommender-system > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # Build vs. Buy: Deciding the Best Approach for Your Recommender System ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova & Ondrej Fiedler Mar 14, 2025 ![](https://www.recombee.com/img/blog/build-vs-buy-deciding-the-best-approach-for-your-recommender-system/main.png) When it comes to deciding between buying a recommender system and building one from scratch, the choice isn’t always straightforward. Both options come with their own set of pros and cons, and the right answer depends heavily on your team, resources, and long-term goals. But if you're weighing the decision, it helps to break things down into a few key factors: time, expertise, cost, and control. ## The Case for Building In-House ![](https://www.recombee.com/img/blog/build-vs-buy-deciding-the-best-approach-for-your-recommender-system/01.png) For some businesses, the main appeal of an in-house recommender system is full control. When you build your own, you decide everything, from the algorithms to the data flow to the key metrics. This allows for deep alignment with business goals, ensuring recommendations fit your exact needs rather than relying on an off-the-shelf solution. If your company has highly specific requirements, whether it’s complex personalization logic, proprietary data models, or strict compliance needs, building in-house might seem like the best approach. An internal system also offers more room for experimentation. While third-party platforms allow some customization, they often have limits on modifying core algorithms or integrating niche data sources. With an in-house solution, your team can test different recommendation strategies freely, optimize for business-specific KPIs, and refine models without external constraints. That said, building a recommendation system from scratch is a massive investment, not just in development but in ongoing maintenance, infrastructure, and talent. It requires a team with specialized expertise, and hiring top-tier machine learning engineers and data scientists is both difficult and expensive. Only companies with significant resources can realistically afford to maintain a high-performing, custom-built system in the long run. For businesses where recommendations are central to engagement, conversions, or customer retention, the ability to customize every aspect of the system can be a real advantage. But the cost and complexity mean it’s an option only for those with the budget, the talent, and the long-term commitment to make it work. ## Challenges of Building In-House That level of control and customization can be a major advantage, but it also comes with significant complexity. The reality is that maintaining and refining an in-house system is not a one-time effort but an ongoing process that demands continuous adjustments and technical oversight. Even small changes, such as boosting specific content or fine-tuning ranking criteria, often require direct involvement from data scientists or developers. If these adjustments necessitate model retraining, performance evaluations, or infrastructure updates, it can take days or even weeks before the update goes live. In contrast, SaaS solutions often provide user-friendly web interfaces where product teams can tweak recommendations instantly without needing technical skills. This allows for more agile experimentation and faster iteration, reducing the dependency on specialized engineering resources for day-to-day adjustments. Beyond the core recommendation model, there is a hidden layer of complexity that many companies overlook. Running a recommender system in production requires: * Monitoring and alerting for uptime, latency, and recommendation quality * 24/7 incident response teams to handle system failures * A/B testing and quality assurance tools to evaluate performance * Scalability planning and capabilities as your user base grows While an in-house team could develop custom tools to manage these needs, it adds significant cost and development time. The reality is that maintaining a high-quality recommendation system is not a one-time build, but a continuous commitment that requires a dedicated team of data scientists, engineers, and operational support. And then, of course, there is the cost. Infrastructure, salaries, ongoing R&D; it all adds up quickly. What initially seems like a cheaper, more flexible solution can easily become more expensive and time-consuming than expected. ## Why Buying Often Makes Sense ![](https://www.recombee.com/img/blog/build-vs-buy-deciding-the-best-approach-for-your-recommender-system/02.png) On the other side of the coin, buying lets you skip many of the headaches that come with building. Instead of spending months (or longer) developing your own solution, you can get started almost immediately by leveraging an existing system. Off-the-shelf systems are often built with scalability in mind. They’ve been tested across industries and user bases, meaning they’re designed to handle large amounts of data and adapt as your business grows. You also get the benefit of ongoing improvements since providers are constantly refining their technology to stay competitive. For businesses that lack the time or resources to hire data scientists or stay up to date with the latest algorithmic trends, outsourcing is a practical option. And just because you’re outsourcing doesn’t mean you have to sacrifice control or flexibility. Many platforms today allow you to fine-tune their systems to meet your needs. For example, Recombee’s advanced filtering options let you control which content gets recommended, whether that means showing only available items or filtering out specific content based on custom business rules. You can boost certain kinds of content, adjust for diversity in recommendations, or bias the system toward newer content. It’s highly flexible, meaning you can adapt it to reflect your priorities without needing to reinvent the wheel. Plus, buying unlocks advanced personalization capabilities without the burden of managing them in-house. ## Dynamic Personalization Without the Hassle ![](https://www.recombee.com/img/blog/build-vs-buy-deciding-the-best-approach-for-your-recommender-system/03.png) A major advantage of buying a recommendation system is achieving deep personalization without the complexity of building and maintaining it yourself. Modern platforms make it easier to adjust recommendations for different business needs. Take [Scenarios](https://www.recombee.com/features/full-control-with-scenario-settings), for example. These are distinct recommendation contexts designed for specific use cases. Whether it’s personalized content for a news homepage, "similar items" on an e-commerce product page, movie recommendations for a streaming platform, or suggested courses on an e-learning site, each Scenario operates with its own logic and rules, ensuring the recommendations align with the user’s intent. Once Scenarios are defined, you can further refine recommendations using user and item data. With Recombee, boosters and filters help adjust which content is prioritized within each Scenario, whether it’s based on region, language, user preferences, or subscription tier. This way, recommendations stay relevant without constant manual updates. The system also adapts in real-time. If a user suddenly interacts with a new category, shows interest in trending topics, or searches for specific products, the recommendations update dynamically. Instead of relying on static, pre-set suggestions, the system continuously learns and evolves, keeping recommendations relevant and engaging. ## The Scalability Advantage Another key benefit of buying is scalability. Whether you’re a small startup just starting out or a larger company dealing with millions of users, Recombee’s horizontally scalable infrastructure is capable of handling over a billion recommendations daily and processing more than 30,000 recommendations per second. It supports extensive catalogs with tens of millions of items and ensures low latency through multiple data centers strategically located across the globe. There’s no need to worry about whether your infrastructure can handle traffic spikes or growth. It’s all handled for you. ## Weighing the Trade-offs The biggest downside of an off-the-shelf solution is often the perceived lack of control. It can feel risky to hand over such an important part of your customer experience to an external provider. However, the level of customization and adaptability available in modern recommendation systems means this concern is less valid than it once was. With tools to fine-tune everything from the algorithms to the user experience, these systems ensure the recommendations still feel uniquely yours and align closely with your business goals. On the flip side, building in-house gives you full control, but at a high cost, both in terms of money and time. Unless personalization is a core competency for your business, it’s worth asking whether the effort is truly worth it. ## A Balanced Perspective Ultimately, the build vs. buy decision comes down to your company’s strengths. If recommendations are central to your business, you have a sizable budget, and you want full control over the system, developing in-house could be a logical choice. For most companies, however, outsourcing provides a faster, more cost-effective way to deliver high-quality recommendations. And with platforms like Recombee, you get the best of both worlds. Expertly designed algorithms with the flexibility to adapt them to your needs. You’re not just adopting a tool, you’re gaining a customizable, scalable, and future-proof solution designed to grow with your business while ensuring the personalized experience your users expect. Recommendation Engine Personalization ## Next Articles [![](https://www.recombee.com/img/blog/2025-sneak-peek.png)](https://www.recombee.com/blog/2025-sneak-peek) ### [2025 Sneak Peek](https://www.recombee.com/blog/2025-sneak-peek) This year is already off to an exciting start, and we’re rolling out new tools to improve efficiency and optimize recommendations. Here’s what’s available and what’s coming next. ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Mar 19, 2025 Recommendation Engine New Features [![](https://www.recombee.com/img/blog/are-you-here-to-stay-unraveling-the-dynamics-of-stable-and-curious-audiences-in-web-systems.png)](https://www.recombee.com/blog/are-you-here-to-stay-unraveling-the-dynamics-of-stable-and-curious-audiences-in-web-systems) ### [Are You Here to Stay? Unraveling the Dynamics of Stable and Curious Audiences in Web Systems](https://www.recombee.com/blog/are-you-here-to-stay-unraveling-the-dynamics-of-stable-and-curious-audiences-in-web-systems) Why do influencers frequently request their subscribers to enable all notifications for their channels? This practice stems from their awareness that not all subscribers are regular... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Mar 03, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/recombee-research-2024.png)](https://www.recombee.com/blog/recombee-research-2024) ### [Recombee Research 2024](https://www.recombee.com/blog/recombee-research-2024) Recombee has always been deeply connected to academia, with four of our six co-founders holding Ph.D. degrees. Over the years, our investment in research has grown alongside our company and the increasing demand for advanced features in the market. ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Feb 23, 2025 Recommendation Engine Personalization --- # How Regionalization-Based Recommendations Can Improve Your Operations > Source: https://www.recombee.com/blog/how-regionalization-based-recommendations-can-improve-your-operations > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # How Regionalization-Based Recommendations Can Improve Your Operations ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Jul 18, 2025 ![](https://www.recombee.com/img/blog/how-regionalization-based-recommendations-can-improve-your-operations/main.png) From ancient trade routes to modern urban planning, geography has consistently shaped human decisions and opportunities. Today, in the world of online business and personalized recommendations, geography remains equally influential, even though the effects aren’t always immediately obvious. At Recombee Research, we believe in the potential of location-based recommendation methods and prioritize their development. In this post, we will discuss how identifying and understanding **geographic regions** using your recommender system can significantly enhance your operational strategies. Whether you’re running a streaming platform, curating local events or tailoring a marketing campaign, location-tailored content can be a game-changer. Audiences in different areas often exhibit distinct tastes (what captures attention in one city might not be the same in another) so integrating regional insights into your recommendation engine can boost relevance and engagement. For instance, by suggesting recipes that resonate with local culinary traditions, region-aware recommendations let you meet users (customers) where they are (literally) and offer them what they truly want. Note that, however, if the recommender isn’t calibrated correctly, globally popular items can drown out local content, depriving regional providers of the visibility they deserve and users of the locally relevant choices they expect. Location-tailored content is crucial for many applications, including streaming services Our recent research introduces a practical and efficient method for incorporating geographical data into recommendations. Using regionalization (spatial clustering) techniques, this approach effectively uncovers meaningful geographic patterns in user preferences and behaviors. These insights can then be applied to personalize content curation, recommend local events that truly resonate with each community, tailor marketing campaigns to regional trends, and deepen the overall understanding of how geography shapes user engagement. ## Why Does Location Matter? Consider a streaming platform looking to refine its content strategy. Regionalized recommendation analysis may reveal that one micro-region is embracing short-form cooking videos, something not yet picked up in global trends, while another area shows growing enthusiasm for interactive live streams and e-sports content. By detecting these nascent trends in your recommender system’s interaction logs, you can, for instance, boost engagement by offering the most relevant titles and deliver a more personalized experience that feels tailor-made for local audiences. The same principle applies, for instance, when curating and promoting local events. Geographic recommendation techniques can uncover neighborhoods where demand for indie music nights, pop-up art shows, or wellness workshops is on the rise. By surfacing these findings to the right user clusters you drive higher attendance, strengthen community bonds, and give event organizers the insights they need to allocate, for instance, marketing budgets most effectively. ## Common Challenges in Identifying and Understanding Regions in Recommenders Recommendation systems often face two main challenges: * **Sparse Data:** In many cases, there is insufficient data from individual regions to clearly identify user preferences. * **Effective Region Grouping:** Unsupervised learning methods (such as clustering) are inherently difficult to evaluate. It can be challenging to define geographical clusters that truly reflect customer behavior, rather than relying on arbitrary or misleading divisions. ## Our Solution ![](https://www.recombee.com/img/blog/how-regionalization-based-recommendations-can-improve-your-operations/1.png) Imagine a city where restaurants in different neighborhoods have distinct tastes: e.g., downtown users favor trendy vegan spots, while suburban customers lean toward hearty, family-style restaurants. Because these groups rarely overlap in their restaurant interactions, there’s no direct data linking their preferences, and many smaller or emerging neighborhoods simply don’t generate enough feedback to learn reliable local signals. Our solution tackles this by identifying **macro-regions**, spatial clusters of areas with similar latent user affinities, and pooling their interaction data. By aggregating sparse signals across each macro-region, even the least active zones inherit robust, region-aware insights, yielding far more reliable, personalized recommendations for every user. Our method was engineered to exactly explore this scenario. First we use **Inductive Matrix Factorization (IMF)** to combine two sources of data: 1. **Customer interaction data:** the matrix of what people, for instance, have rated, ordered, or clicked on. 2. **Geographic information:** a mapping from items (e.g. restaurants) to **regions** (e.g. neighborhoods or delivery zones). At its core, a recommender system often relies on **matrix factorization**: learning two low-dimensional matrices (one for users, one for items) whose product approximates the user-item interaction matrix _X_. But when items are **sparsely rated**, the system struggles. To overcome it, we replace the learned item matrix with **known features** (like a one-hot encoding of the item’s region). That is, instead of learning all features of an item from data, we **inject side information**. The result is: _X ≈ M Y_ * _X_ is the observed interaction matrix (users × items), * _Y_ is the binary matrix that maps items to regions (the item geo-location), * _M_ is what we learn: a matrix of users × regions. Each row of _M_ represents **a user’s latent affinity for each region**. It's like building a profile of how likely a person is to enjoy products from any given area, even if they haven’t rated any items in that region directly. However, even in this case, some regions often remain too sparse to provide useful collaborative filtering signals. ## From Preferences to Patterns: Clustering Regions As discussed above, once we’ve built _M_, each **column represents a region**. This means we can: 1. **Visualize** how customer preferences vary geographically. 2. **Cluster** similar regions based on their latent profiles. In practice, this led us to discover surprising groupings, such as: * Inner-city districts with university campuses clustered with young, tech-savvy suburbs. * Affluent residential areas aligned with certain rural pockets (both preferring premium and organic goods). * Tourist-heavy zones and central business districts formed a “fast-paced” macro-region. ![](https://www.recombee.com/img/blog/how-regionalization-based-recommendations-can-improve-your-operations/2.png) Therefore, by clustering regions (using _M_) into these broader **macro-regions**, fast-paced urban cores, student-centric districts, premium-oriented suburbs, or tourist hubs, we gain two key advantages. First, we surface stronger, more reliable patterns of local preference by pooling data across similar areas. Second, we overcome sparsity in any one neighborhood: even if a particular zone has few direct interactions, its membership in a well-defined macro-region lets us borrow insights from fellow regions with richer feedback. The result is a recommendation engine that delivers highly relevant, geographically aware suggestions for every user, no matter where they live. In technical terms, we construct an updated _YMacro_ matrix that encodes each item by its macro-region membership rather than its original fine-grained zone. This macro-region-based _YMacro_ (items × macro-regions) both reduces dimensionality and amplifies sparse signals across similar areas. We then refactorize _X_ as _X ≈ Mmacro × Ymacro_ where _Mmacro_ (users × macro-regions) captures each user’s latent affinity profile at the macro level. By leveraging these aggregated region features, your recommender naturally delivers robust, geographically aware suggestions in both dense urban cores and sparsely populated zones. For more details on Inductive Matrix Factorization, see our [detailed post](https://www.recombee.com/blog/inductive-matrix-completion-how-to-improve-recommendations-for-cold-start-users-and-items-by-incorporating-their-attributes). For regionalization in recommenders, check out [our paper](https://dl.acm.org/doi/full/10.1145/3656641), and feel free to reach out for a deeper conversation. ## What Did We Find? Our extensive experiments using both simulated data and real-world datasets (from Google Locals dataset), particularly in the context of restaurants, yielded compelling results: * **Accurate Region Identification:** We successfully identified distinct geographic clusters, clearly separating items and behaviors critical for strategic product placement and marketing. * **Robust Predictions:** Recommendations based on regional clusters significantly outperformed traditional methods, demonstrating reliable accuracy even when data was limited or incomplete. * **Actionable Insights:** The method effectively revealed regional trends and preferences, enabling businesses to make informed decisions. ## …and what are the real-world benefits? Adopting our regionalization method can help your business: * **Enhance Customer Satisfaction:** Deliver faster and more relevant recommendations by aligning product offerings with local interests and upcoming trends. * **Personalize Content Curation:** Deliver region-tailored articles, videos, or playlists that resonate with local tastes, boosting click-through and engagement rates. * **Increase Event Attendance:** Surface concerts, meetups, and festivals most likely to appeal to each community, driving higher ticket sales and word-of-mouth. * **Enhance User Retention:** Keep audiences coming back by consistently suggesting experiences and products that feel made for their region. ## Final Thoughts Integrating geography into your recommendation and business strategy is both practical and impactful. Employing regionalization can not only improve operational efficiency but also provide valuable insights into customer preferences, helping businesses spot trends early and adapt accordingly. Recombee Research is committed to advancing geography-based recommendation methods, incorporating interaction data while exploring additional data sources. Stay updated by following us on social media; exciting developments are on the way! If you're interested in our research or wish to collaborate, please reach out to us. ## References Alves, Rodrigo. "Regionalization-based Collaborative Filtering: Harnessing Geographical Information in Recommenders." _ACM Transactions on Spatial Algorithms and Systems_ 10.2 (2024): 1-23. Recommendation Engine Personalization ## Next Articles [![](https://www.recombee.com/img/blog/no-code-search-widget-personalized-powerful-effortless.png)](https://www.recombee.com/blog/no-code-search-widget-personalized-powerful-effortless) ### [No-Code Search Widget: Personalized, Powerful, Effortless](https://www.recombee.com/blog/no-code-search-widget-personalized-powerful-effortless) At Recombee, we don't just excel at recommendations – we provide powerful full-text search capabilities too. Our Quick, No-Code Search Widget exemplifies this, offering a seamless... ![](https://www.recombee.com/img/blog/authors/ondrejfiedler.png) Ondrej Fiedler Jul 25, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/shield-the-universal-framework-making-ai-search-safer-for-everyone.png)](https://www.recombee.com/blog/shield-the-universal-framework-making-ai-search-safer-for-everyone) ### [SHIELD: The Universal Framework Making AI Search Safer for Everyone](https://www.recombee.com/blog/shield-the-universal-framework-making-ai-search-safer-for-everyone) Imagine searching for "glass tubing" and getting recommendations for drug manufacturing equipment. As AI-powered search becomes ubiquitous — from online marketplaces to social networks... ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Jun 24, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/making-recommendations-fairer-a-new-way-to-guarantee-exposure-for-all.png)](https://www.recombee.com/blog/making-recommendations-fairer-a-new-way-to-guarantee-exposure-for-all) ### [Making Recommendations Fairer: A New Way to Guarantee Exposure for All](https://www.recombee.com/blog/making-recommendations-fairer-a-new-way-to-guarantee-exposure-for-all) As recommender systems become more widespread across digital platforms, concerns around fairness are coming to the forefront. Standard relevance-based ranking techniques, while effective... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Apr 29, 2025 Recommendation Engine Personalization --- # Complete Personalization Experience | Recombee AI > Source: https://www.recombee.com/blog/innovative-personalization-features-for-2023 > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # Innovative Personalization Features for 2023 ![](https://www.recombee.com/img/blog/authors/gabrielatakacova.png) Gabriela Takacova Feb 17, 2023 ![](https://www.recombee.com/img/blog/innovative-personalization-features-for-2023/main.png) The digital world is changing; users' expectations for personalization are increasing, and our Recombee features are continuously improving. One of our focuses is to support our clients in providing the best possible user experiences on their sites and platforms. As such, we are excited to share our latest innovations in content organization, which will allow developers and product managers to more effortlessly and intuitively utilize our recommendation services. ## Big Announcement! The Release of Item Segmentations ![](https://www.recombee.com/img/blog/innovative-personalization-features-for-2023/01.png) The release of Item Segmentations is one of our biggest milestones, which significantly improves the recommendation engine usage and increases its personalization capabilities. ### What Can This Bring to Your Users? [Item segmentations](https://www.recombee.com/blog/recombee-item-segmentations) provide users with the ultimate tailored experience while also saving them time when searching for content they want to engage with. Users will no longer have to search through the site or scroll down to find their preferred category, but will instead see the content they are most likely to engage with right at the very top of the page. Item segmentation assists users by working dynamically and adapting to users' preferences, ensuring that the categories they see corresponds to their preferences. ### How Does It Work? Item segmentations work by grouping items (products or pieces of content in your catalog) based on shared conditions. It can not only classify items by authors, genres, vendors, and categories, but also by their combinations and further specific conditions such as year of production and price ranges. For example, a video-based platform could create tailored segments such as “French romantic comedies released between 2000-2010” or “American sci-fi movies produced in Hollywood”. In E-commerce, the products could be grouped into segments based on the intersection between vendors and price ranges. ### And Finally, What Is the Real Benefit of This Feature for Product Managers? Item Segmentations is a tool that will greatly benefit any product manager looking to improve their platform's personalization capacity. It can assist in optimizing the presentation of items from your catalog, increasing the impact of your recommendations, and driving increased engagement and conversions. Item Segmentations will be an important addition to your platform, allowing you to make more personalized and engaging recommendations to your users. Showing a personalized selection of the most relevant categories or achieving personalized re-ordering of the rows on the home page has never been easier! You can create segments specific to your platform directly in [Recombee Admin UI](https://admin.recombee.com/sign-up). For more info explore [Item Segmentations Docs](https://docs.recombee.com/segmentations). ## Additional Recombee Features for a Complete Personalization Experience There is more than one way to personalize, and every client’s needs for personalization may differ. This is also why every client can set up the engine to reach their specific KPIs, be it increased conversion rate, sales, click-through rate, or others. Curious about how to fully utilize the solution? Here is an overview of the most popular features to elevate the UX on your site. ### Personalized Search ![](https://www.recombee.com/img/blog/innovative-personalization-features-for-2023/02.png) Search 2.0 is a powerful feature enabling users to receive personalized search results based on their past interaction history. This high-tech feature enables your search engine to provide tailored search results and works with typos, synonyms, and now newly with segments. Our search works in more than 75 languages, including Arabic and Chinese. You can integrate [Recombee’s personalized search](https://docs.recombee.com/getting_started#getting-started-search) through API. ### Infinite Scroll ![](https://www.recombee.com/img/blog/innovative-personalization-features-for-2023/03.png) The next feature to help increase user engagement is the Infinite scroll, which enables your site to provide an endless stream of content or products as the user scrolls down the site or mobile app. This is a feature commonly known to be used by social media giants like Instagram or Facebook but it can be applied to nearly all domains. Learn more about integration at [Tips and Tricks in our documentation](https://docs.recombee.com/getting_started#tips-and-tricks). ### Newly Toned Logics ![](https://www.recombee.com/img/blog/innovative-personalization-features-for-2023/04.png) Each Recombee client is presented with the option to tailor the behavior of the recommendations to hit their desired KPI. The first step is to create a Scenario, which can be understood as the touch point where the recommendations are placed - the most popular options are having recommendations on the _homepage, detail page, in the search, emailing,_ or _push notifications._ As a second step of [creating the scenario](https://docs.recombee.com/scenarios), users can choose from a [selection of logics](https://docs.recombee.com/scenarios#logic), which tailors the recommendations’ behavior. For Media and VOD, see our [content recommendations set-up guide](https://www.recombee.com/handbook/download-content-recommendations-35kl362g0946239g0er) which shows the most commonly used scenarios, and steps needed to set them up. The most popular scenarios include: _top picks for you, similar content,_ and _read next/watch next._ For E-commerce see the [product recommendations set-up guide](https://www.recombee.com/handbook/download-product-recommendations-dfh09362my2ld4kl32l) including use cases like: _similar products (upsell), bestsellers,_ or _bought together._ ### Predefined and Custom Business Rules ![](https://www.recombee.com/img/blog/innovative-personalization-features-for-2023/05.png) In addition to setting up the logic and the behavior of the recommendations, clients may further customize their use case by adding business rules. The business rules compose of boosters or filters. There is a selection of predefined [business rules](https://docs.recombee.com/scenarios#business-rules) available in the Admin UI, or clients can use ReQL (Recombee query language) and create their own. The filters and boosters help promote or demote certain types of content. For example in VOD, clients may **filter out unavailable content, restrictions on languages, R-rated movies,** or others. Conversely, clients can also **boost visibility of paid content, content from a specific production company, or editorial content.** In E-commerce, you may for example apply filters to not show users products that are out of stock, products with missing images, or products below a specific price or margin. On the other hand, you may use boosters to **promote a specific vendor, products that have been newly added to your catalog, or products with a higher price/margin** than the currently shown product on the detail page. ### Methods of Integration We understand our clients want to integrate the recommendation engine as quick as possible, which is why we strive to make this process simple while also forming meaningful partnerships to provide multiple integration options. So far, we’d like to encourage new clients to consider three existing options. #### Direct Integrations ##### Option 1 - Direct Integration Through API Recombee supports multiple SDK libraries and programming languages. [See API documentation](https://docs.recombee.com/api_clients) ##### Option 2 - Direct Integration Through No-Code Widget Codeless, the fastest way of direct integration - you can use URL to synchronize items catalog and tailor recommendations design through a visual editor. Download a [guidebook for step-by-step guide](https://www.recombee.com/handbook/download-a-step-by-step-guide-to-the-fastest-integration-eyw34h56k2hu56sh). #### Integration Through a Partner - Twilio Segment ![](https://www.recombee.com/img/blog/innovative-personalization-features-for-2023/06.png) ##### Option 3 - Are You a Segment User? Integrate Recombee Through Twilio Segment. Integration through our CDP partner is the fastest way to send user interaction data by adding Recombee as a destination. See the [Recombee Segment recipe for full steps](https://segment.com/recipes/increase-conversions-personalizing-experience-recombee/). ## Special Announcements: 9GAG Aboard! ![](https://www.recombee.com/img/blog/innovative-personalization-features-for-2023/07.png) This year we have partnered with new clients - including the global entertainment giant 9GAG. The cross-platform network with billions of videos has shown incredible uplifts in a very short period of time - by fully personalizing the homepage 9GAG’s overall number of post views increased by 37%, the overall number of interactions by 22%, and total session duration of all users by 7.5%. You can [read the full case study here](https://www.recombee.com/case-studies/9gag). ## Let’s Connect! Interested in a custom personalization roadmap for your business? **Our specialist’s at your disposal** ![](https://www.recombee.com/img/team/petr-popov.png) For business inquiries contact Petr [petr.popov@recombee.com](mailto:petr.popov@recombee.com) ![](https://www.recombee.com/img/team/filip-hanus.png) For integration inquiries contact Filip [filip.hanus@recombee.com](mailto:filip.hanus@recombee.com) New Features Recommendation Engine ## Next Articles [![](https://www.recombee.com/img/blog/breaking-the-news-the-role-of-ai-in-modern-journalism.png)](https://www.recombee.com/blog/breaking-the-news-the-role-of-ai-in-modern-journalism) ### [Breaking the News: The Role of AI in Modern Journalism](https://www.recombee.com/blog/breaking-the-news-the-role-of-ai-in-modern-journalism) Artificial Intelligence (AI) has rapidly transformed the media industry in recent years. From automated news production to trend analysis and personalized content recommendations, AI has brought significant changes to the way media is created, distributed, and consumed. ![](https://www.recombee.com/img/blog/authors/tanalancova.png) Tana Lancova Mar 14, 2023 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/recombee-item-segmentations.png)](https://www.recombee.com/blog/recombee-item-segmentations) ### [Recombee Item Segmentations](https://www.recombee.com/blog/recombee-item-segmentations) Item Segmentations are Recombee's original and elegant solution to various advanced tasks related to hierarchical and relational data. The feature provides a flexible way to group items (products or pieces of content) into segments... ![](https://www.recombee.com/img/blog/authors/tomasrehorek.png) Tomas Rehorek Jan 11, 2023 New Features Recommendation Engine [![](https://www.recombee.com/img/blog/bandit-models-exploiting-popularity-and-curiosity-to-recommend-trending-content.png)](https://www.recombee.com/blog/bandit-models-exploiting-popularity-and-curiosity-to-recommend-trending-content) ### [Bandit Models: Exploiting Popularity and Curiosity to Recommend Trending Content](https://www.recombee.com/blog/bandit-models-exploiting-popularity-and-curiosity-to-recommend-trending-content) Humans are inherently curious. In fact, curiosity is linked to the evolution of humankind. For instance, according to famous historian Yuval Noah Harari in his bestseller book "Sapiens", our language skills evolved as a way of gossiping... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Oct 16, 2022 Recommendation Engine --- # Introducing beeFormer: A Framework for Training Foundational Models for Recommender Systems > Source: https://www.recombee.com/blog/introducing-beeformer-a-framework-for-training-foundational-models-for-recommender-systems > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # Introducing beeFormer: A Framework for Training Foundational Models for Recommender Systems ![](https://www.recombee.com/img/blog/authors/vojtechvancura.png) Vojtech Vancura Oct 15, 2024 [Listen to AI generated podcast](https://drive.google.com/file/d/1ozY3h79xMdlxhVH_T6Y33iV43tkQtbtZ/view?usp=drive_link) ![](https://www.recombee.com/img/blog/introducing-beeformer-a-framework-for-training-foundational-models-for-recommender-systems/main.png) In the fast-evolving world of recommender systems, understanding both how users interact with content and the actual content itself is crucial. Many existing recommender systems struggle to balance these two aspects, especially when there is little interaction data available—commonly known as the "cold-start problem." This issue occurs when new items are introduced, and the system has no past interaction data to rely on for recommendations. _beeFormer_ is an innovative approach designed to solve this challenge by connecting two types of information: what the items are about (semantic similarity) and how users have interacted with them in the past (interaction similarity). By combining these perspectives, _beeFormer_ makes more accurate recommendations, even in situations where data is limited. Let's take a closer look at how _beeFormer_ works and why it has the potential to transform content recommendations. ## The Problem: Semantic vs. Interaction Similarity Traditional recommender systems rely heavily on collaborative filtering (CF) approaches, which use past user interactions (ratings, clicks, views, etc.) to predict future preferences. However, CF methods falter when there is no historical data to draw from—situations like the “cold-start problem” (new users or items) or zero-shot recommendations (where a system must generalize to unseen content without training on interaction data). To address these challenges, many systems turn to content-based filtering (CBF), which uses side information like text descriptions or user reviews. While this helps with semantic similarity—how similar two items are in terms of meaning—it often misses how users actually engage with the items. Users don’t always choose items that are semantically similar but rather those that fit their immediate needs or preferences, creating a gap between what a system thinks is relevant and what is actually interacted with. ## The beeFormer Solution: Merging Two Worlds With _beeFormer_, you are able to recommend not only well established items but also fresh items that do not have any interactions yet. Moreover, _beeFormer_ recommends not only items having semantically similar attributes, but also those that might have high interaction similarity. This significantly improves user experience and lets users discover fresh, diverse and relevant content. This is actually the most important capability of modern recommenders. Similarly to regular sentence transformer based content based filtering methods, _beeFormer_ turns item text descriptions and attributes into a neural item embedding. This embedding, however, is special because it reflects the knowledge of how items are interacted with together. ![](https://www.recombee.com/img/blog/introducing-beeformer-a-framework-for-training-foundational-models-for-recommender-systems/01.png) Training _beeFormer:_ Sentence transformer is improved by knowledge hidden in user interactions. _beeFormer_ integrates the strengths of CF and CBF by training a sentence Transformer model using interaction data, effectively combining semantic knowledge with real-world user behavior represented by interaction data. What makes this approach unique is its ability to transfer knowledge to unseen items, possibly across domains, allowing it to perform well in cold-start and zero-shot scenarios. For example, _beeFormer_ can transfer insights from movie recommendation datasets to book recommendations, a feat that many current recommender systems cannot efficiently accomplish. The key lies in how _beeFormer_ optimizes the Transformer’s understanding of item representations. Instead of merely predicting similarities between items based on their descriptions, _beeFormer_ uses interaction data to fine-tune these representations, capturing behavioral patterns specific to how users interact with items. In a typical _beeFormer_ training workflow: 1. **Text Encoding with Sentence Transformer:** Item descriptions or other side information are encoded into vector representations, providing the semantic basis for recommendations. 2. **Interaction Data Knowledge Transfer with ELSA:** _beeFormer_ uses [ELSA](https://www.recombee.com/blog/making-linear-autoencoders-work-for-large-scale-recommendation-systems), a scalable linear shallow autoencoder, to refine these vectors based on real user interactions. This step injects the behavioral knowledge often missed by semantic-only approaches, allowing the model to better understand actual user preferences at scale. 3. **Optimization with Efficient Gradient Techniques:** To handle large datasets, _beeFormer_ employs techniques like gradient checkpointing and gradient accumulation, enabling the model to scale to massive item catalogs without overwhelming memory resources. ## Experimental Results To verify behavior of _beeFormer_, we have trained a sentence transformer on the Amazon electronic dataset. ![](https://www.recombee.com/img/blog/introducing-beeformer-a-framework-for-training-foundational-models-for-recommender-systems/02.png) On the left, you can see most similar products to Apple macbook (pure content based similarity). On the right, you can see most similar products based on ELSA collaborative filtering. In the middle, most similar items are based on _beeFormer_ training. Text based semantic similarity is clearly combined with interaction similarity. Note that interactions are not used at all during inference of the model making it applicable to pure cold start scenarios. We experimented with _beeFormer_ in three key scenarios: **Time-Split**, **Zero-Shot**, and **Cold-Start**. Here’s an explanation of each, along with interpretation of results. In all scenarios, we measured how well models make recommendations using **R@20, R@50 and N@100:** * **R@20** means the percentage of correct recommendations within the top 20 results. * **R@50** measures this for the top 50 results. * **N@100** indicates a normalized score that takes into account the rank of the recommendation, with a focus on the top 100 results. To increase the reproducibilty of our experiments we generated item-descriptions by Meta-Llama-3.1-8B-Instruct model. LLAMA 3.1 license allows the use of generated output to train new language models. We add the prefix "Llama" to the names of our models to comply with license terms. Using Llama to generate the item descriptions allows us to publish them. More details about the item description generation are available on our GitHub page. ### Time-Split Scenario In this scenario, the model is trained on older interactions and tested on newer ones. It mimics a real-world situation where recommendations need to be made based on historical data for new events or items that appear over time. In the Time-Split setup, both traditional collaborative filtering (CF) models and beeFormer-trained models were compared, with the latter showing superior results.The Amazon Books dataset was sorted by timestamp, and the last 20% of interactions were used as the test set, with the remaining 80% used for training and validation. This way, the model must make recommendations for new books based on older user interactions. ![](https://www.recombee.com/img/blog/introducing-beeformer-a-framework-for-training-foundational-models-for-recommender-systems/03.png) #### Key Observations * **semantic similarity CBF models** like _nomic-embed-text_ try to make recommendations without being trained specifically on the dataset. Their performance is lower (e.g., R@20 = 0.0387). * **Collaborative Filtering (CF)** models like _KNN_ and ELSA recommend books based on user interactions. Their scores are generally moderate, like KNN's R@20 of 0.0370. * **beeFormer:** The standout result here shows that _beeFormer_, which combines text and user interaction data, is able to transfer knowledge from other datasets (like Goodbooks) and perform well on Amazon books. For example, _Llama-amazbooks-mpnet_ achieves R@20 of 0.0706, the highest score in this comparison. ### Zero-Shot Scenario Zero-shot recommendation refers to the model’s ability to make recommendations for items it has never seen before. This scenario measures how well the model can transfer knowledge from one dataset to another without needing additional training on the new dataset. The model was tested in a way where it was trained on one dataset and tested on another. The model had no prior exposure to the test dataset and needed to transfer its learned knowledge from the training dataset(s) to the test dataset. Various **sentence transformer models** were tested on two datasets, **Goodbooks (GB10K)** and **MovieLens (ML20M)**. ![](https://www.recombee.com/img/blog/introducing-beeformer-a-framework-for-training-foundational-models-for-recommender-systems/04.png) #### Key Observations 1. **Goodbooks (GB10K) Results:** * The model **Llama-amazbooks-mpnet** trained on Amazon Books data performs best, with high scores across all metrics (R@20 = 0.2649, R@50 = 0.3957, N@100 = 0.3787). * This result shows that _beeFormer_, which trained on a different dataset (Amazon Books), successfully transfers its knowledge to recommend items from the Goodbooks dataset. This transfer learning improves content-based filtering on Goodbooks compared to other models. 2. **MovieLens (ML20M) Results:** * **Llama-goodbooks-mpnet**, a model trained on Goodbooks data, performs best on MovieLens (R@20 = 0.1589, R@50 = 0.2647, N@100 = 0.2066), indicating that a model trained on book data can still recommend movies, showing its versatility and transferability. Models trained with _beeFormer_ on different datasets (books) can make effective recommendations in entirely different domains (movies), proving the effectiveness of _beeFormer_ in zero-shot scenarios. This capability is critical in real-world systems where new datasets are constantly added, and training from scratch may not be feasible. The key takeaway is that _beeFormer_ facilitates knowledge transfer across domains, allowing models trained on one type of content (books) to successfully recommend items in another (movies), achieving strong performance without requiring specific training on the target dataset. ### Cold-Start Scenario The cold-start problem occurs when a system has to recommend items with little to no interaction data, such as newly added items or users. In this case, the model has to rely heavily on side information, such as item descriptions. The experiment involved using side information (text descriptions of items) to generalize recommendations to new, unseen items. For benchmarking, _Heater_ (a model that maps interaction data to side information) was used, and the models trained with _beeFormer_ demonstrated superior performance by making use of both interaction and content data. ![](https://www.recombee.com/img/blog/introducing-beeformer-a-framework-for-training-foundational-models-for-recommender-systems/05.png) #### Key Observations 1. **Goodbooks (GB10K) Results:** * The **Llama-goodlens-mpnet** trained with _beeFormer_ model performs best in this scenario for GB10K, achieving high recall (R@20 = 0.2710, R@50 = 0.4218, N@100 = 0.4066). * This model combines knowledge from books and movies, this ability to accumulate knowledge from multiple domains marks an important step towards general, multi-domain recommender systems. 2. **MovieLens (ML20M) Results:** * The **Llama-goodlens-mpnet** model again leads the performance in the MovieLens dataset with an outstanding R@20 of 0.4630 and R@50 of 0.6152, indicating it is highly effective in recommending unseen items. * This demonstrates the model's robustness across domains. In cold-start scenarios, models rely heavily on side information (such as item descriptions) to generalize and make recommendations for unseen items. The results highlight that _beeFormer_ models, trained on multiple datasets (MovieLens and Goodbooks), outperform models trained on just a single dataset. This emphasizes the model’s ability to leverage knowledge from multiple sources and handle new items more effectively. **Potential for Foundational Models:** The conclusion suggests that there is potential to build a "foundational" _beeFormer_ model trained on multiple datasets, making it more versatile and applicable across different domains, helping with cold-start issues across a broad range of applications. In summary, _beeFormer_ models, particularly **Llama-goodlens-mpnet**, excel in cold-start scenarios by leveraging text data and performing well across both book and movie datasets. This makes them ideal for situations where new items or users are frequently introduced. ## Conclusion: A Bee in Every System? In conclusion, _beeFormer_ demonstrates remarkable versatility and effectiveness in tackling some of the most challenging scenarios in recommendation systems, such as zero-shot and cold-start. By leveraging the power of text-based side information and combining it with interaction data, _beeFormer_ models outperform traditional methods and other state-of-the-art transformers. The ability to transfer knowledge across domains, as shown in the zero-shot and time-split scenarios, highlights its potential to generalize and perform well on unseen datasets. Furthermore, in cold-start situations where new items or users are introduced with limited interaction data, _beeFormer_ excels by utilizing foundational knowledge from multiple datasets, making it a valuable tool for a wide range of industries and applications. The success of models like Llama-goodlens-mpnet across different domains suggests a promising future for building universal, foundational recommender systems. With further development and expansion into multi-modal datasets, _beeFormer_ could become a cornerstone of modern recommendation technology, delivering highly personalized and accurate recommendations, even in the most data-sparse environments. **Ready to explore beeFormer?** Dive into its source code and learn more on the official [GitHub repository](https://github.com/recombee/beeFormer). Or contact Recombee to try _beeFormer_ on your data without the need of training and maintaining your model and integrating it into the recommender pipeline. This research was performed in collaboration with Czech Technical University in Prague and Charles University in Prague. New Features Recommendation Engine ## Next Articles [![](https://www.recombee.com/img/blog/celestial-tiger-entertainment-launches-new-chinese-movie-app-cmgo-with-diagnal.png)](https://www.diagnal.com/cmgo/) ### [Celestial Tiger Entertainment launches new Chinese Movie app, CMGO, with DIAGNAL](https://www.diagnal.com/cmgo/) With Recombee’s AI-powered recommendation engine working with DIAGNAL Enhance, CMGO serves up personalised experiences for each viewer, driving engagement for the service. ![](https://www.recombee.com/img/blog/authors/diagnal.png) Diagnal Nov 14, 2024 Personalization Partnerships [![](https://www.recombee.com/img/blog/video-recommendations-made-easy-integrating-axinom-mosaic-with-recombee.png)](https://www.axinom.com/webinar/video-backends-with-recommendations) ### [Video Recommendations Made Easy: Integrating Axinom Mosaic with Recombee](https://www.axinom.com/webinar/video-backends-with-recommendations) In this webinar we look into building a data-driven video backend geared towards personalized video recommendations, integration with Axinom Mosaic, and how to transform user experiences on streaming platforms. ![](https://www.recombee.com/img/blog/authors/grigorygrin.png) Grigory Grin (Axinom) Sep 09, 2024 Personalization Partnerships [![](https://www.recombee.com/img/blog/insights-the-next-level-of-analytics-in-recombee-ui.png)](https://www.recombee.com/blog/insights-the-next-level-of-analytics-in-recombee-ui) ### [Insights: The Next Level of Analytics in Recombee UI](https://www.recombee.com/blog/insights-the-next-level-of-analytics-in-recombee-ui) Insights, the analytics section of our Admin UI, offers various predefined and fully customizable reports to track recommended items and how users interact with these recommendations. ![](https://www.recombee.com/img/blog/authors/ondrejfiedler.png) Ondrej Fiedler May 09, 2024 New Features Recommendation Engine --- # Digital Media Personalization | Content Recommendations | Blog > Source: https://www.recombee.com/blog/keeping-up-with-digital-media-convergence > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # Keeping Up With Digital Media Convergence ![](https://www.recombee.com/img/blog/authors/annadolezelova.png) Anna Dolezelova Oct 8, 2022 ![](https://www.recombee.com/img/blog/keeping-up-with-digital-media-convergence/main.png) At Recombee, we felt the transition within the media industry accelerated by the pandemic. [OTT and CTV](https://www.oracle.com/cx/advertising/measurement/ctv-vs-ott/#:~:text=Connected%20TV%20%28CTV%29%20definition,%2C%20Apple%20TV%2C%20and%20more.) consumption ballooned at a significant rate. Viewing is dynamic, ratings change, content is disseminated through multiple platforms, and business models have greater sophistication. We operate under ever-ongoing digital media convergence driven by innovations in technology. As we interact with digital media leaders and professionals day-to-day, learning about the needs and concerns in their industry, we decided to set foot in some of the events that bring them and all their challenges together. During the past couple of months, we immersed ourselves in talks and discussions on current and future digital media trends. We participated in multiple media-oriented conferences (such as IBC or HbbTV Symposium), where we kept our eyes and ears wide open. From changing working model synergies, through automation vs. humans, to finding the key to the ultimate user experience, here are our five key takeaways. ## Five Takeaways ### 1 Holistic Approach to Workflow ![](https://www.recombee.com/img/blog/keeping-up-with-digital-media-convergence/1.png) There was a lot of fuss about the paradigm shift in the workflow and Media2Cloud, a solution that helps streamline and automate the content ingestion process when you migrate your digital assets to the cloud. What was it all about? Well, over time, everything has to evolve around the audience. So instead of thinking about each product separately, there should be an effort to create a working model synergy. Currently, there is a paradigm shift underway. From **an old, platform-based workflow** that is focused on building a workflow around each platform separately (and only leads to creating siloes) to **a new, cloud-centric, converged workflow** that focuses on laying a cloud layer on top of all platforms to unify them and create an uninterruptible working synergy that allows for smooth communication and cooperation between the platforms. Just Hut of KPN, the leading telecommunications and IT provider and market leader in the Netherlands, said during his [panel at IBC 2022](https://www.ibc.org/video/kpn-taking-bold-action-to-power-innovation/8918.article?adredir=1) that they are fully convinced about moving from a data center to the cloud. Still, remember creating a new setup is important, but striking a healthy balance between innovation and stability even more so. ### 2 Human Touch Is Crucial ![](https://www.recombee.com/img/blog/keeping-up-with-digital-media-convergence/2.png) You can decide which content is being shown and automate it, but you still need to keep the human touch. Most digital media leaders agreed on that. In nearly every meeting at each of the conferences where we discussed AI-powered automation and personalization, the other side asked us if their editorial team can have influence over the personalized content being shown to their users. Our answer? That is absolutely essential. With Recombee, we empower companies to utilize content and behavioral analysis to automate the generating of personalized recommendations for their viewers while still allowing for human tuning of the recommendations to meet business needs. How do we manage that? [With our business rules and filters.](https://docs.recombee.com/reql_filtering_and_boosting) ### 3 AVOD on the Rise ![](https://www.recombee.com/img/blog/keeping-up-with-digital-media-convergence/3.png) Third and probably the most frequently mentioned topic was advertising. In 2017, most video-on-demand platforms were based on paying for the content (either with a subscription fee - SVOD or on a pay-per-view basis - TVOD), and ads were 2nd class relative to the paid part of the business. Since 2020, ads took a hit, and even companies like Netflix, who previously have been very vocal about not giving any space to advertisers, have now tapped out on the subscription revenue and are [looking into the ads revenue streams.](https://about.netflix.com/en/news/netflix-partners-with-microsoft) So in the coming years, ads will represent an increasingly larger share among the VOD platform providers. According to the latest research from analysis powerhouse Omdia, AVOD streamers are set to boom with [$259bn](https://tbivision.com/2022/03/23/exclusive-avod-set-to-boom-with-259bn-annual-revenue-by-2025-research-claims/) annual revenue by 2025. We are advocates of the subscription model as it enables product owners to focus on making their platform for users. When ad revenues come into play, they always need to balance having a great product for happy users and making advertisers happy. Still, many of our clients offer both subscription and free/reduced-price services powered by ads with the full support of our recommendation technology. ### 4 Remain Close to the User ![](https://www.recombee.com/img/blog/keeping-up-with-digital-media-convergence/4.png) The future is in making content available in all formats while remaining close to the user. To achieve that, OTT providers should focus on implementing a scalable, agile framework aimed at users' needs. The ultimate win? Consumers can constantly watch the content they long for, loyalty improves, and ROI increases. How to achieve that? Deliver content across all your channels, platforms, and devices, and personalize the user experience one-on-one. **Couple more tips for a customer-centric approach** 1. Act as a partner, not a vendor 2. Align on what are the ambitions and aspirations 3. Keep control of the content ### 5 Personalization Is the Key to Building the Next Level of Experience ![](https://www.recombee.com/img/blog/keeping-up-with-digital-media-convergence/5.png) Personalization plays a vital role in automating and improving the user experience. Having quality products or services ONLY will not cut it. [77%](https://www.slideshare.net/TrackIF/forrester-webinar-individualization-versus-personalization) of consumers will choose, recommend or pay more for a brand that provides personalized service or experience. At Recombee, we want to make the customer journey the ultimate experience. Our engine helps your users [find exactly what they are looking for](https://www.recombee.com/blog/real-time-personalization-of-content-with-ai-powered-recommendations) or discover what they didn't know existed and engage them with the content, generating real-time, AI-powered recommendations. The solution is versatile, yet our focus has greatly shifted towards media, where our technology enables us to cope with challenges such as: * Constantly growing user-generated content * Varying preferences of individual users * Sites in various languages and dialects * Instant recommending of newly added content under large traffic What else is in it for you except for obviously the higher customer retention, loyalty, and satisfaction? Increase in desired KPIs such as conversion rate, click-through rate, emailing click-to-open rate, number of views, as well as a rise in ROI (e.g., profit from the advertisement shown). **To prove our point, here are some of our client success stories:** * [9GAG](https://www.recombee.com/case-studies/9gag) * [Prima](https://www.recombee.com/case-studies/ftv-prima) * [Showmax](https://www.recombee.com/case-studies/showmax) ## Conclusion Digital consumption is increasing rapidly. To thrive and remain competitive, content providers must remain close to customers, bring forth personalized content, and provide engaging, rich, innovative media experiences. Who stands to win from this? The users. And companies that keep up with video-hungry consumers by optimizing their products to meet the users' needs. Who stands to lose? Those who won’t navigate the digital media convergence. So don't lose. Strive for subscribers, product quality, and long-term engagement. Recombee is here to guide you along the way; happy to continue discussing the upcoming trends and help you navigate the changes. We are one email away: [business@recombee.com](mailto:business@recombee.com). Personalization ## Next Articles [![](https://www.recombee.com/img/blog/repsys-opensource-library-for-interactive-evaluation-of-recommendation-systems.png)](https://www.recombee.com/blog/repsys-opensource-library-for-interactive-evaluation-of-recommendation-systems) ### [Visual and Interactive Evaluation of Recommender Systems](https://www.recombee.com/blog/repsys-opensource-library-for-interactive-evaluation-of-recommendation-systems) When building modern real-world artificial intelligence systems, it is increasingly important to validate that the system works correctly. This is however not an easy task. Existing tools for machine learning practitioners... ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Sep 19, 2022 Recommendation Engine [![](https://www.recombee.com/img/blog/bandit-models-exploiting-popularity-and-curiosity-to-recommend-trending-content.png)](https://www.recombee.com/blog/bandit-models-exploiting-popularity-and-curiosity-to-recommend-trending-content) ### [Bandit Models: Exploiting Popularity and Curiosity to Recommend Trending Content](https://www.recombee.com/blog/bandit-models-exploiting-popularity-and-curiosity-to-recommend-trending-content) Humans are inherently curious. In fact, curiosity is linked to the evolution of humankind. For instance, according to famous historian Yuval Noah Harari in his bestseller book "Sapiens", our language skills evolved as a way of gossiping... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Oct 16, 2022 Recommendation Engine [![](https://www.recombee.com/img/blog/recombee-in-e-mail-marketing-a-partner-success-story-with-ryzeo.png)](https://www.recombee.com/blog/recombee-in-e-mail-marketing-a-partner-success-story-with-ryzeo) ### [Recombee in E-mail Marketing: A Partner Success Story with Ryzeo](https://www.recombee.com/blog/recombee-in-e-mail-marketing-a-partner-success-story-with-ryzeo) Do you feel there is a potential to increase your success with customers through an efficient recommender engine? You're highly likely right. Adding a recommender service to your emailing campaigns gives each client tailored product recommendations in all of their emails. ![](https://www.recombee.com/img/blog/authors/russellmiller.png) Russell Miller (Ryzeo) Oct 5, 2022 Partnerships [![](https://www.recombee.com/img/blog/how-we-are-using-ai-to-power-content-recommendations.png)](https://revium.com.au/blog/how-we-are-using-ai-to-power-content-recommendations) ### [How We Are Using AI to Power Content Recommendations](https://revium.com.au/blog/how-we-are-using-ai-to-power-content-recommendations) In this article we walk you through how we are using the AI recommendation engine Recombee embedded in our headless CMS StoryBlok to drive content recommendations throughout our own website. ![](https://www.recombee.com/img/blog/authors/revium.png) Revium Sep 30, 2022 Partnerships --- # Key Trends in News & Media for 2025 and the Role of AI-Powered Solutions > Source: https://www.recombee.com/blog/key-trends-in-news-and-media-for-2025-and-the-role-of-ai-powered-solutions > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # Key Trends in News & Media for 2025 and the Role of AI-Powered Solutions ![](https://www.recombee.com/img/blog/authors/annapetr.png) Gabriela Takacova & Violeta Milarova Feb 20, 2025 ![](https://www.recombee.com/img/blog/key-trends-in-news-and-media-for-2025-and-the-role-of-ai-powered-solutions/main.png) The way news is produced, delivered, and consumed keeps shifting, and as we move through 2025, a few big changes have become impossible to ignore. News organizations are working harder than ever to keep up, figuring out how to connect with audiences, stay financially stable, and make the most of technology without losing their core identity. We got a firsthand look at these shifts through collaboration with some of the leading sites in the media sphere. Our insights were also confirmed during the INMA Media Innovation Week in Helsinki, where some of the biggest names in the industry, like Axel Springer, Mediahuis, and DMG Media, came together to talk about where things are headed. What stood out the most? Two major challenges: * Maintaining a strong editorial voice in an era where AI is reshaping everything. * Finding better ways to bring in subscribers and, more importantly, keep them. These aren’t small issues, but AI-driven personalization, when done right, can help tackle both. Let’s break it down. ## Insight #1: Editorial Control Still Matters a Lot ![](https://www.recombee.com/img/blog/key-trends-in-news-and-media-for-2025-and-the-role-of-ai-powered-solutions/01.png) Even as AI becomes more common in newsrooms, editors are still the ones shaping what people see. This is especially true on homepages, where first impressions matter. The trick is finding the balance between using AI to improve efficiency and keeping the human touch that gives a news brand its voice. AI can suggest personalized content at scale, but editorial decisions ultimately shape brand identity, add context and relevance, and provide a sense of purpose that algorithms simply can’t replicate. That’s why editors need to stay in the driver’s seat. ### How Recombee Supports Editorial Oversight This is exactly why Recombee’s platform is built the way it is. Instead of replacing human judgment, its recommendation engine works alongside editors, making sure they still have control over what gets featured. With its “human-in-the-loop” approach, editors can tweak AI-generated recommendations in real time, adjusting specific content visibility, prioritizing important stories, and ensuring the homepage reflects the publication’s editorial values. It’s about using AI to enhance editors’ voices, not override them. ## Insight #2: The Subscription Game Is Tougher Than Ever ![](https://www.recombee.com/img/blog/key-trends-in-news-and-media-for-2025-and-the-role-of-ai-powered-solutions/02.png) Getting people to pay for news is one thing. Keeping them subscribed? That’s an entirely different battle. With so many media options out there, audiences expect content that feels tailored to them, otherwise they’ll move on. Personalization isn’t just a nice feature anymore; it’s what keeps readers engaged. If a news platform consistently delivers stories that match someone’s interests, they’re way more likely to stick around. ### How Recombee Helps Keep Readers Hooked Recombee’s AI-driven recommendation engine takes personalization up a notch by analyzing browsing habits, content interactions, and even location in real time to surface the most relevant stories for each user. Beyond on-site recommendations, it also fine-tunes email newsletters, ensuring subscribers receive content that genuinely matches their interests instead of a generic list of articles. Infinite scroll feeds in news apps are another example of AI-driven content delivery, a concept familiar from platforms like TikTok, Instagram, and Facebook. The result? Higher engagement, better retention, and more loyal readers. ## What a Recommender Engine Needs to Deliver in 2025 ![](https://www.recombee.com/img/blog/key-trends-in-news-and-media-for-2025-and-the-role-of-ai-powered-solutions/03.png) For news organizations trying to keep up with changing audience expectations, having a solid recommendation system isn’t optional anymore, it’s a necessity. Based on what’s happening in the industry, here are four key things a good recommender engine should do: ### 1\. Deliver Breaking News Instantly with Real-Time Updates People expect to see the latest stories the moment they happen. A strong recommender engine must push breaking news to readers instantly, ensuring they stay informed without delay. Recombee’s real-time engine keeps media platforms ahead of the competition by delivering fresh stories as soon as they are published, making them the go-to source for up-to-the-minute updates. ### 2\. Handle Traffic Spikes Without Slowing Down Major news events like elections, championship games, and political scandals can drive huge surges in traffic. If a platform can’t keep up, readers will look elsewhere. Recombee’s system is built to scale effortlessly, handling tens of thousands of recommendations per second so sites stay responsive and reliable, even at peak demand. ### 3\. Create Personalized News Feeds for Every Reader No two readers are the same. Some want political deep dives, while others follow tech trends or entertainment gossip. Recombee ensures that every user gets a feed tailored to their interests by analyzing reading habits, engagement patterns, and device preferences. Whether browsing on a laptop or skimming headlines on mobile, readers get a seamless, highly relevant experience. ### 4\. Boost Engagement with Cross-Platform and Cross-Site Recommendations Readers don’t just stop at one article. They explore, follow links, and engage with related content across different formats. A smart recommender engine should connect the dots, guiding users toward relevant stories, opinion pieces, videos, and even classifieds or e-commerce pages. Through recirculation, Recombee helps publishers keep audiences engaged longer by continuously suggesting content that matches their interests, increasing both readership and revenue. ### 5\. Turn Data into Actionable Insights with Real-Time Analytics A good recommender engine isn’t just about delivering content. It should also provide publishers with deep insights into user behavior. Recombee’s Insights tool, available in the Admin UI, allows publishers to track how readers engage with recommendations and explore a range of analytical views or create custom reports. By understanding which topics are gaining traction and where engagement spikes, publishers can fine-tune recommendation rules and align the system with their editorial strategy. ### 6\. Upgrade Search with AI-Powered Semantic Understanding Finding the right news shouldn’t feel like a guessing game. Traditional keyword-based searches often deliver results that are too broad or miss the real intent behind a query. Recombee’s semantic search changes that by interpreting meaning, context, and user intent rather than just matching keywords. With LLM-powered natural language processing, users can search as if they’re having a conversation, getting smarter, more relevant results that align with what they actually mean, not just what they type. ## AI + Human Judgment = Smarter News Delivery ![](https://www.recombee.com/img/blog/key-trends-in-news-and-media-for-2025-and-the-role-of-ai-powered-solutions/04.png) The future of news isn’t about choosing between automation and human expertise. It is about making them work together. AI-driven personalization can improve efficiency and engagement, but only when it complements the human side of journalism rather than replacing it. For media companies looking to stay ahead, the key is choosing AI solutions that respect editorial control, enhance personalization, and scale with audience needs. Recombee was built with these challenges in mind, helping publishers create smarter, more engaging news experiences without losing their unique voice. ## Want to see how AI-powered recommendations play out in a real-world newsroom? Check out how [Prima](https://www.recombee.com/case-studies/ftv-prima-content), the largest media group in the Czech market, used Recombee to transform its content strategy. [Schedule a Demo](https://www.recombee.com/request-demo) or [Sign Up](https://admin.recombee.com/sign-up) for a free 30-day trial. Personalization ## Next Articles [![](https://www.recombee.com/img/blog/recombee-research-2024.png)](https://www.recombee.com/blog/recombee-research-2024) ### [Recombee Research 2024](https://www.recombee.com/blog/recombee-research-2024) Recombee has always been deeply connected to academia, with four of our six co-founders holding Ph.D. degrees. Over the years, our investment in research has grown alongside our company and the increasing demand for advanced features in the market. ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Feb 23, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/2024-wrap-up.png)](https://www.recombee.com/blog/2024-wrap-up) ### [2024 Wrap-Up](https://www.recombee.com/blog/2024-wrap-up) As we wrap up 2024, let’s take a moment to celebrate the milestones we’ve achieved together. Your collaboration and feedback have been at the heart of our progress. Here’s a quick recap of last year’s highlights. ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Jan 27, 2025 New Features Personalization [![](https://www.recombee.com/img/blog/celestial-tiger-entertainment-launches-new-chinese-movie-app-cmgo-with-diagnal.png)](https://www.diagnal.com/cmgo/) ### [Celestial Tiger Entertainment launches new Chinese Movie app, CMGO, with DIAGNAL](https://www.diagnal.com/cmgo/) With Recombee’s AI-powered recommendation engine working with DIAGNAL Enhance, CMGO serves up personalised experiences for each viewer, driving engagement for the service. ![](https://www.recombee.com/img/blog/authors/diagnal.png) Diagnal Nov 14, 2024 Personalization Partnerships --- # Looking Back at 2025 > Source: https://www.recombee.com/blog/looking-back-at-2025 > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # Looking Back at 2025 ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Jan 20, 2026 ![](https://www.recombee.com/img/blog/looking-back-at-2025/main.png) 2025 marked **10 years of Recombee**. A decade of building personalization from first principles, shaped by research, real-world deployments, and close collaboration with partners across industries. **Here’s what defined the year.** ## New Success Stories We had the privilege of welcoming new partners who push the boundaries of personalization at scale. ![](https://www.recombee.com/img/blog/looking-back-at-2025/01.png) ### The Telegraph One of the UK’s most influential news brands now personalizes article discovery across their website and app with Recombee. The integration helps millions of readers find stories that match their interests while supporting editorial strategy and onward journey optimization. [Read More](https://www.recombee.com/blog/recombee-partners-with-the-telegraph-to-deliver-ai-driven-personalisation-to-millions-of-readers) ### DAZN A global sports streaming powerhouse bringing personalization to sports content where timing, loyalty, and context matter more than anything. We’re proud to support DAZN in delivering 1:1 experiences to millions of passionate fans. These collaborations underline the growing need for real-time, adaptive personalization in both news and sports - two of the most dynamic content ecosystems in the world. ## Events & Community 2025 was a big year for connecting with partners, product leaders, and the wider media and streaming community. ![](https://www.recombee.com/img/blog/looking-back-at-2025/02.png) ### The Reflections: Future of AI & Personalization A standout moment of the year was hosting **The Reflections** \- our first-ever industry event in Prague. It brought together more than 80 leaders from media, technology, and AI for an evening of conversations about the future of recommendation systems, product innovation, and the evolving role of AI in content discovery. It marked a major step into building our own community-driven formats. ![](https://www.recombee.com/img/blog/looking-back-at-2025/03.png) ### Industry Presence Alongside launching our own event, we continued engaging with the industry at leading global conferences, including **INMA Media Innovation Week**, **IBC**, and **SportsPro Media Summit**, as well as local meetups and partner-led sessions across Europe. These gatherings helped us exchange insights, share our work, and stay connected to the challenges and goals of our partners. ## Recombee Science 2025 was another strong year for the Recombee Research team, with new papers published, prototypes tested, and several advancements making their way into production features. Our research continued to focus on deep representation learning, semantic understanding, and models that adapt quickly to fast-changing content. ![](https://www.recombee.com/img/blog/looking-back-at-2025/04.png) A major highlight was our presence at **ACM RecSys 2025 in Prague**, where our research team contributed five papers across theory, modeling, fairness, and real-world scalability, alongside a [joint paper](https://research.idi.ntnu.no/NewsTech/INRA/program.html) with **The Telegraph** showcasing how large-scale personalization can deliver measurable results while respecting editorial values. It was a meaningful milestone that demonstrated the strength of combining research-driven innovation with real-world deployments. Alongside this, our team continued contributing to journals, conferences, and applied research initiatives - reinforcing our commitment to bridging the gap between academic breakthroughs and production-ready recommendation technology. ## New Offices in the Heart of Prague One of the most exciting milestones of 2025 was moving into our new office space in the center of Prague. The new location has already become a hub for collaboration, learning, and community moments. ![](https://www.recombee.com/img/blog/looking-back-at-2025/05.png) The new space gives our growing team room to build, experiment, and welcome more guests in the year ahead, while marking an important step in our continued growth. ## Thank You for an Amazing 2025 Each milestone, from scientific breakthroughs to new partnerships, has been shaped by your input and collaboration. Thank you for being part of our journey this past year. We’re excited to keep building the future of real-time personalization with you in 2026\. Onward to another year of innovation, bold ideas, and meaningful impact. ✨ Recommendation Engine Personalization ## Next Articles [![](https://www.recombee.com/img/blog/a-2025-research-retrospective.png)](https://www.recombee.com/blog/a-2025-research-retrospective) ### [A 2025 Research Retrospective](https://www.recombee.com/blog/a-2025-research-retrospective) In 2025, I became Head of Research at Recombee. I knew it would be a challenge: and I wasn’t wrong. Recombee is a leader in recommendation-as-a-service, with thousands of clients... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Jan 29, 2026 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/product-highlights-from-2025.png)](https://www.recombee.com/blog/product-highlights-from-2025) ### [Product Highlights from 2025](https://www.recombee.com/blog/product-highlights-from-2025) In 2025, we focused on making advanced personalization easier to implement, scale, and maintain across products and platforms. Here’s a look at the key product updates we released last year. ![](https://www.recombee.com/img/blog/authors/ondrejfiedler.png) Ondrej Fiedler Jan 08, 2026 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/ai-news-and-outlook-for-2026.png)](https://www.recombee.com/blog/ai-news-and-outlook-for-2026) ### [AI News and Outlook for 2026](https://www.recombee.com/blog/ai-news-and-outlook-for-2026) Here’s what caught my attention in AI research lately, and where things might be heading in 2026\. After 25+ years in this field, the pace has gotten hard to keep up with. I’m trying to make sense of... ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Dec 17, 2025 Recommendation Engine Personalization --- # Making Recommendations Fairer: A New Way to Guarantee Exposure for All > Source: https://www.recombee.com/blog/making-recommendations-fairer-a-new-way-to-guarantee-exposure-for-all > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # Making Recommendations Fairer: A New Way to Guarantee Exposure for All ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Apr 29, 2025 ![](https://www.recombee.com/img/blog/making-recommendations-fairer-a-new-way-to-guarantee-exposure-for-all/main.png) ## The Problem: Exposure Bias in Recommendations As recommender systems become more widespread across digital platforms, concerns around fairness are coming to the forefront. Standard relevance-based ranking techniques, while effective at optimizing user satisfaction, often reinforce existing biases and contribute to the systematic underexposure of certain item groups. For example, on music or book platforms, content from independent or minority creators may be consistently ranked lower due to historical popularity patterns, even when it's just as relevant to the user's interests. This phenomenon is known as exposure bias, and it has the potential to perpetuate visibility gaps, reduce content diversity, and further marginalize creators who are already underrepresented in the ecosystem. Exposure bias in recommender systems arises due to position bias: users are more likely to interact with items that appear at the top of recommendation lists. Even if an item is included in a recommendation list, its rank position strongly influences whether it receives user attention. Existing fairness-aware ranking methods have made progress in increasing group representation. However, many fail to address exposure explicitly, or do so without providing guarantees on its distribution across groups. Some methods also rely on large optimization models, which makes them too resource-heavy for real-world systems with large item catalogs. ## Our Approach: A Scalable Fairness Framework Our solution for this problem is a post-processing framework based on Integer Linear Programming (ILP) that re-ranks recommendation lists to satisfy fairness constraints. The framework is centered around two types of exposure constraints: * **Minimum Exposure:** Ensures that each group receives at least a fixed proportion of the total exposure. * **Relative Exposure:** Guarantees that the exposure of a protected group is at least a number of times the exposure of a non-protected group. * **Exposure is quantified** using a logarithmic discount factor to account for position bias, consistent with established metrics like nDCG. Notably, our ILP formulation involves at most K×|I| binary variables (where K is the number of recommendation positions and |I| the number of items), making it significantly more scalable than existing quadratic-size models. The technical details of the development can be quite complex for the general public. For those interested, we recommend reading our paper and welcome direct contact for further discussion. ## What We Learned: Results and Insights **Our key empirical findings in offline data are:** 1. The framework increased exposure of disadvantaged items significantly, especially in long-tail scenarios. 2. As expected, there was some trade-off in accuracy, especially with offline data. But the drop was modest and could be adjusted with a tunable parameter. 3. More active users with longer interaction history often maintained the same level of accuracy, likely because they already engage with long-tail items. **Why It Matters?** In many real-world scenarios (like music platforms, online marketplaces, or news apps) fair exposure isn't just nice to have; it's essential. Sometimes it's even part of business rules or regulatory compliance (e.g., guaranteeing visibility to all content providers). Our research provides a flexible, practical solution that balances fairness and accuracy, and can be deployed without redesigning the entire system. Moreover, promoting fair exposure can have long-term benefits for user engagement by encouraging exploration and increasing satisfaction across a broader range of content. Over time, this can also help diversify consumption patterns and expand the reach of underrepresented creators or products. ## Use-Case Examples ![](https://www.recombee.com/img/blog/making-recommendations-fairer-a-new-way-to-guarantee-exposure-for-all/01.png) For instance, consider a music streaming service where millions of tracks, from blockbuster hits to indie tracks, compete for listener attention. Moreover, in the original setup, 90 percent of plays cluster around the top 1 percent of songs. But by slipping our re‑ranking module into the pipeline, treating “independent artist” as the protected group and tuning our relative‑exposure weight, the service suddenly could boost those long‑tail tracks surfacing in users’ top‑20 recommendations. In e‑commerce domain, for instance, an online marketplace could aim to give eco‑friendly and women‑owned brands a fighting chance against mass‑market names. By running our post‑processor model and enforcing a relative‑exposure boost for certified producers, the platform can nudge its carousel of items toward a greener balance. The expected result? Certified‑green products will see their click‑through rate climb, while shoppers’ average basket size remained steady. Social feeds can also benefit from fairer exposure. For example, a photo-sharing app can use our framework to ensure that emerging creators (i.e., those with few followers) make up at least 40 percent of the top positions in each personalized feed. By boosting this group, the app can significantly increase new-creator engagement. ## Conclusion In summary, recommender systems go much beyond a single model. They are complex ecosystems involving not only prediction accuracy but also fairness, diversity, user experience, and alignment with broader organizational goals. Our work focuses on the often-overlooked post-processing stage, demonstrating that fairness can be effectively introduced without compromising system architecture or scalability. As the field moves toward more responsible and human-centered recommendation, incorporating exposure guarantees is a concrete step toward making these systems more inclusive, accountable, and aligned with real-world constraints. At Recombee, we align with this direction by incorporating principles of transparency, fairness, and practical deployment into our research and development efforts, with the goal of building recommender systems that are both effective and socially responsible. ## References Lopes, Ramon, Rodrigo Alves, Antoine Ledent, Rodrygo Santos, and Marius Kloft. "Recommendations with minimum exposure guarantees: A post-processing framework." _Expert Systems with Applications_ 236 (2024): 121164. Recommendation Engine Personalization ## Next Articles [![](https://www.recombee.com/img/blog/shield-the-universal-framework-making-ai-search-safer-for-everyone.png)](https://www.recombee.com/blog/shield-the-universal-framework-making-ai-search-safer-for-everyone) ### [SHIELD: The Universal Framework Making AI Search Safer for Everyone](https://www.recombee.com/blog/shield-the-universal-framework-making-ai-search-safer-for-everyone) Imagine searching for "glass tubing" and getting recommendations for drug manufacturing equipment. As AI-powered search becomes ubiquitous — from online marketplaces to social networks... ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Jun 24, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/2025-sneak-peek.png)](https://www.recombee.com/blog/2025-sneak-peek) ### [2025 Sneak Peek](https://www.recombee.com/blog/2025-sneak-peek) This year is already off to an exciting start, and we’re rolling out new tools to improve efficiency and optimize recommendations. Here’s what’s available and what’s coming next. ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Mar 19, 2025 Recommendation Engine New Features [![](https://www.recombee.com/img/blog/build-vs-buy-deciding-the-best-approach-for-your-recommender-system.png)](https://www.recombee.com/blog/build-vs-buy-deciding-the-best-approach-for-your-recommender-system) ### [Build vs. Buy: Deciding the Best Approach for Your Recommender System](https://www.recombee.com/blog/build-vs-buy-deciding-the-best-approach-for-your-recommender-system) When it comes to deciding between buying a recommender system and building one from scratch, the choice isn’t always straightforward. Both options come with their own set of pros and cons... ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova & Ondrej Fiedler Mar 14, 2025 Recommendation Engine Personalization --- # Mid-Year Roundup: 2026 So Far > Source: https://www.recombee.com/blog/mid-year-roundup-2026-so-far > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # Mid-Year Roundup: 2026 So Far ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Jul 20, 2026 ![](https://www.recombee.com/img/blog/mid-year-roundup-2026-so-far/main.png) The digital world continues to change at an incredible pace. As technology evolves and user expectations keep rising, creating experiences that feel relevant has never been more important. The first half of 2026 brought new partnerships, industry recognition, research milestones, and fresh ways to help platforms connect with their audiences. **Here’s what shaped the year so far.** ## New Success Stories More platforms are advancing their personalization strategies to help users discover the right content, products, and opportunities at the right moment. We’re excited to welcome new partners including: ![](https://www.recombee.com/img/blog/mid-year-roundup-2026-so-far/01.png) ### BeatStars The leading digital music marketplace for artists and producers is using Recombee to strengthen track discovery, increase engagement, and help users find beats and sounds that match their creative needs. ### Perlego The global learning platform partnered with Recombee to improve content discovery, helping learners navigate its extensive library and find resources that align with their goals and interests. [Explore Case Study](https://www.recombee.com/case-studies/perlego) ### Poslovi Infostud Serbia’s leading employment platform is using Recombee to help job seekers discover more relevant opportunities while helping employers connect with the right candidates faster. These collaborations reflect the growing role of real-time, adaptive personalization in creating digital experiences that feel more relevant for every user. [Explore Case Study](https://www.recombee.com/case-studies/poslovi-infostud) ## Vibe Coding Challenge Innovation often starts with experimentation. That’s exactly what inspired our Vibe Coding Challenge, where teammates across Recombee took a day to turn ideas into working prototypes. ![](https://www.recombee.com/img/blog/mid-year-roundup-2026-so-far/02.png) From new feature concepts to creative use cases and internal improvements, colleagues from both technical and non-technical teams explored what’s possible with today’s tools. Stay tuned to see which ideas make it into production. 👀 ## Events & Community The first half of the year brought plenty of opportunities to exchange ideas with product leaders, media innovators, and the wider community. ![](https://www.recombee.com/img/blog/mid-year-roundup-2026-so-far/03.png) A major highlight came at the **INMA World Congress of News Media**, where our solution [Segment-Aware Analytics](https://ceur-ws.org/Vol-4056/short1.pdf), developed together with The Telegraph, was awarded second place in the **Best Use of Generative AI** category. The project shows how AI can help teams move beyond surface-level insights and better understand audience behavior, content performance, and personalization opportunities. We also continued conversations around the future of digital experiences at **SportsPro London** and other industry events. ## Recombee Science Research remains one of the foundations behind everything we do. This year, our Recombee Research team contributed **three papers** to **ACM UMAP 2026**, one of the leading conferences in user modeling, adaptation, and personalization. ![](https://www.recombee.com/img/blog/mid-year-roundup-2026-so-far/04.png) The papers explore how recommendation systems can better understand content, users, and preferences. [Language Embeddings Meet Shallow Autoencoders](https://dl.acm.org/doi/full/10.1145/3774935.3806192), **nominated for the Best Paper Award** and recognized among the **Best Paper Candidates in the Short Papers category**, looks at improving recommendations in cold-start scenarios through semantic understanding. [The Stars Align](https://dl.acm.org/doi/10.1145/3774935.3812702) examines how systems can better interpret differences in user rating behavior, while [Leveraging Artist Catalogs for Cold-Start Music Recommendation](https://dl.acm.org/doi/10.1145/3774935.3806178) demonstrates how an artist’s existing catalog can be used to deliver more accurate recommendations for newly released music. ![](https://www.recombee.com/img/blog/mid-year-roundup-2026-so-far/05.png) None of these milestones would be possible without our partners, customers, and community around us. Your collaboration continues to inspire what we build next. We’re excited to keep experimenting, innovating, and advancing personalization throughout the rest of 2026 and beyond. ✨ Recommendation Engine Personalization ## Next Articles [![](https://www.recombee.com/img/blog/new-feature-ab-testing.png)](https://www.recombee.com/blog/new-feature-ab-testing) ### [New Feature: A/B Testing](https://www.recombee.com/blog/new-feature-ab-testing) Personalization is never finished. Every change to your recommendation strategy raises new questions. Should you boost fresh content? Change used Logic? Introduce new filters? ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Aug 14, 2026 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/a-2025-research-retrospective.png)](https://www.recombee.com/blog/a-2025-research-retrospective) ### [A 2025 Research Retrospective](https://www.recombee.com/blog/a-2025-research-retrospective) In 2025, I became Head of Research at Recombee. I knew it would be a challenge: and I wasn’t wrong. Recombee is a leader in recommendation-as-a-service, with thousands of clients... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Jan 29, 2026 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/looking-back-at-2025.png)](https://www.recombee.com/blog/looking-back-at-2025) ### [Looking Back at 2025](https://www.recombee.com/blog/looking-back-at-2025) 2025 marked 10 years of Recombee. A decade of building personalization from first principles, shaped by research, real-world deployments, and close collaboration with partners across industries. ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Jan 20, 2026 Recommendation Engine Personalization --- # New Feature: A/B Testing > Source: https://www.recombee.com/blog/new-feature-ab-testing > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # New Feature: A/B Testing ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Aug 14, 2026 ![](https://www.recombee.com/img/blog/new-feature-ab-testing/main.png) Personalization is never finished. Every change to your recommendation strategy raises new questions. Should you boost fresh content? Change used Logic? Introduce new filters? Even small changes can influence engagement, conversions, and revenue. That's why we've introduced **A/B Testing.** A/B Testing lets you compare different recommendation configurations within a Scenario and measure their impact before rolling changes out to all users. Instead of relying on assumptions, teams can validate every optimization with real user behavior. ## Test Every Part of Your Recommendation Strategy Each experiment runs against your current Scenario configuration, with traffic automatically split between a Control and one or more Variants. Variants can modify any combination of recommendation Logic, Filters, Boosters, and Constraints, making it easy to compare different strategies while keeping everything else consistent. The Admin UI also highlights exactly what changed in each Variant, giving teams a clear overview of every experiment. ![](https://www.recombee.com/img/blog/new-feature-ab-testing/1.png) ## Measure the Metrics That Matter Choose from a library of predefined metrics tailored to your industry, including Click-Through Rate (CTR), Conversion Rate, Watch Time from Recommendations, and Income from Recommendations. Need something more specific? You can also create custom metrics using Insights, allowing you to measure highly targeted business outcomes, from purchases within selected product categories to engagement with specific types of content. ## Make Decisions with Confidence Once an experiment is running, Recombee continuously evaluates each Variant and presents the results in an intuitive report. ![](https://www.recombee.com/img/blog/new-feature-ab-testing/2.png) Compare improvements against the Control, see the probability that each Variant outperforms the baseline, understand whether the results are statistically significant, and identify the best-performing configuration at a glance. Whether you're fine-tuning recommendation models or experimenting with entirely new strategies, A/B Testing helps every optimization move forward with confidence rather than guesswork. Recommendation Engine Personalization ## Next Articles [![](https://www.recombee.com/img/blog/mid-year-roundup-2026-so-far.png)](https://www.recombee.com/blog/mid-year-roundup-2026-so-far) ### [Mid-Year Roundup: 2026 So Far](https://www.recombee.com/blog/mid-year-roundup-2026-so-far) The digital world continues to change at an incredible pace. As technology evolves and user expectations keep rising, creating experiences that feel relevant has never been more important. ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Jul 20, 2026 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/a-2025-research-retrospective.png)](https://www.recombee.com/blog/a-2025-research-retrospective) ### [A 2025 Research Retrospective](https://www.recombee.com/blog/a-2025-research-retrospective) In 2025, I became Head of Research at Recombee. I knew it would be a challenge: and I wasn’t wrong. Recombee is a leader in recommendation-as-a-service, with thousands of clients... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Jan 29, 2026 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/looking-back-at-2025.png)](https://www.recombee.com/blog/looking-back-at-2025) ### [Looking Back at 2025](https://www.recombee.com/blog/looking-back-at-2025) 2025 marked 10 years of Recombee. A decade of building personalization from first principles, shaped by research, real-world deployments, and close collaboration with partners across industries. ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Jan 20, 2026 Recommendation Engine Personalization --- # New Features Real-Time Recommendation Engine | Blog > Source: https://www.recombee.com/blog/new-features-for-a-better-personalization-experience > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # New Features for a Better Personalization Experience ![](https://www.recombee.com/img/blog/authors/ondrejfiedler.png) Ondrej Fiedler Mar 07, 2022 ![](https://www.recombee.com/img/blog/new-features-for-a-better-personalization-experience/01.png) Like most of the world, the majority of 2021 was spent on home office or in isolation - which left us with all the time to be invested in work (and Netflix :) ) and continue improving UX for our clients. We are now happy to share what our team was able to build and what advanced features we can offer to reach new levels of personalization. ## Advanced Features for Content Recommendations (Video, Music, Articles...) ### Netflix-Like and Showmax-Like Rows As the popularity of streaming platforms spirals, so does the complexity of personalization. We have noticed multiple trends across the industry, where most platforms are trying to implement features alike to the global giant of video streaming platforms - Netflix, Showmax. We collaborate with our premium clients on integrating their video streaming platforms (VOD, SVOD, AVOD) in a new way of showcasing content using Netflix-like rows (which is how Netflix achieved 80% stream time through personalization). Utilizing a two-tiered row-based ranking, the recommendations are organized vertically as well as horizontally. Meaning, within each row, the strongest recommendations are on the left, and across the rows, the strongest recommendations are on the top. Resulting in a fully personalized experience where each user’s homepage is assembled from the most relevant categories and titles based on their unique tastes and preferences. ![](https://www.recombee.com/img/blog/new-features-for-a-better-personalization-experience/11.png) ### Infinite Scroll Personalization Besides Netflix, another platform that lately earned its name as a master of personalization - TikTok, inspired us to innovate the content recommendations game. We have developed a unique technology enabling real-time infinite scroll personalization that will provide users with TikTok, Instagram, or Twitter-like experiences. The more and more popular Infinite scroll enables bottomless content loading as the user scrolls down the page, eliminating the need for pagination. This technique makes every user feel like there are endless content options specific to their taste with every scroll. ![](https://www.recombee.com/img/blog/new-features-for-a-better-personalization-experience/10.png) ### Personalized Search 2.0 Have you ever searched for a foreign artist, but weren’t quite sure how to spell their name? Or typed in “MaDDona” for the 6th time, and left wondering why there are no albums being shown? Not every AI is trained to understand typos and understand the user meant “MadoNNa”. As we are working with many media houses, podcast platforms, and other sites providing content on their side that can be found with a search function, we ease the user’s experience with a [personalized full-text search.](https://www.recombee.com/where-to-use#full-text-search) Our personalized search 2.0 helps users find what they have in mind as long as the search query remotely resembles any piece of content that was uploaded in the catalog feed. This function now also includes [Search synonyms](https://docs.recombee.com/api#synonyms). ![](https://www.recombee.com/img/blog/new-features-for-a-better-personalization-experience/05.png) ### Content Personalization in Action The new features aim for users to enjoy their streaming experience with each login. Providing personalized guidance helps users find what they are looking for much faster and saves them from the frustration of aimless scrolling. With the competitiveness of the online landscape and hundreds of alternative options online, a tailored experience can convert unsubscribed users into happy loyal streamers. To see a fully personalized experience from the first click, try our [media demo site](https://demo-content.recombee.com/). ### Scenario Set-up Guide Throughout the years we understood every client has different pain points and is trying to achieve different KPIs. That is also the logic of each scenario which can be easily configured in a codeless way and changed over time. Some of the most popular scenarios (with explanations) can be found in our [content recommendations scenario set-up guide](https://www.recombee.com/handbook/download-content-recommendations-35kl362g0946239g0er). ![](https://www.recombee.com/img/blog/new-features-for-a-better-personalization-experience/02.png) _E.g. See a personalized site for a user interested in family content. All scenarios are personalized to a specific genre, even though the platform offers romance, thriller, comedy, and many others._ ## Advanced Features for Product Recommendations (E-Commerce, Marketplaces, Real-Estate...) ### Infinite Scroll Personalization A lot of features from our content recommendations can also be applied to our product recommendations - for example, infinite scroll can be used in e-commerce or marketplaces, or even in real estate to increase the chance of finding the right product. The continuous load personalization is a feature that can be now easily integrated through our [Admin UI](https://admin.recombee.com/). ### Personalized Search 2.0 For marketplaces or online stores with huge catalogs of products well personalized internal search is a game-changer! It does not only save users time but can also provide a competitive advantage through seamless UX. Recombee’s [personalized full-text search](https://www.recombee.com/where-to-use#full-text-search) enables [Search synonyms](https://docs.recombee.com/api#synonyms) and the inclusion of typos. For example, an online shop that offers both “coffee tables” and “nightstands” can show a user seeking “ a table” both options available, despite it not being the table seeker's exact command. ### Product Recommendations in Action Real AI powers all our scenarios and undergoes constant retraining to meet each client’s KPIs. As there are more numbers of e-commerce and marketplaces scenarios to choose from, we have prepared a [demo store](https://recombeedemostore.myshopify.com/) to see in practice some of our most frequently used ones (i.e. “just for you”, “popular & trending”, or “recently viewed”). For no coding configuration of the most popular scenarios, our clients can now follow our [Product recommendations scenario set-up guide.](https://www.recombee.com/handbook/download-product-recommendations-dfh09362my2ld4kl32l) ![](https://www.recombee.com/img/blog/new-features-for-a-better-personalization-experience/03.png) ## Even Faster and More Efficient Integration Understanding huge variability in our client's use-cases and technical capabilities, we always strive to provide multiple integration options, so each client can find the most suitable solution. To enable new clients to effortlessly implement Recombee, we have prepared new ways for integration - the fastest one can be done in a matter of minutes! ### No-Code Widget & Built-in User Identifier Using 1st Party Cookies The simplest, fastest way to integrate Recombee with no coding is through [our No-Code Widget](https://docs.recombee.com/no-code-widgets) that allows for item synchronization using URL. When first released, clients preferring the No-Code Widget method were identifying users from their side. With the growing trend of concerns of user privacy, we have developed a built-in user identifier using 1st party cookies. Working with first-party cookies enables Recombee to provide the highest level of personalization while adhering to the strictest privacy regulations. Switching to 1st party cookies is one of many behind-the-scenes steps Recombee initiates to help our clients fulfill current and upcoming data privacy policies (see differences between 1st and 3rd party cookies [here](https://clearcode.cc/blog/difference-between-first-party-third-party-cookies/)). ### Extended Support for Catalog Feeds We have completely redone the Catalog feeds section and besides the Google Merchant feed, we now also support RSS and custom XML or CSS feeds. If you generate a catalog feed in almost any format, you can now just set its URL in the Admin UI, and Recombee will take care of periodical parsing and processing of the feed, ensuring that your items catalog is always up to date. ### Integration Issues Section There is nothing more common during software development than introducing a bug. A new section of our Admin UI helps developers find & fix these issues by showing error messages from all the API calls in a single listing. ### Segment Integration ![](https://www.recombee.com/img/blog/new-features-for-a-better-personalization-experience/06.png) For clients using [Segment,](https://segment.com/) or wishing to use Segment, we have developed an easy, minimum coding needed integration through the CDP platform. The integration is done on Segment’s platform and can be configured in Recombee’s admin UI. We chose Segment because it greatly simplifies data collection from digital properties, such as websites or apps, which helps clients gather clean data for multiple purposes. Following the [integration steps,](https://docs.recombee.com/segment#segment-integration) clients can send views, purchases of products, or video watch time information to Recombee, and start personalizing within a few clicks. Read more information in our [Recombee Segment blog post](https://www.recombee.com/blog/recombee-real-time-ai-recommendations-as-the-new-destination-in-segment). ### Kentico Software Integration For clients looking for multiple marketing services, we have enabled easy integration through a DXP platform [Kentico Xperience,](https://xperience.io/?_ampl=c2207080-d59f-4aea-9709-54aa2e6a51caR&_ga=1705261133.1637069013) where Recombee AI recommendations can be added through the [Kentico Xperience Recombee module.](https://github.com/Kentico/xperience-module-recombee) Similarly Kentico Xperience, there’s a simple integration available through [Kentico Kontent,](https://kontent.ai/?_ampl=c2207080-d59f-4aea-9709-54aa2e6a51caR&_ga=1705261133.1637069013) a headless CMS for content management at scale, now available with Recombee personalization. ## Continuous AI Research Our AI is the core of our system, and there is no expense spared when it comes to developing new models. Our researchers can be typically found on the grounds of our partner institutes, European AI hub [prg.ai,](https://prg.ai/en/) or the [Czech Technological University in Prague,](https://www.cvut.cz/en) where Recombee develops new models and algorithms. ### Improving Linear Methods for Recommendation In our article on [Linear Methods and Autoencoders in Recommender Systems](https://www.recombee.com/blog/linear-methods-and-autoencoders-in-recommender-systems), you can read about linear models and how we improved these models by a non-linear part realized by variational autoencoder. ![](https://www.recombee.com/img/blog/linear-methods-and-autoencoders-in-recommender-systems/autoencoder-5.png) You can check out our code from the [repository.](https://paperswithcode.com/sota/collaborative-filtering-on-movielens-20m) ### Next Basket Prediction With Neural Networks Continuing on our research into next basket prediction, Recombee reaped recognition at the annual Recommender System Conference, RecSys 2021 in Amsterdam. Our Machine Learner Developer Vojtech Vancura presented research into [Neural Basket Embedding for Sequential Recommendation,](https://dl.acm.org/doi/abs/10.1145/3460231.3473896) whose goal is to develop models that would automatically refill customers’ shopping carts based on their habits. We are on the tip of our toes to be amongst the firsts to present these revolutionary models to our clients. We have also organized the next round of challenges for senior data science students, where the number of active participants who submitted a solution was almost 50\. We are currently supporting the research of the winning student, who is introducing new architectures of popular transformer models to handle long purchasing sequences. Read more in our [Deep Learning for Recommender Systems: Next basket prediction and sequential product recommendation.](https://medium.com/recombee-blog/deep-learning-for-recommender-systems-next-basket-prediction-and-sequential-product-recommendation-796228b34dee) ![](https://www.recombee.com/img/blog/new-features-for-a-better-personalization-experience/08.png) ## Let’s Connect! Interested in a custom personalization roadmap for your business? Meet our team and let’s talk about recommendations. ![](https://www.recombee.com/img/team/gabriela-takacova.png) For business inquiries contact Gabriela [gabriela.takacova@recombee.com](mailto:gabriela.takacova@recombee.com) ![](https://www.recombee.com/img/team/filip-hanus.png) For technical inquiries contact Filip [filip.hanus@recombee.com](mailto:filip.hanus@recombee.com) ![](https://www.recombee.com/img/team/karen-harazimova.png) For partnership opportunities contact Karen [karen.harazimova@recombee.com](mailto:karen.harazimova@recombee.com) New Features ## Next Articles [![](https://www.recombee.com/img/blog/making-linear-autoencoders-work-for-large-scale-recommendation-systems.png)](https://www.recombee.com/blog/making-linear-autoencoders-work-for-large-scale-recommendation-systems) ### [Making Linear Autoencoders Work for Large Scale Recommendation Systems](https://www.recombee.com/blog/making-linear-autoencoders-work-for-large-scale-recommendation-systems) Linear autoencoders for collaborative filtering in recommender systems are simple and surprisingly accurate as we explained in our blogpost on how linear methods work. The critical disadvantage of methods like EASE is that they are not applicable to real-world problems... ![](https://www.recombee.com/img/blog/authors/vojtechvancura.png) Vojtech Vancura Aug 29, 2022 Recommendation Engine [![](https://www.recombee.com/img/blog/advancing-your-career-in-artificial-intelligence-with-prg-ai-and-recombee.png)](https://www.recombee.com/blog/advancing-your-career-in-artificial-intelligence-with-prg-ai-and-recombee) ### [Advancing Your Career in Artificial Intelligence with prg.ai and Recombee](https://www.recombee.com/blog/advancing-your-career-in-artificial-intelligence-with-prg-ai-and-recombee) At Recombee, we have always collaborated with academia — after all, five of our co-founders graduated from the Czech Technical University in Prague, one of the largest and oldest technical universities in Europe, and most of them hold a Ph.D. degree. ![](https://www.recombee.com/img/blog/authors/karenharazimova.png) Karen Harazimova Nov 21, 2021 [![](https://www.recombee.com/img/blog/linear-methods-and-autoencoders-in-recommender-systems.png)](https://www.recombee.com/blog/linear-methods-and-autoencoders-in-recommender-systems) ### [Linear Methods and Autoencoders in Recommender Systems](https://www.recombee.com/blog/linear-methods-and-autoencoders-in-recommender-systems) Linear regression is probably the simplest and surprisingly efficient machine learning method. It should be the method of your first choice, according to the famous KISS principle. Also, it often works better than sophisticated methods, because it is... ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Sep 7, 2021 Recommendation Engine --- # No-Code Search Widget: Personalized, Powerful, Effortless > Source: https://www.recombee.com/blog/no-code-search-widget-personalized-powerful-effortless > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # No-Code Search Widget: Personalized, Powerful, Effortless ![](https://www.recombee.com/img/blog/authors/ondrejfiedler.png) Ondrej Fiedler Jul 25, 2025 ![](https://www.recombee.com/img/blog/no-code-search-widget-personalized-powerful-effortless/main.png) At Recombee, we don't just excel at recommendations – we provide powerful full-text search capabilities too. Our Quick, No-Code Search Widget exemplifies this, offering a seamless, customizable search experience that's quick to integrate and enhances the utility of our recommendation engine. ## Our Search Engine Recombee's search is fully **personalized**, taking into account both the **user's query** and **their interactions**. This ensures that users receive highly relevant results, allowing them to quickly find exactly what they’re looking for. When two users type the same query, they won't necessarily see the same results. For instance, after typing “Ha,” User A might see “Hannibal” first if they tend to watch thrillers while User B might see “Harry Potter” if they prefer fantasy. Additionally, our search supports multiple languages and is typo-tolerant, so even if users misspell a director's name, their movies will still appear. Let's explore the key benefits and functionalities of our no-code search widget feature. ## Customizable Search Experience ![](https://www.recombee.com/img/blog/no-code-search-widget-personalized-powerful-effortless/1.png) The Quick Search Widget is a user interface element designed to provide a customizable search experience. It allows you to tailor the search functionality to your specific needs and is **fully configurable in the Admin UI**. It supports multiple sections – Items, Hero, and Segments – each connected to specific recommendation scenarios. You can easily rearrange these sections using a drag-and-drop interface: 1. **Items Section**: Displays found items, customizable to show images, titles, subtitles, and highlighted information like prices. 2. **Hero Section**: Optionally highlights the top item, making it stand out. 3. **Segments Section**: Optionally shows matching Item Segments (categories, genres, vendors), allowing users to navigate directly to relevant pages. ## Leverage Your Existing Data For customers already using Recombee's recommendation solution for product and content recommendations, transitioning to the no-code Quick Search Widget is effortless. The widget automatically uses existing catalog and user data, eliminating the need to send new datasets. This integration maximizes the value of your data, delivering accurate and relevant search results to users from the start. ## Effortless No-Code Search Integration Integrating our Quick Search Widget couldn’t be simpler. With just a few clicks in the Admin panel, followed by copying and pasting an **embed code**, you can add robust, no-code search functionality to your platform. This ease of integration makes it accessible, even for those with **minimal technical expertise**. ## Fine-Tune Visual Appearance We understand that maintaining a consistent brand appearance is crucial. The Quick Search Widget allows for extensive visual customization through CSS. You can adjust the widget's appearance to align perfectly with your site's design, ensuring a seamless user experience. ![](https://www.recombee.com/img/blog/no-code-search-widget-personalized-powerful-effortless/2.png) ## Conclusion The Quick Search Widget is a powerful addition to the Recombee recommendation engine, offering a streamlined, customizable, data-integrated, and no-code search solution. By simplifying integration and enhancing the search experience, it enables you to provide users with accurate and relevant search results effortlessly. Try the no-code Quick Search Widget today and see how it can transform your platform's search functionality. For more detailed information and to start integrating the Quick Search Widget, visit our [documentation here](https://docs.recombee.com/no-code-widgets#html-widget-full-text-search). Recommendation Engine Personalization ## Next Articles [![](https://www.recombee.com/img/blog/the-building-blocks-of-privacy-friendly-personalization.png)](https://www.recombee.com/blog/the-building-blocks-of-privacy-friendly-personalization) ### [The Building Blocks of Privacy-Friendly Personalization](https://www.recombee.com/blog/the-building-blocks-of-privacy-friendly-personalization) Personalization can be achieved without compromising user privacy. While many personalization systems have historically relied on practices now considered intrusive, like... ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Aug 07, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/how-regionalization-based-recommendations-can-improve-your-operations.png)](https://www.recombee.com/blog/how-regionalization-based-recommendations-can-improve-your-operations) ### [How Regionalization-Based Recommendations Can Improve Your Operations](https://www.recombee.com/blog/how-regionalization-based-recommendations-can-improve-your-operations) From ancient trade routes to modern urban planning, geography has consistently shaped human decisions and opportunities. Today, in the world of online business and personalized recommendations... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Jul 18, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/shield-the-universal-framework-making-ai-search-safer-for-everyone.png)](https://www.recombee.com/blog/shield-the-universal-framework-making-ai-search-safer-for-everyone) ### [SHIELD: The Universal Framework Making AI Search Safer for Everyone](https://www.recombee.com/blog/shield-the-universal-framework-making-ai-search-safer-for-everyone) Imagine searching for "glass tubing" and getting recommendations for drug manufacturing equipment. As AI-powered search becomes ubiquitous — from online marketplaces to social networks... ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Jun 24, 2025 Recommendation Engine Personalization --- # Product Highlights from 2025 > Source: https://www.recombee.com/blog/product-highlights-from-2025 > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # Product Highlights from 2025 ![](https://www.recombee.com/img/blog/authors/ondrejfiedler.png) Ondrej Fiedler Jan 08, 2026 ![](https://www.recombee.com/img/blog/product-highlights-from-2025/main.png) In 2025, we focused on making advanced personalization easier to implement, scale, and maintain across products and platforms. Here’s a look at the key product updates we released last year. ![](https://www.recombee.com/img/blog/product-highlights-from-2025/01.png) ## Composite Recommendations: Dynamic Homepages Made Easy **Composite Recommendations** make it easy to build impactful personalization use cases like **Because You Watched** or [Fully personalized homepages](https://www.recombee.com/features/recommendations-search#fully-personalized-homepage). Teams can create dynamic rows and sections that automatically re-order based on each user’s interests and real-time context, enabling richer discovery experiences without adding complexity as personalization setups scale. [Explore Composite Recommendations](https://docs.recombee.com/scenarios#composite-recommendations) ![](https://www.recombee.com/img/blog/product-highlights-from-2025/02.png) ## Semantic Search and Semantic Segmentations **2025 saw a major surge in the adoption of Semantic Search**, our premium feature that goes beyond simple keyword matching. Powered by large language models (LLMs), it understands the meaning behind both catalog metadata and user queries - enabling users to search naturally using intent, concepts, and full questions rather than exact keywords. We also introduced a brand-new capability: **Semantic Segmentations**. Using advanced AI, the system automatically uncovers niche item clusters and assigns them meaningful, human-readable names by analyzing their content with LLMs. With Semantic Segmentations, we can dynamically define homepage rows, such as micro-genres or trending topics, that update in real time and continuously reflect what's most relevant. [Explore Semantic Segmentations](https://www.recombee.com/features/recommendations-search#semantic-segmentations) ![](https://www.recombee.com/img/blog/product-highlights-from-2025/04.png) ## Swift SDK for iOS & New Demo App Our SDK family expanded with the release of the **Swift SDK for iOS**, giving mobile teams a modern and intuitive way to integrate Recombee directly into native apps. Swift remains the go-to for iOS development, and this SDK makes adding personalized recommendations seamless across any content-rich mobile experience. To help developers get started quickly, we also launched a new **iOS demo app** built on a movie dataset. It showcases key use cases like personalized Top Picks for You, related titles, real-time interaction tracking, and instant recommendation updates - all powered by Recombee. [Explore Swift SDK](https://docs.recombee.com/swift_client) [Try the Demo App](https://github.com/recombee/ios-demo) ![](https://www.recombee.com/img/blog/product-highlights-from-2025/03.png) ## Smarter Scenario Management Managing Scenarios is now easier than ever. We introduced **tags** and **icons** that help teams instantly understand what each Scenario is for - whether it powers the homepage, content detail pages, mobile, web, or any custom flow. With clearer grouping and visual labeling, organizing and navigating your Scenarios becomes faster, more intuitive, and far more scalable as your personalization setup grows. ![](https://www.recombee.com/img/blog/product-highlights-from-2025/05.png) ## Widget SDKs: Faster UI Development Across Devices We also launched **Widget SDKs** \- ready-made building blocks for adding personalized recommendations and search to any website or app. The SDKs include customizable components like carousels, grids, feeds, and quick search, handling the heavy lifting of data fetching, loading states, responsiveness, and user interaction out of the box. Teams can style the widgets freely or integrate them with modern frameworks like React to build rich, dynamic, personalized experiences in a fraction of the time. [Explore Widget SDKs](https://docs.recombee.com/widget-sdks) More updates are already in motion, and 2026 is shaping up to build on these foundations with new capabilities already underway. Recommendation Engine Personalization ## Next Articles [![](https://www.recombee.com/img/blog/looking-back-at-2025.png)](https://www.recombee.com/blog/looking-back-at-2025) ### [Looking Back at 2025](https://www.recombee.com/blog/looking-back-at-2025) 2025 marked 10 years of Recombee. A decade of building personalization from first principles, shaped by research, real-world deployments, and close collaboration with partners across industries. ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Jan 20, 2026 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/ai-news-and-outlook-for-2026.png)](https://www.recombee.com/blog/ai-news-and-outlook-for-2026) ### [AI News and Outlook for 2026](https://www.recombee.com/blog/ai-news-and-outlook-for-2026) Here’s what caught my attention in AI research lately, and where things might be heading in 2026\. After 25+ years in this field, the pace has gotten hard to keep up with. I’m trying to make sense of... ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Dec 17, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/recombee-partners-with-the-telegraph-to-deliver-ai-driven-personalisation-to-millions-of-readers.png)](https://www.recombee.com/blog/recombee-partners-with-the-telegraph-to-deliver-ai-driven-personalisation-to-millions-of-readers) ### [Recombee Partners with The Telegraph to Deliver AI-Driven Personalisation to Millions of Readers](https://www.recombee.com/blog/recombee-partners-with-the-telegraph-to-deliver-ai-driven-personalisation-to-millions-of-readers) Prague, 11th September 2025 – Recombee, a leading AI-powered recommendation platform, has announced a strategic partnership with The Telegraph... ![](https://www.recombee.com/img/blog/authors/recombee.png) Recombee Oct 02, 2025 Company News --- # Increase Your Customer’s Success With Recombee Engine | Blog > Source: https://www.recombee.com/blog/recombee-in-e-mail-marketing-a-partner-success-story-with-ryzeo > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # Recombee in E-mail Marketing: A Partner Success Story with Ryzeo ![](https://www.recombee.com/img/blog/authors/russellmiller.png) Russell Miller (Ryzeo) Oct 5, 2022 ![](https://www.recombee.com/img/blog/recombee-in-e-mail-marketing-a-partner-success-story-with-ryzeo/main.png) Do you feel there is a potential to increase your success with customers through an efficient recommender engine? You're highly likely right. Adding a recommender service to your emailing campaigns gives each client tailored product recommendations in all of their emails. With the limited time customers spend on your site, it is all the more crucial to build a rapport. Making sure you can re-engage and bring the buyers back to your site with one-on-one personalization is key. Here are examples of automated emails where we’ve added Recombee recommendations to drive thousands of dollars in sales for our customers. ## Abandoned Cart When you send an abandoned cart email, you should also recommend three other products for your customers to consider. Why? Because the odds are that they may not like the item they abandoned, so you need to show them an alternative. Recommended products can increase your conversion rate by 10-20%. This email had a 9% conversion rate, and made over $80,000 in a month. A big part of that was due to Recombee product recommendations. ![](https://www.recombee.com/img/blog/recombee-in-e-mail-marketing-a-partner-success-story-with-ryzeo/1.png) What other email templates can you add product recommendations to? ## Product View This email is triggered whenever someone views a product. They don’t even have to add it to their cart. Although this email series has a lower conversion rate than the abandoned cart, it typically has 5x the volume, producing more overall revenue. These emails can convert at 5-11% when you add in product recommendations to increase your odds of a sale. ## Welcome Series This email is triggered whenever someone signs up for your website in a popup. It’s typically a series of emails introducing your site. If you include personalized product recommendations, you can greatly increase your conversion rate. In consumer products, we have seen conversion rates as high as 16% when product recommendations are used. ![](https://www.recombee.com/img/blog/recombee-in-e-mail-marketing-a-partner-success-story-with-ryzeo/2.png) ## Post-purchase Email Another email automation that works very well with product recommendations is a **post-purchase email.** This email sequence is sent out after a customer has made a purchase. We use Recombee for our customers to sell add-on products (cross-sell) for the product they just purchased. Combined with a time-limited coupon, this can significantly boost your sales. ## Newsletters Finally, what about **Newsletters?** You can also insert personalized product recommendation blocks in your newsletters that go out to your entire email list. Even better, you can create **automated newsletters** that just have product recommendations, no work required from you. We find that newsletters with product recommendations can generate up to 6% conversion rate. ![](https://www.recombee.com/img/blog/recombee-in-e-mail-marketing-a-partner-success-story-with-ryzeo/3.png) The most powerful use of newsletters combines product recommendations with sophisticated segments. For example, segmenting out B2B vs B2C customers, high LTV customers, or customers with a strong brand affinity. A custom segment allows you to create a subject line that speaks directly to your audience, increasing your open and clickthrough rate. When combined with personalized product recommendations, it can be a powerful marketing tool. Taken altogether, we believe that adding product recommendations to both automated one-to-one emails and email newsletters can greatly increase your overall sales, by offering customers products they are most likely to be interested in. At Ryzeo, we have years of experience integrating Recombee product recommendations into our E-commerce email system. We help our customers make 20-40% of their sales from our emails. If you’re looking for a “done for you” option, and are an established E-commerce business, [set a discovery session with us](https://calendly.com/ryzeo/call) to see how we can supercharge your sales using Recombee product recommendations and our behavioral segmentation email system. _Russell Miller is the COO of Ryzeo, an E-commerce Email Platform, and Recombee Partner._ Partnerships ## Next Articles [![](https://www.recombee.com/img/blog/keeping-up-with-digital-media-convergence.png)](https://www.recombee.com/blog/keeping-up-with-digital-media-convergence) ### [Keeping Up With Digital Media Convergence](https://www.recombee.com/blog/keeping-up-with-digital-media-convergence) At Recombee, we felt the transition within the media industry accelerated by the pandemic. OTT and CTV consumption ballooned at a significant rate. ![](https://www.recombee.com/img/blog/authors/annadolezelova.png) Anna Dolezelova Oct 8, 2022 Personalization [![](https://www.recombee.com/img/blog/how-we-are-using-ai-to-power-content-recommendations.png)](https://revium.com.au/blog/how-we-are-using-ai-to-power-content-recommendations) ### [How We Are Using AI to Power Content Recommendations](https://revium.com.au/blog/how-we-are-using-ai-to-power-content-recommendations) In this article we walk you through how we are using the AI recommendation engine Recombee embedded in our headless CMS StoryBlok to drive content recommendations throughout our own website. ![](https://www.recombee.com/img/blog/authors/revium.png) Revium Sep 30, 2022 Partnerships [![](https://www.recombee.com/img/blog/repsys-opensource-library-for-interactive-evaluation-of-recommendation-systems.png)](https://www.recombee.com/blog/repsys-opensource-library-for-interactive-evaluation-of-recommendation-systems) ### [Visual and Interactive Evaluation of Recommender Systems](https://www.recombee.com/blog/repsys-opensource-library-for-interactive-evaluation-of-recommendation-systems) When building modern real-world artificial intelligence systems, it is increasingly important to validate that the system works correctly. This is however not an easy task. Existing tools for machine learning practitioners... ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Sep 19, 2022 Recommendation Engine --- # Recombee Partners With Axinom to Enhance Video Streaming Experiences > Source: https://www.recombee.com/blog/recombee-partners-with-axinom-to-enhance-video-streaming-experiences > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # Recombee Partners With Axinom to Enhance Video Streaming Experiences ![](https://www.recombee.com/img/blog/authors/janvaluch.png) Jan Valuch Mar 20, 2024 **Prague, March 18, 2024** – Recombee, the leading innovator in AI-driven recommendation technologies for video and media platforms, announces a new partnership with Axinom, an expert in video streaming backends, to revolutionize digital user experiences. ![](https://www.recombee.com/img/blog/recombee-partners-with-axinom-to-enhance-video-streaming-experiences/main.png) This collaboration is set to introduce a new era of personalized and engaging digital user experiences by integrating Recombee's advanced machine-learning algorithms with Axinom's robust [video streaming backed development platform, Axinom Mosaic.](https://www.axinom.com/products/mosaic) At the heart of this collaboration are Recombee's advanced machine-learning algorithms, which power dynamic content and search recommendations. These algorithms enable streaming platforms to construct unparalleled user experiences, boosting engagement and satisfaction. Recombee's recommendation engine is celebrated for its ability to extract value from your data and instantly captivate users with real-time, personalized content suggestions from the first click. Recombee's solution offers companies complete control over their recommendations, allowing them to customize their behavior and leverage [unique features designed for video platforms](https://www.recombee.com/domains/video). Axinom Mosaic stands out with its flexible, microservice-based architecture, which seamlessly combines managed and open-source services. This unique architecture is crucial for the integration of Recombee, providing multiple API endpoints for various Mosaic services like Media, Catalog, Image, and User Management. Additionally, Mosaic services furnish essential metadata and usage insights to enrich the user experience further. Gabriela Takacova, Co-founder and CBO of Recombee, highlights the significance of this partnership: "In today’s market, personalized user experiences are not just preferred; they are expected. Our collaboration with Axinom allows companies to build an innovative, comprehensive solution for the video industry, marking a significant leap forward in meeting and exceeding these expectations. Axinom’s backend platform, known for its versatility and modularity, complements our vision, enabling streaming platforms to significantly enhance both the developer and end-user experience." "Recombee’s state-of-the-art recommendation engine is a great counterpart for our Mosaic services, and this partnership allows us to showcase Axinom Mosaic’s strengths in backend development and easy integration. Utilizing both solutions, companies can not only enhance user experiences but can also benefit from best-of-breed solutions for both backend workflows and recommendation engines," commented Stefanie Schuster, CCO, Axinom. ## About Recombee Recombee revolutionizes digital platforms with its sophisticated, AI-powered recommendation engine, empowering product managers to customize recommendations to align with their unique strategic business goals. Established by leading data and machine learning experts in 2015, Recombee has rapidly become a key player in the recommendation engine market, serving a diverse global client base, including industry giants like 9GAG and Showmax. For more information, visit [www.recombee.com](https://www.recombee.com) ## About Axinom Axinom helps media businesses overcome digital challenges and succeed in a rapidly changing landscape. We offer building blocks that enable the development of content-first backends. Media companies, broadcasters, and telcos worldwide rely on Axinom's products to tackle workflows for processing, managing, securing, and delivering video content. For more information, visit [www.axinom.com](http://www.axinom.com) Partnerships Personalization ## Next Articles [![](https://www.recombee.com/img/blog/insights-the-next-level-of-analytics-in-recombee-ui.png)](https://www.recombee.com/blog/insights-the-next-level-of-analytics-in-recombee-ui) ### [Insights: The Next Level of Analytics in Recombee UI](https://www.recombee.com/blog/insights-the-next-level-of-analytics-in-recombee-ui) Insights, the analytics section of our Admin UI, offers various predefined and fully customizable reports to track recommended items and how users interact with these recommendations. ![](https://www.recombee.com/img/blog/authors/ondrejfiedler.png) Ondrej Fiedler May 09, 2024 New Features Recommendation Engine [![](https://www.recombee.com/img/blog/elevate-your-personalization-strategy-with-recombees-innovative-features.png)](https://www.recombee.com/blog/elevate-your-personalization-strategy-with-recombees-innovative-features) ### [Elevate Your Personalization Strategy with Recombee's Innovative Features](https://www.recombee.com/blog/elevate-your-personalization-strategy-with-recombees-innovative-features) The digital landscape and customer preferences and behavior are changing faster than ever now. To help our clients stay on top of the game, our team has focused on developing innovative features... ![](https://www.recombee.com/img/blog/authors/janvaluch.png) Jan Valuch Mar 13, 2024 New Features Recommendation Engine [![](https://www.recombee.com/img/blog/modern-recommender-systems-part-2-data.png)](https://www.recombee.com/blog/modern-recommender-systems-part-2-data) ### [Modern Recommender Systems - Part 2: Data](https://www.recombee.com/blog/modern-recommender-systems-part-2-data) Data used by modern recommenders and how we can measure progress towards goals. ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Mar 07, 2024 Recommendation Engine Personalization --- # Recombee Partners with The Telegraph to Deliver AI-Driven Personalisation to Millions of Readers > Source: https://www.recombee.com/blog/recombee-partners-with-the-telegraph-to-deliver-ai-driven-personalisation-to-millions-of-readers > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # Recombee Partners with The Telegraph to Deliver AI‑Driven Personalisation to Millions of Readers ![](https://www.recombee.com/img/blog/authors/recombee.png) Recombee Oct 02, 2025 ![](https://www.recombee.com/img/blog/recombee-partners-with-the-telegraph-to-deliver-ai-driven-personalisation-to-millions-of-readers/main.png) **Prague, 11th September 2025 – Recombee**, a leading AI-powered recommendation platform, has announced a strategic partnership with **The Telegraph**, one of the United Kingdom’s most respected media brands. The collaboration will enable real-time, personalised content discovery for millions of readers across The Telegraph’s digital platforms, reinforcing its focus on innovation and reader-centric publishing. Recombee’s technology will power personalised content recommendations across **The Telegraph’s website and app**, surfacing the most relevant stories for each individual based on reading history, engagement patterns, and editorial goals. With this integration, The Telegraph aims to **enrich the reader experience**, improve engagement, and deliver a **highly personalised content journey** while staying true to its values. This solution connects each reader’s uniqueness with the breadth of The Telegraph’s journalism. From the first click, readers will be presented with stories that match their interests, writers they follow, and new articles similar to ones they’ve enjoyed before. **Combined with editorial expertise**, real-time insights, and diversification-focused models, readers will also be introduced to perspectives they may not otherwise encounter. _“At The Telegraph, driving deeper engagement with our journalism through sophisticated personalisation is a core part of our AI strategy,”_ said **Tom Kelleher, Director of Technology at The Telegraph**. _“We’re excited about the potential of Recombee’s capabilities to integrate with and enhance our editorial processes, offering new ways to surface relevant content and understand performance trends in real time.”_ The partnership supports The Telegraph with data-driven tools for smarter content planning and onward journey optimisation. With features such as semantic content segmentation and multimodal analysis of both text and images, Recombee delivers actionable insights into user behavior and content performance across digital channels. _“We’re excited to work with one of the most iconic names in journalism to build a more personalised, reader-centric experience,”_ said **Gabriela Takacova, Co-Founder of Recombee**. _“Our goal is to empower The Telegraph’s editorial teams with intelligent AI tools that enhance engagement while upholding the quality and integrity their brand represents.”_ Company News ## Next Articles [![](https://www.recombee.com/img/blog/ai-news-and-outlook-for-2026.png)](https://www.recombee.com/blog/ai-news-and-outlook-for-2026) ### [AI News and Outlook for 2026](https://www.recombee.com/blog/ai-news-and-outlook-for-2026) Here’s what caught my attention in AI research lately, and where things might be heading in 2026\. After 25+ years in this field, the pace has gotten hard to keep up with. I’m trying to make sense of... ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Dec 17, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/modern-recommender-systems-part-3-objectives.png)](https://www.recombee.com/blog/modern-recommender-systems-part-3-objectives) ### [Modern Recommender Systems - Part 3: Objectives](https://www.recombee.com/blog/modern-recommender-systems-part-3-objectives) Learning objectives of recommender systems and personalized search. ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Sep 03, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/the-building-blocks-of-privacy-friendly-personalization.png)](https://www.recombee.com/blog/the-building-blocks-of-privacy-friendly-personalization) ### [The Building Blocks of Privacy-Friendly Personalization](https://www.recombee.com/blog/the-building-blocks-of-privacy-friendly-personalization) Personalization can be achieved without compromising user privacy. While many personalization systems have historically relied on practices now considered intrusive, like... ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Aug 07, 2025 Recommendation Engine Personalization --- # Recombee Real-Time AI Recommendations in Segment | Blog > Source: https://www.recombee.com/blog/recombee-real-time-ai-recommendations-as-the-new-destination-in-segment > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # Recombee Real-Time AI Recommendations as the New Destination in Segment ![](https://www.recombee.com/img/blog/authors/adelasloupenska.png) Adela Sloupenska Mar 05, 2024 Segment has enabled its users to enjoy Recombee personalization services without the need to leave their platform and with minimum coding involved. With a few simple clicks, domains using Segment can upgrade their services offered through [Recombee as Segment’s Destination](https://segment.com/integrations/recombee-ai/) and maximize the digital experience for their customers. ![](https://www.recombee.com/img/blog/recombee-real-time-ai-recommendations-as-the-new-destination-in-segment.png) ## Recombee + Segment: Personalization Services in the Customer Data Platform To ensure premium services and the most accurate recommendations to a wider audience, Recombee can now be easily integrated through the Segment platform, available to its users as an additional feature. **[Segment](https://segment.com/) is a popular Customer Data Platform (CDP)** that simplifies data collection from digital properties, such as websites or apps. While Segment oversees data collection and has a strong track record with clients such as IBM and Domino’s, Recombee ensures tailored recommendations and personalization services on the client’s web or app. **What makes Recombee a great Segment destination?** Customer satisfaction is the driving force for customer retention, and personalization is key in decreasing churn rates. Our engine utilizes machine learning algorithms to generate [product](https://www.recombee.com/product-recommendations) and [content recommendations](https://www.recombee.com/content-recommendations) specific to every user's likes. Recombee analyzes the user’s onsite behavior from the very first click, together with text descriptions, images, and other metadata. This gives the user a tailored one-on-one experience that adjusts according to their real-time changes in preferences. This personalized approach to individual users increases customer satisfaction and improves the site’s competitiveness. ## How to Fully Leverage Recombee - Segment Integration ![](https://www.recombee.com/img/blog/recombee-real-time-ai-recommendations-as-the-new-destination-in-segment-2.png) **Personalization can be applied to any digital domain and can help editors enhance their on-site offering for individual users.** Domains for online shops, VoD, media houses, marketplaces, real estate - across all sectors, web platforms using personalization from an AI-powered recommender typically report increased conversion rates, higher traffic, greater user engagement, and fuller shopping carts. **This newly formed partnership makes the integration of Recombee very easy.** While Segment users enjoy using services such as collection, transformation, sending, and archiving of first-party customer data, with just a few extra clicks Recombee Destination can be added allowing Segment to send its data to the Recombee engine. **Start at your Segment platform and select ‘Recombee’ as the destination.** Follow the [Recombee Segment recipe](https://segment.com/recipes/increase-conversions-personalizing-experience-recombee/) and [Recombee AI Documentation](https://segment.com/docs/connections/destinations/catalog/recombee-ai/) steps that allow Segment to send previously collected interaction data. These are then used in the Recombee engine to generate recommendations for the Segment/Recombee clients. Data supplied to the engine include real-time and historical interactions such as purchases, views, cart additions, bookmarks, likes, and also view portions. The ingested interactions can then be monitored in the Recombee KPI console. ## Key Take away Points ![](https://www.recombee.com/img/blog/recombee-real-time-ai-recommendations-as-the-new-destination-in-segment-3.png) **Using this extended service is highly beneficial for online vendors offering extensive product catalogs or content on websites or apps.** Users entering sites with vast offerings can easily be overwhelmed by the sheer number of options to choose from. This is where Recombee comes into its own. It creates an organized space in which the user can find what they are looking for. In real-time, from the very first click, Recombee analyses and responds to the users’ changing preferences to give them the right personalized recommendations every time. Are you a Segment customer keen on trying Recombee personalization services? We are happy to assist with any general inquires at [business@recombee.com](mailto:business@recombee.com) or provide integration support at [support@recombee.com](mailto:support@recombee.com). If you just want to learn more about how personalization can be applied to your use case, get inspired in our [Case Study](https://www.recombee.com/case-studies) section, and explore the application of AI recommendations in various domains. Personalization Integrations Partnerships ## Next Articles [![](https://www.recombee.com/img/blog/modern-recommender-systems-part-2-data.png)](https://www.recombee.com/blog/modern-recommender-systems-part-2-data) ### [Modern Recommender Systems - Part 2: Data](https://www.recombee.com/blog/modern-recommender-systems-part-2-data) Data used by modern recommenders and how we can measure progress towards goals. ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Mar 07, 2024 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/is-this-comment-useful-enhancing-personalized-recommendations-by-considering-user-rating-uncertainty.png)](https://www.recombee.com/blog/is-this-comment-useful-enhancing-personalized-recommendations-by-considering-user-rating-uncertainty) ### [Is This Comment Useful? Enhancing Personalized Recommendations by Considering User Rating Uncertainty](https://www.recombee.com/blog/is-this-comment-useful-enhancing-personalized-recommendations-by-considering-user-rating-uncertainty) Picture this: you're on the hunt for the perfect new smartphone, browsing through your favourite online electronics store. The online store’s recommendation engine pops up with what it thinks could be your possible next gadget love... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Mar 01, 2024 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/recombeelabs-2023-research-publications.png)](https://www.recombee.com/blog/recombeelabs-2023-research-publications) ### [Recombeelab's 2023 Research Publications](https://www.recombee.com/blog/recombeelabs-2023-research-publications) Recombeelab, a joint research laboratory of Recombee and the Faculty of Information Technology at the Czech Technical University in Prague, experienced a highly productive year in 2023, publishing a series of insightful and impactful papers in the field of recommendation systems. ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Jan 19, 2024 Recommendation Engine Personalization --- # Recombee Research 2024 > Source: https://www.recombee.com/blog/recombee-research-2024 > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # Recombee Research 2024 ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Feb 23, 2025 ![](https://www.recombee.com/img/blog/recombee-research-2024/main.png) ## Academic Foundations Recombee has always been deeply connected to academia, with four of our six co-founders holding Ph.D. degrees. Over the years, our investment in research has grown alongside our company and the increasing demand for advanced features in the market. This commitment is exemplified by the achievements of **RecombeeLab**, our joint research laboratory with the Faculty of Information Technology at the Czech Technical University in Prague. In 2024, RecombeeLab had another highly productive year. Beyond this collaboration, we’ve built an internal team dedicated to transitioning innovations from academia into practical applications within our recommender systems. This focus has allowed us to effectively integrate cutting-edge research into products that deliver real impact for our clients, aligning with the latest AI trends. ## Breakthroughs in Recommendation Systems ### Introducing beeFormer ![](https://www.recombee.com/img/blog/recombee-research-2024/01.png) One of our standout innovations of 2024 is [beeFormer](https://www.recombee.com/blog/introducing-beeformer-a-framework-for-training-foundational-models-for-recommender-systems), an open-source sentence transformer designed to meet diverse generative AI needs. Presented at **ACM RecSys 2024**,\[1\] beeFormer combines transformer architecture with a unique methodology for distilling interaction data. This approach improves natural language model capabilities for recommendation tasks, such as item segmentation and semantic search, paving the way for greater precision and flexibility. [Explore beeFormer](https://www.recombee.com/blog/introducing-beeformer-a-framework-for-training-foundational-models-for-recommender-systems) ### Understanding Trends and Regional Interactions ![](https://www.recombee.com/img/blog/recombee-research-2024/02.png) At **WWW 2024** in Singapore,\[2\] we introduced a method to assess shifts in item popularity over time, helping decode users’ temporal behaviors and detect changing trends in catalogs. Our **regionalization-based interactions algorithm**\[3\] identifies optimal warehouse locations and other logistics opportunities based on user-item interaction patterns. ### Generalization and Cognitive Science in AI ![](https://www.recombee.com/img/blog/recombee-research-2024/03.png) Our research efforts extend to understanding how personalized machine learning models generalize. At **ICML 2024** in Vienna, we presented a study analyzing the generalization mechanisms of deep factorization methods,\[4\] which are widely used in recommendation systems. We’ve also pioneered cognitive science modeling within the recommender domain.\[5\] One of our studies explores users' decision-making processes when interacting with item lists. Using similar methodologies, we analyzed how large language models (LLMs), such as GPT models, perceive object concepts.\[6\] This research reveals how certain LLMs adapt to human-like reasoning after fine-tuning, aiding in model selection and ethical AI development. ## The Impact of Our Research ### Prestigious Publications and Presentations ![](https://www.recombee.com/img/blog/recombee-research-2024/04.png) In 2024, our work appeared in high-impact journals like **IEEE Transactions on Neural Networks and Learning Systems** and **ACM Transactions on Intelligent Systems and Technology**. We also presented papers at leading conferences, including **ACM RecSys** and **ICML**, one of the "big three" AI conferences alongside NeurIPS and ICLR. ### Sponsorship of ACM RecSys 2024 ![](https://www.recombee.com/img/blog/recombee-research-2024/05.png) As part of our dedication to the field, we proudly sponsored **RecSys 2024**. Our researchers were active participants, presenting papers, sharing innovations, and contributing to discussions that shape the future of recommendation systems. ## What’s Next? ### Exciting Opportunities in 2025 ![](https://www.recombee.com/img/blog/recombee-research-2024/06.png) Research will remain central to Recombee’s mission as we continue to push the boundaries of AI. A key highlight of the year will be **ACM RecSys 2025**, hosted in Prague which is also home to Recombee’s headquarters. Our team is looking forward to contributing to the organization of the event and exploring new ideas to advance the future of recommendation technologies. [Explore RecSys 2025 in Prague](https://recsys.acm.org/recsys25/) ## Collaborate with Us ![](https://www.recombee.com/img/blog/recombee-research-2024/07.png) At Recombee, we value partnerships with researchers and innovators. If you’re interested in our work or exploring potential collaborations, reach out! We’re always open to exchanging ideas and advancing research together. Contact us at [research@recombee.com](mailto:research@recombee.com). Recommendation Engine Personalization ## Resources \[1\] Vancura, Vojtech, Pavel Kordik, and Milan Straka. "beeFormer: Bridging the Gap Between Semantic and Interaction Similarity in Recommender Systems." In Proceedings of the 18th ACM Conference on Recommender Systems, 2024 \[2\] Alves, Rodrigo, Antoine Ledent, Renato Assunção, Pedro Vaz-De-Melo, and Marius Kloft. "Unraveling the Dynamics of Stable and Curious Audiences in Web Systems." In Proceedings of the ACM on Web Conference, 2024 \[3\] Alves, Rodrigo. "Regionalization-based Collaborative Filtering: Harnessing Geographical Information in Recommenders." ACM Transactions on Spatial Algorithms and Systems, 2024 \[4\] Ledent, Antoine, and Rodrigo Alves. "Generalization analysis of deep nonlinear matrix completion." Proceedings of the International Conference on Machine Learning (ICML), 2024 \[5\] Alves, Rodrigo, and Antoine Ledent. "Context-Aware REpresentation: Jointly Learning Item Features and Selection From Triplets." IEEE Transactions on Neural Networks and Learning Systems, 2024 \[6\] Hrytsyna, Anastasiia, and Rodrigo Alves. "From Representation to Response: Assessing the Alignment of Large Language Models with Human Judgment Patterns." ACM Transactions on Intelligent Systems and Technology, 2024 (To Appear) ## Next Articles [![](https://www.recombee.com/img/blog/are-you-here-to-stay-unraveling-the-dynamics-of-stable-and-curious-audiences-in-web-systems.png)](https://www.recombee.com/blog/are-you-here-to-stay-unraveling-the-dynamics-of-stable-and-curious-audiences-in-web-systems) ### [Are You Here to Stay? Unraveling the Dynamics of Stable and Curious Audiences in Web Systems](https://www.recombee.com/blog/are-you-here-to-stay-unraveling-the-dynamics-of-stable-and-curious-audiences-in-web-systems) Why do influencers frequently request their subscribers to enable all notifications for their channels? This practice stems from their awareness that not all subscribers are regular... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Mar 03, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/key-trends-in-news-and-media-for-2025-and-the-role-of-ai-powered-solutions.png)](https://www.recombee.com/blog/key-trends-in-news-and-media-for-2025-and-the-role-of-ai-powered-solutions) ### [Key Trends in News & Media for 2025 and the Role of AI-Powered Solutions](https://www.recombee.com/blog/key-trends-in-news-and-media-for-2025-and-the-role-of-ai-powered-solutions) The way news is produced, delivered, and consumed keeps shifting, and as we move through 2025, a few big changes have become impossible to ignore. ![](https://www.recombee.com/img/blog/authors/annapetr.png) Gabriela Takacova & Violeta Milarova Feb 20, 2025 Personalization [![](https://www.recombee.com/img/blog/2024-wrap-up.png)](https://www.recombee.com/blog/2024-wrap-up) ### [2024 Wrap-Up](https://www.recombee.com/blog/2024-wrap-up) As we wrap up 2024, let’s take a moment to celebrate the milestones we’ve achieved together. Your collaboration and feedback have been at the heart of our progress. Here’s a quick recap of last year’s highlights. ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Jan 27, 2025 New Features Personalization --- # Analysing Training Data Using RepSys | Blog > Source: https://www.recombee.com/blog/repsys-opensource-library-for-interactive-evaluation-of-recommendation-systems > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # Visual and Interactive Evaluation of Recommender Systems ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Sep 19, 2022 When building modern real-world artificial intelligence systems, it is increasingly important to validate that the system works correctly. This is however not an easy task. Existing tools for machine learning practitioners are bad in providing insights into **how AI models work for individual groups of end users.** This is particularly important in **detecting and mitigating AI bias** of most widespread AI models: **recommender systems.** We have built and open sourced [RepSys](https://github.com/cowjen01/repsys) a tool for interactive evaluation of recommender systems to help the machine learning community to build better models. Here we demonstrate what you can achieve with our tool. We have selected a dataset well known to the community, but you can easily replace it by your datasets and your models. ## Recommending Books (GoodReads-10k) Most popular machine learning tools such as Jupyter notebooks offer tools for basic exploratory data analysis, model training and evaluation. Typical example is presented in this [notebook.](https://github.com/Spartee/Book-recommendation-system/blob/master/goodreads-book-recsys-kaggle-v1.ipynb) ### Analysing Training Data Using RepSys Our library projects the rating (interaction) matrix into two dimensions revealing similarities among items (columns) or users (rows). Then you can use attributes of items to explore e.g. where clusters of older books are located in the projection. The same way you can find clusters of users that enjoy reading old books. When you select a cluster, you can look at most frequent values of attributes or distribution of numerical values. ![](https://www.recombee.com/img/blog/repsys-opensource-library-for-interactive-evaluation-of-recommendation-systems/1.png) Using this approach, you can get a really good understanding of the item latent space and how individual clusters of items are related and separated. ### Analysing Models Using RepSys When you do not see any flaws in user-item interactions, you can proceed and build some recommendation models. Typically the dataset is divided to the train and test sets and you can observe the metrics aggregated over all test users. As you see, on this dataset, the best performing model in terms of recall is EASE followed by KNN. It is also useful to look at other criteria as well, for example APL that states for the average percentage of long tail items recommended by model. The higher value, the better. Bestseller model (POP) is recommending just popular items, and non-personalized Random model (RAND) suggests lots of bizzare long tail items. EASE is able to offer both high recall and APL which is challenging. Single number such as an average or median is not enough to understand how models work for particular users and items. RepSys can visualize distribution of all metrics and relate it to specific clusters in the latent space. As the image shows, EASE can recommend particularly well to users in upper clusters (paranormal, recent fiction books), because they are better predictable. What is even more interesting is to compare two different models (e.g. EASE and POP) and observe the relative difference. Clearly, the Bestseller model (POP) is **strongly biased** towards a niche cluster of readers enjoying paranormal books. They are not popular enough to be recommended by this model. On the other hand, EASE can make these users very satisfied. In this way, you can tune hyperparameters of your models to get reasonable performance for all user groups. ![](https://www.recombee.com/img/blog/repsys-opensource-library-for-interactive-evaluation-of-recommendation-systems/2.png) For the GoodReads dataset, the optimal number of neighbors for the user based recommendation is around 50\. We can investigate what will happen, if this hyperparameter is set too low (overfitted model) or too high (underfitted model). We have chosen the APL@10 metric telling us what is the percentage of long tail items in 10 most relevant items recommended by a model. Apparently, 1NN model recommends too many long-tail items to mainstream users in the center of the latent space who probably prefer more popular items. On the other hand, the 5000NN model is not capable of recommending long tail items at all. When compared to 50NN, the error (bias) of this model will be much higher on niche users in the surroundings. Note that green users have a similar proportion of long tail items recommended as by the model with K close to optimum so their recall is also good. ![](https://www.recombee.com/img/blog/repsys-opensource-library-for-interactive-evaluation-of-recommendation-systems/3.png) The error of overfitted model (1NN) is higher for almost all users — the model often recommends bizzare items to niche users and it does not recommend popular items much — harming its performance for mainstream users. Underfitted model (5000NN) correctly recommends popular content for the mainstream and the main source of its error is the bias towards niche clusters in the surroundings of the latent space. ### Comparing Recommendations Using RepSys The relative performance on a cluster of users or items is still just a number. Offline evaluation will always be biased. It is impossible to predict what would happen if a user was exposed to recommendations suggested by an evaluated model, which are not present in offline training data. For this reason, RepSys supports you in comparing recommendations of multiple models for individual users from the training set and for simulated users in the interactive mode. You can click on the suggested items and instantly observe how the recommendations change. [![](https://www.recombee.com/img/blog/repsys-opensource-library-for-interactive-evaluation-of-recommendation-systems/4.png)](https://repsys.recombee.net/) Explore [RepSys Demo here](https://repsys.recombee.net/) As you can see, after two interactions, the KNN model still recommends the bestselling Hunger Games book in the first position while EASE is able to suggest niche content that seems to be more relevant. If you like your work, cite us, give us feedback. We are also looking for contributors to the [RepSys library.](https://github.com/cowjen01/repsys) Recommendation Engine ## Next Articles [![](https://www.recombee.com/img/blog/how-we-are-using-ai-to-power-content-recommendations.png)](https://revium.com.au/blog/how-we-are-using-ai-to-power-content-recommendations) ### [How We Are Using AI to Power Content Recommendations](https://revium.com.au/blog/how-we-are-using-ai-to-power-content-recommendations) In this article we walk you through how we are using the AI recommendation engine Recombee embedded in our headless CMS StoryBlok to drive content recommendations throughout our own website. ![](https://www.recombee.com/img/blog/authors/revium.png) Revium Sep 30, 2022 Partnerships [![](https://www.recombee.com/img/blog/real-time-personalization-of-content-with-ai-powered-recommendations.png)](https://www.recombee.com/blog/real-time-personalization-of-content-with-ai-powered-recommendations) ### [Real-Time Personalization of Content With AI-Powered Recommendations](https://www.recombee.com/blog/real-time-personalization-of-content-with-ai-powered-recommendations) Do you manage a publishing company, online gaming platform, or a streaming site with a content-heavy catalog and are thinking about how to improve the user experience? ![](https://www.recombee.com/img/blog/authors/karenharazimova.png) Karen Harazimova Sep 16, 2022 Personalization [![](https://www.recombee.com/img/blog/ai-powered-content-recommendations-with-a-headless-cms.png)](https://www.luminary.com/blog/ai-content-recommendations-headless-cms-recombee) ### [AI-Powered Content Recommendations With a Headless CMS](https://www.luminary.com/blog/ai-content-recommendations-headless-cms-recombee) Thanks to its API-first nature, it is quite straightforward to integrate your headless CMS with the most powerful AI-powered content recommendations available on the market. Luminary just did that with their own website, Kontent.ai and Recombee. ![](https://www.recombee.com/img/blog/authors/andythompson.png) Andy Thompson (Luminary) Aug 31, 2022 Partnerships --- # SHIELD: The Universal Framework Making AI Search Safer for Everyone > Source: https://www.recombee.com/blog/shield-the-universal-framework-making-ai-search-safer-for-everyone > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # SHIELD: The Universal Framework Making AI Search Safer for Everyone ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Jun 24, 2025 ![](https://www.recombee.com/img/blog/shield-the-universal-framework-making-ai-search-safer-for-everyone/main.png) _Presented at UMAP 2025 in New York City_ Imagine searching for "glass tubing" and getting recommendations for drug manufacturing equipment. As AI-powered search becomes ubiquitous — from online marketplaces to social networks — the stakes for getting it wrong have never been higher. ## The Rise of Natural Language Expectations The success of ChatGPT and similar conversational AI has fundamentally changed user behavior. People have learned that machines can understand natural language, and they're applying this expectation everywhere. Instead of typing "red shoes size 8," users now search with complex, conversational queries like "comfortable red shoes for my evening work commute that won't hurt after 10 hours." ![](https://www.recombee.com/img/blog/shield-the-universal-framework-making-ai-search-safer-for-everyone/1.png) Companies like Linked-in are convincing users that they can use natural language queries in their digital products. This shift is forcing online services to rapidly enhance their capabilities. Traditional keyword-based search feels primitive when users expect semantic understanding. Companies are turning to advanced solutions like **Recombee's semantic search** to meet these new expectations, enabling their platforms to interpret intent rather than just match words. But here's the challenge: when your catalog contains sensitive or potentially harmful items — from chemicals and tools to adult content — natural language queries can lead users (and your recommendation algorithms) into dangerous territory. A conversational query might seem innocent but actually be seeking something problematic. **This is where alignment becomes critical.** Our research proposes a systematic approach to ensure that semantic search systems understand not just what users are asking for, but whether they should be helped to find it. ![](https://www.recombee.com/img/blog/shield-the-universal-framework-making-ai-search-safer-for-everyone/2-1.png) Presenting SHIELD at UMAP 2025, a paper by [Filip Spacek](https://www.linkedin.com/in/filip-spacek/), [Vojtech Vancura](https://www.linkedin.com/in/vojtech-vancura/) and [Pavel Kordik](https://www.linkedin.com/in/kordik/) That's why we built **SHIELD**: a framework that teaches AI systems to recognize harmful, sensitive, and safe queries before they cause damage. ## The Hidden Dangers of Semantic Search Semantic search has transformed digital interactions by understanding natural language rather than relying on exact keywords. This makes it powerful for guiding users to relevant content — but also risky. In environments like marketplaces or social networks, ambiguous or malicious queries can trigger harmful or offensive recommendations. Over-filtering sensitive content may alienate users seeking it intentionally, while under-filtering risks exposing others to unsafe material. Without safeguards, even well-meaning AI systems can inadvertently cause harm. ## What Makes SHIELD Different **SHIELD** (Semantic Harmful-content Identification and Ethical Labeling Dataset) isn't just another content moderation tool — it's a complete methodology for building ethical AI systems from the ground up. SHIELD is not a static dataset but rather a **methodology** for constructing and curating labeled data for training classification models. The methodology involves three key stages: ### 1\. Hierarchical Query Generation The process begins by defining a **hierarchical structure of content types** relevant to the intended application. For instance, this may involve: * A set of **broad categories** (e.g., safety violations, legal issues, misinformation) * A corresponding set of **fine-grained subcategories** Next, for each subcategory, **large language models** are used to generate realistic, diverse **query examples**. Typically, a fixed number of candidate queries are produced per subcategory (e.g., 20 per subcategory), ensuring sufficient breadth and coverage. ### 2\. Quality Scoring and Filtering with a Reward Model Not all generated queries are equally informative or representative. To refine the dataset: * Each query is evaluated using a **reward model** (such as Skywork-Reward), which assigns a relevance or quality score. * Based on these scores, only the top-ranked examples are **retained for downstream use.** This filtering step ensures: * Higher **semantic clarity** and **intent precision** * Better **separation between target classes** * Reduced labeling noise in the final dataset Importantly, the exact number of categories and classes can vary between implementations. In one use case, a **three-class system** (safe, sensitive, harmful) was used — but the SHIELD methodology can easily be extended to more classes based on the application’s ethical and operational needs. ### 3\. Training and Deployment The curated dataset is then used to train AI models capable of classifying incoming queries based on learned semantic and ethical signals. These models can be integrated into moderation pipelines, chatbot backends, or safety layers for various applications. ## Use Case: Moderating User Input in Online Marketplaces To evaluate SHIELD in a real-world context, we applied the methodology to the**moderation of user-generated queries in an online marketplace** — a setting where misuse can take many forms: scams, illegal activity, or inappropriate communication. ![](https://www.recombee.com/img/blog/shield-the-universal-framework-making-ai-search-safer-for-everyone/3.png) Example of SHIELD powered detection in online marketplace environment The dataset used for this evaluation was built through structured generation and then filtered using a reward model to retain only high-confidence examples. After this filtering process, the final dataset comprised 17,170 queries classified as harmful and 8,871 as sensitive, selected from an initially much larger set of generated examples. These queries reflect realistic search behavior in marketplace contexts and are cleanly separated into well-aligned classes suitable for model training. Using this dataset, three classification approaches were tested. The first, based on BM25 keyword similarity, achieved 93.2% accuracy, offering a computationally light yet reasonably effective solution. The second approach, which employed semantic embeddings with FAISS for nearest neighbor classification, improved performance to 96.5%, demonstrating the benefit of deeper contextual understanding. The highest accuracy — 98.4% with a 98.6% F1-score — was obtained using MoralBERT, a fine-tuned transformer model trained directly on the SHIELD dataset. This method provided the most robust generalization, especially in handling subtle or adversarial queries, though it came with higher computational costs. These results demonstrate that SHIELD can provide a strong foundation for query-level content moderation in semantic search systems, particularly when dealing with complex ethical boundaries in commercial platforms. See more in our research [paper](https://dl.acm.org/doi/10.1145/3699682.3728329) \[1\]. ## Extending SHIELD to Other Domains: Ideas and Examples The strength of SHIELD lies in its **flexibility** — its modular pipeline for query generation, filtering, and classification can be adapted to many real-world applications. Below are several domains where SHIELD can be used, along with examples of potential **category structures** and **alignment objectives** for each. ### Applications Beyond Marketplaces: Domain-Specific Examples 🏥 **Healthcare & Wellness** SHIELD can support chatbots and digital health tools by identifying ethically risky queries such as unsafe medical advice (e.g., self-medicating, harmful home remedies), misdiagnosis risks (ambiguous or misleading symptoms), mental health crises (suicidal ideation, self-harm), and misinformation (vaccine myths, dietary pseudoscience). The goal is to flag or escalate such queries for human oversight when necessary. 🏛️ **Government and Civic Tech** Civic platforms benefit from filtering legally ambiguous queries (e.g., exploiting legal loopholes), threats, hate speech, and manipulative content (e.g., conspiracy-laden inputs). SHIELD can promote civil discourse and policy compliance while maintaining trust in public services. 🎓 **Educational Technology** AI tutors can detect academic dishonesty (e.g., cheating requests), cyberbullying, and inappropriate content (e.g., explicit jokes). SHIELD helps reinforce ethical learning behavior and safe classroom interactions. 💼 **Enterprise Security & Compliance** For internal tools, SHIELD can flag policy violations (e.g., bypass attempts, leaking sensitive data), toxic communication (e.g., harassment), and compliance risks (e.g., GDPR misuse). These capabilities aid in early intervention and internal risk management. 🧠 **Mental Health & Wellbeing** Self-help apps can use SHIELD to identify crisis signals (suicidal language, hopelessness), interpersonal distress (abuse, trauma), and harmful coping patterns (substance abuse, isolation), enabling appropriate referrals and support. 📰 **Media and News Moderation** Comment sections can be moderated by detecting inflammatory language (e.g., incitement, hate speech), disinformation (false claims, conspiracies), and manipulation tactics (deep fakes, revisionism). SHIELD can help maintain healthy discourse and safeguard editorial integrity. SHIELD enables developers to design **domain-specific query classifiers** by simply adjusting: 1. **The taxonomic structure** (categories and subcategories) 2. **The query generation prompts** 3. **The reward model filtering criteria** 4. **The downstream model training and deployment strategy** This adaptability means SHIELD can serve as a **building block** for creating AI systems that understand and enforce ethical boundaries — whether in education, healthcare, public policy, or enterprise. ## Beyond Content Moderation: Building Trust SHIELD represents a fundamental shift in how we think about AI safety. Rather than reactive content filtering, it enables **proactive alignment** — training systems to understand ethical boundaries before they encounter real users. This approach builds something even more valuable than safety: **trust**. When users know an AI system has been designed with their wellbeing in mind, adoption increases and satisfaction soars. ## Open Source, Open Model, Ready to Deploy The entire SHIELD framework is available for immediate use: * 📦 **Dataset & Code**: [github.com/flpspacek/SHIELD](https://github.com/flpspacek/SHIELD) * 🤖 **Pre-trained Model**: [huggingface.co/spacefi1/moralBERT](https://huggingface.co/spacefi1/moralBERT) Whether you're building the next generation of search engines, virtual assistants, or specialized AI tools, SHIELD provides the ethical foundation your users deserve. ## The Bottom Line Every AI system that interacts with users in natural language needs ethical guardrails. SHIELD makes implementing those guardrails not just possible, but practical and scalable. The question isn't whether you need content moderation for your AI system—it's whether you can afford to deploy without it. **Ready to build safer AI?** Start with SHIELD. _This research was supported by the FIT CTU Student Research Program, Recombee, and the PoliRuralPlus project (EU Grant 101136910)._ ## References \[1\] F. Spacek, V. Vancura, and P. Kordik, “Mitigating Risks in Marketplace Semantic Search: A Dataset for Harmful and Sensitive Query Alignment,” in Proceedings of the 33rd ACM Conference on User Modeling, Adaptation and Personalization, 2025, pp. 329–334\. doi: 10.1145/3699682.3728329. Recommendation Engine Personalization ## Next Articles [![](https://www.recombee.com/img/blog/how-regionalization-based-recommendations-can-improve-your-operations.png)](https://www.recombee.com/blog/how-regionalization-based-recommendations-can-improve-your-operations) ### [How Regionalization-Based Recommendations Can Improve Your Operations](https://www.recombee.com/blog/how-regionalization-based-recommendations-can-improve-your-operations) From ancient trade routes to modern urban planning, geography has consistently shaped human decisions and opportunities. Today, in the world of online business and personalized recommendations... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Jul 18, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/making-recommendations-fairer-a-new-way-to-guarantee-exposure-for-all.png)](https://www.recombee.com/blog/making-recommendations-fairer-a-new-way-to-guarantee-exposure-for-all) ### [Making Recommendations Fairer: A New Way to Guarantee Exposure for All](https://www.recombee.com/blog/making-recommendations-fairer-a-new-way-to-guarantee-exposure-for-all) As recommender systems become more widespread across digital platforms, concerns around fairness are coming to the forefront. Standard relevance-based ranking techniques, while effective... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Apr 29, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/2025-sneak-peek.png)](https://www.recombee.com/blog/2025-sneak-peek) ### [2025 Sneak Peek](https://www.recombee.com/blog/2025-sneak-peek) This year is already off to an exciting start, and we’re rolling out new tools to improve efficiency and optimize recommendations. Here’s what’s available and what’s coming next. ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Mar 19, 2025 Recommendation Engine New Features --- # The Building Blocks of Privacy-Friendly Personalization > Source: https://www.recombee.com/blog/the-building-blocks-of-privacy-friendly-personalization > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # The Building Blocks of Privacy-Friendly Personalization ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Aug 07, 2025 Personalization can be achieved without compromising user privacy. ![](https://www.recombee.com/img/blog/the-building-blocks-of-privacy-friendly-personalization/main.png) While many personalization systems have historically relied on practices now considered intrusive, like third-party cookies, cross-site tracking, or opaque data sharing, Recombee has taken a different approach from the start. With increased scrutiny and the ongoing tightening of regulations like GDPR in Europe and CCPA in California, coupled with users becoming more conscious of what’s being collected and how it’s being used, organizations are under growing pressure to rethink how they handle personal data. In response, leading companies are shifting focus; not just on personalization, but on doing it in ways that respect privacy boundaries while still delivering clear value to users. As a result, organizations themselves are becoming more stringent in how they collect and apply data. So, what does a more privacy-conscious approach to personalization actually involve? This article outlines the key concepts, categories of data, and technical principles that make responsible personalization possible. It also explains how Recombee applies these principles to offer intelligent personalization without disregarding user trust. ## Understanding the Difference Between First-Party and Third-Party Data At the center of the privacy discussion is the question of who collects the data and how. Third-party data refers to information gathered by organizations that do not have a direct relationship with the user. A typical example would be advertising platforms that track user activity across multiple websites. For instance, after visiting a product page on one site, you might start seeing ads for that product on completely unrelated websites, often without knowing you were being tracked in the first place. This kind of tracking, usually powered by third-party cookies, raises privacy concerns because it happens behind the scenes and often without clear user consent. As a result, this category of data is becoming increasingly difficult to use, partly due to evolving browser standards and partly because of compliance requirements under laws such as the GDPR and CCPA. In contrast, first-party data is collected directly by the website or application a user is interacting with. This includes observed behavior such as clicks, searches, viewed content, or completed transactions. When collected transparently and with user consent, this data is typically more reliable and actionable. ## Defining What Constitutes “Safe” Data Effective personalization doesn’t always require knowing exactly who someone is. Instead, it can rely on recognizing patterns in how users interact with content, typically without attaching those patterns to real-world identities. Modern systems can operate effectively using the following types of data: * Interaction data, such as clicks or views * Purchase history or conversion events * Time spent on various types of content * Contextual metadata, for example, device type or content category * User-defined preferences, such as favorite genres When this data is collected with consent, processed in a privacy-conscious way, and linked using pseudonymous or session-based identifiers rather than personally identifiable information, it can still enable meaningful personalization while reducing privacy risks. The goal isn't to avoid identification altogether, but to avoid unnecessary exposure of real identities. ## How Event-Based Personalization Functions Recombee and similar platforms rely on explicitly defined events provided by the customer’s system, rather than passive tracking or background data collection. This means data is only processed when a user interacts with something, like playing a video, scrolling through content, or adding an item to their cart, and the event is sent to Recombee by the application itself. These actions are observable user behaviors, not silently collected signals. Over time, these events contribute to a behavioral profile based on intent and engagement. Crucially, this profile does not require knowing the user’s identity; it reflects what they do, not who they are. This approach keeps personalization focused on relevance, while respecting privacy boundaries and clearly aligning with GDPR definitions of a data processor. ## Evaluating the Relationship Between Privacy and Performance There is a common assumption that removing passive tracking reduces the effectiveness of personalization. In practice, the opposite can be true. When systems rely on recent, high-quality behavioral data rather than large but imprecise third-party profiles, the result is often more accurate. Additionally, when users believe that their data is being handled responsibly, they are more inclined to engage and to share preferences willingly. In this sense, privacy is not a limitation. It can become an operational advantage. ## How Transparency Contributes to Trust Responsible data practices affect more than just compliance outcomes. They shape how users perceive and engage with digital services. This includes: * Clearly communicating what data is collected and why * Providing options to adjust personalization preferences * Respecting user rights, such as the right to access or delete their data These measures contribute to user confidence. When individuals feel informed and in control of their data, they’re more likely to trust and continue using the service over time. ## Privacy by Design: Recombee’s Approach ![](https://www.recombee.com/img/blog/the-building-blocks-of-privacy-friendly-personalization/1.png) Recombee’s platform is built with privacy-respecting personalization as a foundational principle. * It does not utilize third-party cookies or background trackers * It works with clearly defined, user-initiated events, such as clicks, purchases, or explicitly given preferences * It processes data in real time to reflect immediate user behavior * It supports both short-term session data and longer-term user profiles, without tracking users across unrelated websites or applications This design enables personalization that’s both effective and aligned with modern privacy standards. Recommendation Engine Personalization ## Next Articles [![](https://www.recombee.com/img/blog/modern-recommender-systems-part-3-objectives.png)](https://www.recombee.com/blog/modern-recommender-systems-part-3-objectives) ### [Modern Recommender Systems - Part 3: Objectives](https://www.recombee.com/blog/modern-recommender-systems-part-3-objectives) Learning objectives of recommender systems and personalized search. ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Sep 03, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/no-code-search-widget-personalized-powerful-effortless.png)](https://www.recombee.com/blog/no-code-search-widget-personalized-powerful-effortless) ### [No-Code Search Widget: Personalized, Powerful, Effortless](https://www.recombee.com/blog/no-code-search-widget-personalized-powerful-effortless) At Recombee, we don't just excel at recommendations – we provide powerful full-text search capabilities too. Our Quick, No-Code Search Widget exemplifies this, offering a seamless... ![](https://www.recombee.com/img/blog/authors/ondrejfiedler.png) Ondrej Fiedler Jul 25, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/how-regionalization-based-recommendations-can-improve-your-operations.png)](https://www.recombee.com/blog/how-regionalization-based-recommendations-can-improve-your-operations) ### [How Regionalization-Based Recommendations Can Improve Your Operations](https://www.recombee.com/blog/how-regionalization-based-recommendations-can-improve-your-operations) From ancient trade routes to modern urban planning, geography has consistently shaped human decisions and opportunities. Today, in the world of online business and personalized recommendations... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Jul 18, 2025 Recommendation Engine Personalization --- # 2024 Wrap-Up > Source: https://www.recombee.com/blog/2024-wrap-up > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # 2024 Wrap-Up ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Jan 27, 2025 ![](https://www.recombee.com/img/blog/2024-wrap-up/main.png) As we wrap up 2024, let’s take a moment to celebrate the milestones we’ve achieved together. Your collaboration and feedback have been at the heart of our progress. Here’s a quick recap of last year’s highlights. **Last Year Was Full of Exciting Updates!** ## Features ![](https://www.recombee.com/img/blog/2024-wrap-up/01.png) ### Insights: Data-Driven Decisions Made Simple We introduced the **Insights** module within the Recombee Admin UI, empowering you with real-time analytics and customizable reports to track recommendation performance and user engagement. Tailor your data views with stacked charts, tables, and compound metrics, ensuring your insights are as specific and actionable as your business needs. [Explore Insights](https://www.recombee.com/features/real-time-analytics-insights) ![](https://www.recombee.com/img/blog/2024-wrap-up/02.png) ### News Logics: Smarter Content Delivery We’ve fine-tuned content personalization for your users with our new suite of specialized News Logics models. Whether it’s delivering **Daily News** for high-visibility content or offering **Personalized, Related,** and **Categories for You** recommendations, we’ve made sure your audience gets the most relevant, up-to-date content. [See News Logics Docs](https://docs.recombee.com/recommendation_logics#news) ## Integration ![](https://www.recombee.com/img/blog/2024-wrap-up/03.png) ### Kotlin SDK & Go SDK: Expanding Developer Tools We’re making it easier for you to integrate Recombee into your systems with our new SDKs. **Kotlin SDK:** Seamlessly integrate Recombee into Android apps with our [Kotlin SDK](https://github.com/recombee/kotlin-api-client) and [Demo App](https://github.com/recombee/android-demo), making it simpler for you to deliver personalized experiences on mobile. **Go SDK:** Build scalable, reliable applications easily with Recombee’s [Go SDK](https://github.com/recombee/go-api-client), optimizing your product’s performance. [Explore SDKs](https://docs.recombee.com/api_clients) ![](https://www.recombee.com/img/blog/2024-wrap-up/04.png) ### New Segment Destination Sending user interactions from [Segment](http://segment.com/) to Recombee has never been easier. Our new integration, built on Segment’s Destination Actions framework, streamlines sending events like views, purchases, and watch progress. With custom event mappings, you can define your rules and quickly validate them in the Segment web app, all while benefiting from enhanced functionality and flexibility. [Explore Segment Integration](https://docs.recombee.com/segment) ## Conferences & Research ![](https://www.recombee.com/img/blog/2024-wrap-up/05.png) ### Recombee Science 2024 was another notable year for our research team. We published work in high-impact journals like **IEEE TNNLS** and **ACM TIST**, presented at major conferences such as the **ACM Web Conference** and **ICML**, and had the opportunity to showcase our sentence transformer-based recommendation model beeFormer. We accelerated the integration of the latest research findings into our product. Research will continue to be key to our innovation, and we look forward to sharing our progress at [ACM RecSys 2025](https://recsys.acm.org/recsys25/) in Prague. ![](https://www.recombee.com/img/blog/2024-wrap-up/06.png) ### Conferences in 2024 Alongside our research contributions, we had the privilege of attending leading industry events like [NAB](https://www.nabshow.com/), [IBC](https://show.ibc.org/), and [INMA Media Innovation Week](https://www.inma.org/modules/event/2024MediaInnovationWeek/index.html). These events provided great opportunities to connect, exchange insights, and explore how we can keep innovating together. ## Community & Support ![](https://www.recombee.com/img/blog/2024-wrap-up/07.png) ### Empowering Communities and Changing Lives Together In our continued support of non-profits, Recombee has donated to [SONS / vodicipsi.cz](https://www.vodicipsi.cz/), [Linka Bezpeci](https://www.linkabezpeci.cz/), and [Czechitas](https://www.czechitas.cz/en). These contributions provide essential services, from aiding individuals with visual impairments to empowering women through subsidized tech courses. The Czechitas initiative alone will help 80-90 women gain the skills and confidence they need to open doors to new opportunities for themselves and their families. We’re proud to be part of this journey, fostering growth and inclusion through technology. ## Partnerships ![](https://www.recombee.com/img/blog/2024-wrap-up/08.png) ### Collaborating with Exceptional Partners We’ve had the privilege of working with extraordinary partners like [Axinom](https://www.axinom.com/), [Diagnal](https://www.diagnal.com/), [ExpertSender](https://expertsender.com/), [ModernTV](https://www.moderntv.eu/), [Conversion.gr](http://conversion.gr/), [Vianeos](https://vianeos.com/en/home-vianeos), and many others. Their collaboration has been instrumental in achieving shared success, driving innovation, and creating a lasting impact. Thank you for supporting us and being key to our success in 2024\. Here’s to reaching new milestones and making an even bigger impact together in the year ahead! ✨ [Explore Partnerships](https://www.recombee.com/partnerships) ## Interested in Exploring Our New Features in More Detail? ![](https://www.recombee.com/img/team/filip-hanus.png) Schedule a demo for a product tour and more information [filip.hanus@recombee.com](mailto:filip.hanus@recombee.com) New Features Personalization ## Next Articles [![](https://www.recombee.com/img/blog/key-trends-in-news-and-media-for-2025-and-the-role-of-ai-powered-solutions.png)](https://www.recombee.com/blog/key-trends-in-news-and-media-for-2025-and-the-role-of-ai-powered-solutions) ### [Key Trends in News & Media for 2025 and the Role of AI-Powered Solutions](https://www.recombee.com/blog/key-trends-in-news-and-media-for-2025-and-the-role-of-ai-powered-solutions) The way news is produced, delivered, and consumed keeps shifting, and as we move through 2025, a few big changes have become impossible to ignore. ![](https://www.recombee.com/img/blog/authors/annapetr.png) Gabriela Takacova & Violeta Milarova Feb 20, 2025 Personalization [![](https://www.recombee.com/img/blog/celestial-tiger-entertainment-launches-new-chinese-movie-app-cmgo-with-diagnal.png)](https://www.diagnal.com/cmgo/) ### [Celestial Tiger Entertainment launches new Chinese Movie app, CMGO, with DIAGNAL](https://www.diagnal.com/cmgo/) With Recombee’s AI-powered recommendation engine working with DIAGNAL Enhance, CMGO serves up personalised experiences for each viewer, driving engagement for the service. ![](https://www.recombee.com/img/blog/authors/diagnal.png) Diagnal Nov 14, 2024 Personalization Partnerships [![](https://www.recombee.com/img/blog/introducing-beeformer-a-framework-for-training-foundational-models-for-recommender-systems.png)](https://www.recombee.com/blog/introducing-beeformer-a-framework-for-training-foundational-models-for-recommender-systems) ### [Introducing beeFormer: A Framework for Training Foundational Models for Recommender Systems](https://www.recombee.com/blog/introducing-beeformer-a-framework-for-training-foundational-models-for-recommender-systems) In the fast-evolving world of recommender systems, understanding both how users interact with content and the actual content itself is crucial. Many existing recommender systems struggle to balance these two aspects... ![](https://www.recombee.com/img/blog/authors/vojtechvancura.png) Vojtech Vancura Oct 15, 2024 New Features Recommendation Engine --- # Partnerships That Open New Opportunities | Blog > Source: https://www.recombee.com/blog/advancing-your-career-in-artificial-intelligence-with-prg-ai-and-recombee > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # Advancing Your Career in Artificial Intelligence with prg.ai and Recombee ![](https://www.recombee.com/img/blog/authors/karenharazimova.png) Karen Harazimova Nov 21, 2021 At Recombee, we have always collaborated with academia — after all, five of our co-founders graduated from the Czech Technical University in Prague, one of the largest and oldest technical universities in Europe, and most of them hold a Ph.D. degree. We believe that education is a continual process, one that does not end when you get your diploma. Especially in AI, learning new things every day is the only way to keep pace with the dynamic field if you want to belong to the top league. That is why we maintain a strong research culture in the company and support our staff in their educational and research activities. ## Partnerships That Open New Opportunities In addition to our collaboration with the CTU, Recombee maintains a close relationship with [prg.ai](https://prg.ai/en/about-us/), a non-profit organization aiming to transform Prague into a significant European AI hub. Their vision is an ambitious one, and its realization requires, among other components, building a community of AI-educated young people who would use their knowledge to create solutions with beneficial impacts on the Czech economy and society. Being a part of this community helps Recombee reach opportunities that escalate individual career growth. Needless to say, Recombee’s CEO Pavel Kordik, one of prg.ai's co-founders, actively supports the initiative as we believe that academic programs should go hand in hand with practical, real use case projects led by companies. ## Career Growth With Recombee Let us tell you a Recombee success story that shows the journey of our own machine learning researcher Petr Kasalicky. We started working with Petr thanks to [a joint research laboratory](https://fit.cvut.cz/en/science-and-research/facilities/laboratories/8360-recombee-research-laboratory-recombeelab) between Recombee and the FIT CTU several years ago. ![](https://www.recombee.com/img/blog/advancing-your-career-in-artificial-intelligence-with-prg-ai-and-recombee-1.png) Soon after that, Petr applied to the [prg.ai Minor](https://prg.ai/minor/) program, which he successfully graduated in September 2021 as a member of the second cohort. The program offers a mix of bachelor and master courses that focus on an in-depth understanding of and applying the newly gained skills into practice. Petr was able to develop deep theoretical insights into several different machine learning techniques and thanks to the collaboration with Recombee, he was able to apply these insights in practice. ![](https://www.recombee.com/img/blog/advancing-your-career-in-artificial-intelligence-with-prg-ai-and-recombee-2.png) As a tech company, investing in employees’ self-development directly impacts the quality of our solution — as expressed by our CEO Pavel Kordik, Recombee strives to be the leading recommendation platform with top-notch technical solutions. To keep ourselves ahead of the competition, we support employees that wish to improve their skills in their field by giving them funding, data, computing resources, and infrastructure for online experiments. This sponsorship is also possible thanks to our partnership with the NVIDIA [Inception programme](https://www.nvidia.com/en-us/deep-learning-ai/startups/) that supports Recombee's research activities with students. Petr, as a fresh graduate enrolled in the **industrial Ph.D. program** at [CTU FIT](https://fit.cvut.cz/en/studies/programs-and-specializations/doctoral/dsp-informatics) supported by Recombee. Apart from [industrial PhDs](mailto:research@recombee.com), we also offer joint [postdoc positions](mailto:research@recombee.com) with the [CTU in Prague](https://fit.cvut.cz/en/science-and-research/what-we-do/research-topics/8874-machine-learning-and-data-processing), where you can work on fundamental research, disseminate results in the research community, and solve real challenges at Recombee at the same time, getting feedback on how your algorithms work in the real world. [Rodrigo Alves](https://scholar.google.com/citations?hl=cs&user=86t4YwYAAAAJ&view_op=list_works&sortby=pubdate), who is just finishing his doctoral study in Germany, is joining our team from February. ## Available Doctoral Research Topics In the field of recommender systems, we can offer many topics that can be elaborated into a dissertation thesis. From developing deep neural networks to reducing cold start problems, designing new recommendation algorithms or transformers for sequential prediction of next basket, there is always something exciting to dive deep into, be it reinforcement learning to optimize longer-term metrics or using AutoML to optimize the architecture and hyperparameters of the recommendation system. Come to Prague as it is not only a [great place to live](https://www.timeout.com/news/prague-has-just-been-voted-the-most-beautiful-city-in-the-world-091521), it is also home to a [vibrant AI community](https://prg.ai/en/category/prg-ai-news/) where fundamental AI research turns into global AI products. ## Want to Join Do you have a tech background? Are you interested in DevOps, backend development, or machine learning? More than that, do you want to work in an environment that supports you and helps you grow? Get in touch with us at [career@recombee.com](mailto:career@recombee.com) and join the family! ## Next Articles [![](https://www.recombee.com/img/blog/new-features-for-a-better-personalization-experience.png)](https://www.recombee.com/blog/new-features-for-a-better-personalization-experience) ### [New Features for a Better Personalization Experience](https://www.recombee.com/blog/new-features-for-a-better-personalization-experience) Like most of the world, the majority of 2021 was spent on home office or in isolation - which left us with all the time to be invested in work (and Netflix :) ) and improving UX for our clients. We are now happy to share new features we can offer to reach new levels of personalization. ![](https://www.recombee.com/img/blog/authors/ondrejfiedler.png) Ondrej Fiedler Mar 07, 2022 New Features [![](https://www.recombee.com/img/blog/linear-methods-and-autoencoders-in-recommender-systems.png)](https://www.recombee.com/blog/linear-methods-and-autoencoders-in-recommender-systems) ### [Linear Methods and Autoencoders in Recommender Systems](https://www.recombee.com/blog/linear-methods-and-autoencoders-in-recommender-systems) Linear regression is probably the simplest and surprisingly efficient machine learning method. It should be the method of your first choice, according to the famous KISS principle. Also, it often works better than sophisticated methods, because it is... ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Sep 7, 2021 Recommendation Engine [![](https://www.recombee.com/img/blog/recombee-xperience-kentico.png)](https://xperience.io/discover/blog/recombee-kentico-xperience-1-on-1-personalization) ### [Recombee and Kentico Xperience: Guide to One-On-One Personalization](https://xperience.io/discover/blog/recombee-kentico-xperience-1-on-1-personalization) Recombee expanded its integration options - and now is available at the Kentico Xperience platform! Analyzing different types of personalization, we look into why Kentiko chose our AI-powered recommendation engine over manual segmentation. ![](https://www.recombee.com/img/blog/authors/gabrielatakacova.png) Gabriela Takacova May 6, 2021 Personalization Recommendation Engine --- # AI News and Outlook for 2024 | Blog > Source: https://www.recombee.com/blog/ai-news-and-outlook-for-2024 > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # AI News and Outlook for 2024 ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Jan 16, 2024 ![](https://www.recombee.com/img/blog/ai-news-and-outlook-for-2024/main.png) We look at the most interesting research directions and assess the state of knowledge in key areas of AI. We'll also estimate future developments in 2024 so you know what to prepare for. I have been involved in artificial intelligence research, specifically artificial neural networks, for 25 years. All this time, I have been trying to keep up with the latest trends and approaches, which, especially lately, has become quite challenging given the breakneck pace of development in this field. In this article I try to summarize in an accessible way the most interesting recent discoveries that will have a significant impact on how the world around us will look like. After all, in the field of AI, the time from discovery, to application development, to global adoption is incredibly short. That's why we can expect to see the latest innovations in AI research come to fruition in a matter of months. So if you want to be at least a little bit prepared for what's coming in 2024, read the following. ## Transformers in Data Centres The advent of so-called large language models (LLMs), which we talk to through, for example, [ChatGPT](https://chat.openai.com/) or [Bing](https://www.bing.com/search?q=Bing+AI&showconv=1&FORM=hpcodx), is due to advances in deep neural network architectures. So-called deep neural autoencoders learn to add words to sentences or reconstruct parts of images on huge datasets. This forces them to create their own internal representation of the world and understand how language, art, music or spoken word works. This is why we may think they are intelligent, even if they are only very good at predicting the following text or generating images from text. ![](https://www.recombee.com/img/blog/ai-news-and-outlook-for-2024/01.png) The so-called [Transformer](https://jalammar.github.io/illustrated-transformer/) is a special neural autoencoder that stores information during learning in so-called attention matrices, which directly express, for example, connections between words. It turns out that when we increase the capacity of the transformer (number and size of maps) the results keep improving. Modern LLMs are therefore trained and run on [large computing infrastructures](https://developer.nvidia.com/blog/new-video-what-runs-chatgpt/). ## Economics of AI Products There is a cost to training and running large models. Despite all the optimizations, running good LLM models is still quite costly. That's why only large corporations like Microsoft, Google or [Amazon](https://www.aboutamazon.com/news/company-news/amazon-aws-anthropic-ai) offer them for free and make a lot of money. As with other AI products, they are looking for a way to make money from running them. Interestingly, there aren't many [production AI systems](https://www.evidentlyai.com/ml-system-design) yet, other than recommendation and search algorithms, that pay real money to run. But it seems that users are [willing to pay](https://a16z.com/the-economic-case-for-generative-ai-and-foundation-models/) for access to good language models and services built on top of them, because it will pay them back. I think that this year will see the emergence of a lot of companies offering customers subscription-based time-saving services built on top of LLM. Both cheap ones for the mass market (ala [ChatGPT Plus](https://openai.com/blog/chatgpt-plus) or [Gitlab Copilot](https://github.blog/2022-06-21-github-copilot-is-generally-available-to-all-developers/)) and more expensive ones that add a lot of value for a small group of users. And this is where there is huge potential, while at the same time not being in such [danger](https://www.businessinsider.com/openai-chatgpt-pdfs-ai-startups-wrappers-2023-10?op=1) from global players. It's easy to build a service on top of [ready-made models](https://platform.openai.com/docs/models/chatgpt), you just need to think about what your competitive advantage will be. Or you can run your smaller language model locally, [even on mobile phones](https://huggingface.co/papers/2401.02385), so applications can work even without internet access. Of course, the capabilities of these models, especially the smaller ones, are quite limited. Let's now take a look at how they could be improved this year. ## What Is in Store for Us This Year? Of course, it is difficult to estimate given the pace of development, but I will at least try to make a rough prediction. Although LLMs are [starting to be used as reasoning engines](https://auto-rt.github.io/), it is their limited ability to reason analytically that is their biggest weakness. Here we can expect LLMs to be enriched with technologies we know from decades of [reinforcement learning research](https://hunch.net/~jl/projects/RL/RLTheoryTutorial.pdf), and AI assistants to maintain their [model of the world](https://worldmodels.github.io/). They can predict future events, use advanced attentional mechanisms to track important changes and perform actions based on them. This will bring them closer to how humans operate. Here are some interesting directions ([1](https://arxiv.org/pdf/2305.14992.pdf), [2](https://proceedings.mlr.press/v119/badia20a/badia20a.pdf), [3](https://arxiv.org/pdf/1802.07740.pdf), [4](https://truyentran.github.io/kdd2021-tute.html)). While LLMs can store a lot of information in their weight matrices, because they function similarly to biological neural networks, they are often inaccurate in their equipment. However, many tasks require accurate memorization and fitting of information, and therefore, e.g., LLMs with [additional memory](https://arxiv.org/pdf/2305.16338.pdf) are being investigated, which can then [function similarly to a Turing machine](https://arxiv.org/pdf/2301.04589.pdf) and thus can solve more complex tasks in the future. In addition to transformers, promising new neural autoencoder architectures such as [SSMs](https://arxiv.org/pdf/2303.09489.pdf) or [diffusion models](https://arxiv.org/pdf/2311.18257.pdf) are also emerging. ![](https://www.recombee.com/img/blog/ai-news-and-outlook-for-2024/02.png) [Source](https://arxiv.org/pdf/2302.01560.pdf) The ability to plan and generalise the acquired knowledge to solve [new unfamiliar tasks](https://arxiv.org/pdf/2302.01560.pdf) and the ability to solve [multiple tasks](https://arxiv.org/pdf/2205.06175.pdf) in [different domains](https://arxiv.org/pdf/2303.03378.pdf) also becomes a very important aspect. The way in which people can convey their preferences to the models is also improving, thus improving the generated [texts](https://arxiv.org/pdf/2305.18290.pdf) and [images](https://arxiv.org/pdf/2311.12908.pdf). Or perhaps to perform more [complex robotic tasks](https://mobile-aloha.github.io/). Being able to measure how well models perform is crucial for AI model research and development. This is very difficult in the case of general AI, so the development of so-called [benchmarks](https://arxiv.org/pdf/2304.06364.pdf) will be very helpful. [These](https://arxiv.org/pdf/2301.00493.pdf) are of course also essential for the development of [self-driving cars](https://arxiv.org/pdf/2309.17080.pdf), for example. Sometimes it is even useful for the AI to [generate the tasks](https://arxiv.org/pdf/2301.07608.pdf) on which it improves itself. Well, and since all this requires a lot of computing resources, there will be a lot of work on how to make both [hardware](https://www.nature.com/articles/s41467-023-42981-1) and especially software faster and cheaper (see [LoRa](https://arxiv.org/pdf/2106.09685.pdf), [QLoRa](https://arxiv.org/pdf/2305.14314.pdf), [Flash Attention](https://github.com/Dao-AILab/flash-attention)). Thanks to all these advances, we will soon have an army of different digital [personas and assistants](https://arxiv.org/pdf/2309.11696.pdf) at our disposal. They will no longer be passive, but proactive. I reckon that those of us who pay for more and better assistants in the future will be more efficient, they will make more money because their assistants will be advocating for their interests in the digital world. Of course, the assistants will be connected to the Internet and will [improve themselves](https://proceedings.neurips.cc/paper/2020/file/6b493230205f780e1bc26945df7481e5-Paper.pdf) by getting real-time information. ![](https://www.recombee.com/img/blog/ai-news-and-outlook-for-2024/03.png) [Source](https://arxiv.org/pdf/2312.10997.pdf) They will also [iteratively improve](https://arxiv.org/pdf/2305.06983.pdf) their outputs and actions, similar to the [inputs](https://arxiv.org/pdf/2309.08532.pdf) and [training datasets](https://arxiv.org/pdf/2401.01335.pdf). Those who have quick access to [new validated information](https://openai.com/blog/axel-springer-partnership) and quality data repositories will be ahead of the game. Of course, it will also be about finding the most effective interface between humans and AI. Not everyone will be willing to have Elon make a [neural implant](https://www.wired.com/story/everything-we-know-about-neuralinks-brain-implant-trial/). And humans are not yet very willing to communicate with machines even in natural language. Maybe it's because the machines haven't understood us for so long. Either way, our recent experience suggests it's probably gonna take some time. At Recombee we have trained a simple LLM model for a US client for a service where you get a recommendation just by asking in natural language. Although the service has found loyal users, the mainstream probably won't end up being impressed in this form. And so we need to experiment with other ways in which new technologies can make our lives more efficient, and hopefully also better. I wish everyone well for the year of the "full-blown technological AI revolution". I believe it will bring more good than bad to humanity, and that we all can contribute to it too. Recommendation Engine Personalization ## Next Articles [![](https://www.recombee.com/img/blog/recombeelabs-2023-research-publications.png)](https://www.recombee.com/blog/recombeelabs-2023-research-publications) ### [Recombeelab's 2023 Research Publications](https://www.recombee.com/blog/recombeelabs-2023-research-publications) Recombeelab, a joint research laboratory of Recombee and the Faculty of Information Technology at the Czech Technical University in Prague, experienced a highly productive year in 2023, publishing a series of insightful and impactful papers in the field of recommendation systems. ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Jan 19, 2024 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/ai-assistants-know-your-preferences-even-better-than-you-do.png)](https://www.recombee.com/blog/ai-assistants-know-your-preferences-even-better-than-you-do) ### [AI Assistants Know Your Preferences, Even Better Than You Do](https://www.recombee.com/blog/ai-assistants-know-your-preferences-even-better-than-you-do) Recommender systems and ethical controversies ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Nov 23, 2023 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/the-ai-revolution-in-the-media-industry.png)](https://www.recombee.com/blog/the-ai-revolution-in-the-media-industry) ### [The AI (R)Evolution in the Media Industry](https://www.recombee.com/blog/the-ai-revolution-in-the-media-industry) In today's digital age, personalization has become the cornerstone of the media industry. Whether it's tailoring content recommendations, refining marketing strategies, or enhancing user experiences... ![](https://www.recombee.com/img/blog/authors/annapetr.png) Anna Dolezelova & Petr Popov Oct 23, 2023 Recommendation Engine Personalization --- # Company News | Blog > Source: https://www.recombee.com/blog/company-news > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # Blog ## Company News [All](https://www.recombee.com/blog) [Recommendation Engine](https://www.recombee.com/blog/recommendation-engine) [Personalization](https://www.recombee.com/blog/personalization) [Integrations](https://www.recombee.com/blog/integrations) [New Features](https://www.recombee.com/blog/new-features) [Partnerships](https://www.recombee.com/blog/partnerships) [Company News](https://www.recombee.com/blog/company-news) [![](https://www.recombee.com/img/blog/recombee-partners-with-the-telegraph-to-deliver-ai-driven-personalisation-to-millions-of-readers.png)](https://www.recombee.com/blog/recombee-partners-with-the-telegraph-to-deliver-ai-driven-personalisation-to-millions-of-readers) ### [Recombee Partners with The Telegraph to Deliver AI-Driven Personalisation to Millions of Readers](https://www.recombee.com/blog/recombee-partners-with-the-telegraph-to-deliver-ai-driven-personalisation-to-millions-of-readers) Prague, 11th September 2025 – Recombee, a leading AI-powered recommendation platform, has announced a strategic partnership with The Telegraph... ![](https://www.recombee.com/img/blog/authors/recombee.png) Recombee Oct 02, 2025 Company News --- # Elevate Your Personalization Strategy with Recombee's Innovative Features > Source: https://www.recombee.com/blog/elevate-your-personalization-strategy-with-recombees-innovative-features > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # Elevate Your Personalization Strategy with Recombee's Innovative Features ![](https://www.recombee.com/img/blog/authors/janvaluch.png) Jan Valuch Mar 13, 2024 The digital landscape and customer preferences and behavior are changing faster than ever now. To help our clients stay on top of the game, our team has focused on developing innovative features that give a competitive advantage to companies aiming to utilize cutting-edge AI and fine-tuned algorithms to meet their goals. The result is an arsenal of features designed to increase engagement, satisfaction, sales, and overall growth. ![](https://www.recombee.com/img/blog/elevate-your-personalization-strategy-with-recombees-innovative-features/main.gif) ## Features Designed to Improve Your Recommendations ### Constraints ![](https://www.recombee.com/img/blog/elevate-your-personalization-strategy-with-recombees-innovative-features/1.png) [Constraints](https://docs.recombee.com/scenarios#constraints) is a feature that enables you to configure how items with shared conditions (e.g., product category or movie genre) may repeat within the individual recommendation boxes. You might want to recommend at most 50% of movies from a single genre, just one product per brand, or at most 2 items from each category. With Constraints, you can create an even more diverse and engaging experience. **Examples from different industries:** * **VOD and similar** companies can ensure that movies from multiple genres will be recommended in one row. * **Media companies** can mix free articles with premium articles on one page. * **E-commerce businesses** can ensure that a variety of products from different vendors will be recommended. ### Item Segmentations ![](https://www.recombee.com/img/blog/elevate-your-personalization-strategy-with-recombees-innovative-features/2.png) Your quest for great personalization doesn’t stop with merely recommending products or individual pieces of content. Thanks to [Item Segmentations](https://www.recombee.com/blog/recombee-item-segmentations), you can get the top categories, artists, or brands according to each user’s preferences. You can even personalize the order of the rows on the homepage using this unique feature! Users will no longer have to sift through the site or scroll endlessly to find their favorite category. Instead, they can see the content they are most likely to engage with right at the top of the page. ### Search No-Code Widget ![](https://www.recombee.com/img/blog/elevate-your-personalization-strategy-with-recombees-innovative-features/3.png) The [Quick Search No-Code Widget](https://docs.recombee.com/no-code-widgets#html-widget-full-text-search) offers you an innovative way to seamlessly integrate Recombee's personalized search functionality into your website, providing an alternative to using the API. Thanks to this no-code editor solution, you can customize it to match your overall website design in just a few clicks. You can choose which items or item segments (e.g., categories, genres, or artists) are recommended, the order in which the sections are displayed, and even the visual appearance of an item in the widget. ### New Ready-To-Use Logics ![](https://www.recombee.com/img/blog/elevate-your-personalization-strategy-with-recombees-innovative-features/4.png) Logics help Recombee users tailor the behavior of recommendations to meet their specific goals. After picking a Scenario, the place where you want to show the recommendations (e.g., home page, emailing, detail page, full-text search), you can choose from our [selection of Logics](https://docs.recombee.com/recommendation_logics). **Examples of most popular Logics:** * **Video and Media:** just for you, similar content, read/watch next, continue watching, and popular * **E-commerce:** similar products, bestsellers, or complementary products (cross-sell) ### AI ReQL Code Assistant ![](https://www.recombee.com/img/blog/elevate-your-personalization-strategy-with-recombees-innovative-features/5.png) One of the ways to get the desired behavior or recommendations is to customize by adding business rules ([filters and boosters](https://docs.recombee.com/reql_filtering_and_boosting)). The use of rules varies by industry, but here are some of the most common examples. **E-commerce companies** might use boosters when they need to promote certain brands or high-margin products. On another note, a filter might help them to recommend only products that are currently available. **Video platforms** such as [FTV Prima](https://www.recombee.com/case-studies/ftv-prima) can use boosters to favor certain content (e.g., increase the visibility of paid content or content selected by editors). Filters might be used to show different content in each row on the homepage (e.g., only award-winning movies or only episodes published in the last 7 days). When applying these business rules, you can choose from our library of predefined rules or use the [Recombee Query Language (ReQL)](https://docs.recombee.com/reql). But don't worry about coding complex queries - our [AI ReQL Code Assistant](https://docs.recombee.com/reql#reql-code-assistant) will do the work for you. Simply describe how you want your business rules to behave in your natural language and watch our assistant generate the ReQL code. You can also ask for a simple explanation of your current business rules or see the history of the conversation. ## Ready to Reach the Next Level of Personalization? Let’s Connect! **Our specialist’s at your disposal** ![](https://www.recombee.com/img/team/filip-hanus.png) For more details about our solution contact [filip.hanus@recombee.com](mailto:filip.hanus@recombee.com) ![](https://www.recombee.com/img/team/petr-popov.png) For partnerships and cooperation contact [petr.popov@recombee.com](mailto:petr.popov@recombee.com) New Features Recommendation Engine ## Next Articles [![](https://www.recombee.com/img/blog/recombee-partners-with-axinom-to-enhance-video-streaming-experiences.png)](https://www.recombee.com/blog/recombee-partners-with-axinom-to-enhance-video-streaming-experiences) ### [Recombee Partners With Axinom to Enhance Video Streaming Experiences](https://www.recombee.com/blog/recombee-partners-with-axinom-to-enhance-video-streaming-experiences) This collaboration is set to introduce a new era of personalized and engaging digital user experiences. ![](https://www.recombee.com/img/blog/authors/janvaluch.png) Jan Valuch Mar 20, 2024 Partnerships Personalization [![](https://www.recombee.com/img/blog/modern-recommender-systems-part-2-data.png)](https://www.recombee.com/blog/modern-recommender-systems-part-2-data) ### [Modern Recommender Systems - Part 2: Data](https://www.recombee.com/blog/modern-recommender-systems-part-2-data) Data used by modern recommenders and how we can measure progress towards goals. ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Mar 07, 2024 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/recombee-real-time-ai-recommendations-as-the-new-destination-in-segment.png)](https://www.recombee.com/blog/recombee-real-time-ai-recommendations-as-the-new-destination-in-segment) ### [Recombee Real-Time AI Recommendations as the New Destination in Segment](https://www.recombee.com/blog/recombee-real-time-ai-recommendations-as-the-new-destination-in-segment) Segment has enabled its users to enjoy Recombee personalization services without the need to leave their platform and with minimum coding involved. With a few simple clicks, domains using Segment can upgrade their services to maximize the digital experience for their customers. ![](https://www.recombee.com/img/blog/authors/adelasloupenska.png) Adela Sloupenska Mar 05, 2024 Personalization Integrations Partnerships --- # How Interdisciplinary Collaboration Can Accelerate AI Innovation | Blog > Source: https://www.recombee.com/blog/how-interdisciplinary-collaboration-can-accelerate-ai-innovation > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # How Interdisciplinary Collaboration Can Accelerate AI Innovation ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Sep 1, 2020 In a world where innovation is the new standard, Recombee uses the power of interdisciplinary collaboration to stay at the cutting edge of innovation. Partnering up with the leading player in the food industry (Bofrost) and academia (FIT CTU), allowed Recombee to hold a student competition to create AI which can shape the future of the food industry. ![](https://www.recombee.com/img/blog/how-interdisciplinary-collaboration-can-accelerate-ai-innovation.png) ## Intro At Recombee, we welcome a bit of healthy competition. After all, the only way in which to achieve truly scalable innovation in the area of AI is to iterate, build, hack, and come up with new ideas to make sure the market and associated technologies keep shifting forward. So we decided to create a competition. We gave students an opportunity to work with real-world data, threw real-world problems at them, and gave them real-world deadlines (they only had a few hours to come up with a solution...sounds like a work deadline, anyone?). We feel this is the only way in which talented young people can prepare for the world of applied AI and transfer theory into practice. What was the theme of the competition, we hear you eagerly ask? If there was a time where quick deployment of technologies to solve problems was crucial than ever before, it was the time of COVID. Hackathons, competitions, ideathons and everything in between started taking place at a rocket pace. Suddenly, older structures and status quo technologies are no longer sufficient and the hunger for new, better innovations has been larger than ever before. One of those things has been the alpha and omega of modern living - food delivery. Especially at a time when restaurants have been forced to physically shut and social distancing was enforced across society. At the time of COVID19 crisis and quarantine, the popularity of services that bring food to the house has grown exponentially (yep, you’re not the only person who enjoys chain-ordering pizza or repetitively orders the same chocolate cake every week…). **For most companies, the so-called online groceries, the volume of sales is growing several times and it’s likely this trend will only continue now that the crisis is almost over. Customers have gotten used to the ease in which they can consume these new services** (both metaphorically and literally…). Now; let’s not forget that AI feeds on data. Given the growing popularity of these online ordering tools, especially for weekly shopping, the amount of data flowing in about customer behavior has been substantial. Using a range of AI algorithms and methods, the startups of today can predict, based on past behavioral data, what a customer may need and order the shopping box to the house without the customer having to order its contents. Think of it as a predictive food delivery butler. If people then don’t collect this box (ie the algorithm gets it wrong and sends you a collection of dried seaweed instead of pepperoni) the company will collect the box back and of course this feedback will be incorporated in the algorithm which will get more accurate over time as this data flows in. Of course, this method can only be **used with durable foods** \- you don’t want to risk something like this with raw chicken being left on your doorstep. Nonetheless, it still places great demands on the accuracy of the AI algorithm since you don’t want food to get wasted or customers to be dissatisfied, let alone the cost of recovering wrongly predicted products. **Enter stage right: shopping cart prediction and Recombee’s methodology.** The deployment of AI methods has huge commercial potential in this area and both buyers and sellers in the food industry, as well as customers would welcome a more predictive way in which they can plan what they buy, **plan what they sell, streamline processes, save time, logistic resources, and also not have to waste time online** adding the same items you buy every week to your shopping basket. The problem the market is currently facing in this regard however, is that the prediction methods are not yet reliable enough to allow, for example, the aforementioned predictive shopping. This is mainly due to the absence of quality data over which research into suitable algorithms could be performed. **Recombee has a joint research laboratory with the Faculty of Information Technologies (FIT CTU) and aims to crack these types of problems through an interdisciplinary, academia-meets-business approach.** Recombee’s customer Bofrost allowed us to use its unique dataset, which is unparalleled in the size and quality (several hundreds millions of purchases) when predicting a shopping cart. We wanted to give students an opportunity to work on such an interesting problem and this competition was a unique way of doing that. ## So What Was the Competition Itself? The entrants were given a clear task to complete; and for the sake of sentiment, let’s recap it here so we can share with you the real-world use case behind our thinking. **Bofrost is a company selling frozen food to customers in 13 countries.** The network of drivers delivers goods to customers' homes in deliveries with built-in refrigerators. Customers are mostly loyal, like Bofrost, and buy Bofrost’ goods regularly (several entries per shopper - the ‘AI dream’). Also, when a Bofrost van visits their street, they usually buy several different products at once. Given that deliveries have limited capacity and for many other reasons of optimization, it would be very beneficial for Bofrost to predict what goods individual customers will buy on the delivery route so that they can load the right amount of goods. Your goal is to produce an algorithm for forecasting the purchase (basket) based on historical data. The goal is to build a model that accepts a set of historical transactions (baskets) made by a particular user along with the current time (,today’) to predict a list of items that the user will purchase today (basket). The algorithm should be as accurate as possible in the cart estimate. The optimal model predicts all the items that are actually purchased but doesn’t predict anything else, so the delivery can only be loaded with the necessary number of correct items. In order to be able to create a model with the greatest predictive power (perhaps there is no need to explain to anyone that the models will be evaluated on unpublished test data), Recombee and FIT CTU staff provided to students several large and well-labelled data sets. ## Data, Data, Data For the competition, we have prepared several tables and datasets for the entrants to use to crack this enigma. This was picked up well by the attendees, who immediately understood that data quantity isn’t everything but instead putting the correct emphasis on correct labelling and categorisation which will improve the accuracy of the algorithm. The transaction data categories used for the competition were: 1. Anonymous User ID 2. Purchase date 3. Purchased goods (ID) Crucially, Bofrost, being such a large distribution network, not only had large quantities of data about their deliveries, but also a **wide diversity of entries, ranging from returned items, undelivered items, successful deliveries, times of delivery** et cetera. This was able to provide us with a rich database based on which more sophisticated algorithms could be built, leading to very interesting entries from all participants. This emphasis on data and making it publically available underpins a core aspect of AI as a whole; without correct data and correct labelling, AI isn’t magic and will not fashion solutions out of thin air. We were very happy to report a mature, interdisciplinary methodical approach the students took to this task and are excited about the new cohort of AI professionals which are incubating at our universities today. ## Results and Next Steps? As much as we love rewarding all enthusiasm and talent, there can only be one winner. Well, two in our case because there were two phenomenal candidates who we recognised as excelling in both the first and the second round of this competition out of 31 students who actively participated by submitting their solutions. The first one was **Matyas Skalicky** \- for technical details of his solution winning first round check out [**his report**](http://stoked.cz/files/tmp/skalimat.pdf), as well as **Filip Dolnik** who decided not to go down the deep learning methodology route, but instead went for more traditional data modeling methods and smart heuristics. This shows that ‘AI’ while a fantastic development in human history, has many faces, and deep learning methods are just one of the many approaches to achieve great results. Often, taking a hybrid approach of for example combining a deep learning basis with smart heuristics ‘on top’ of it can add the extra accuracy edge needed for a top-shelf product. We’re immensely proud that this competition has so far brought this to the surface and highlighted more great talent in our community. Keep your eyes peeled for next steps and more cool competitions coming your way from the Recombee gang… Also, if you like to learn more about machine learning techniques involved in predicting the next shopping basket from past purchases check out the second part of our [**intro blogpost.**](https://medium.com/recombee-blog/machine-learning-for-recommender-systems-part-2-deep-recommendation-sequence-prediction-automl-f134bc79d66b) We are also about to publish more about deep learning recommender systems soon as we’ve heard from many of you, you’d like to know more about this. You can look at the [**master thesis of Radek Bartyzal**](https://dspace.cvut.cz/bitstream/handle/10467/82308/F8-DP-2019-Bartyzal-Radek-thesis.pdf) as a bit of a taster teaser of the type of cool content we’re prepping for you. Last but not least; we’re planning a new competition of this type to keep the momentum going and to motivate more students to show us what they’ve got and how they apply what they’ve learned in their courses so far in real life. Also, the Recsys research community is hungry for new challenges that will have tremendous business impact. ## Next Articles [![](https://www.recombee.com/img/blog/23d22w82u3d52as.png)](https://medium.com/recombee-blog/deep-learning-for-recommender-systems-next-basket-prediction-and-sequential-product-recommendation-796228b34dee) ### [Deep Learning for Recommender Systems: Next Basket Prediction and Sequential Product Recommendation](https://medium.com/recombee-blog/deep-learning-for-recommender-systems-next-basket-prediction-and-sequential-product-recommendation-796228b34dee) Accurate “next basket prediction” will be enabling next generation e-commerce — predictive shopping and logistics. In this blogpost, we will discuss the deep learning technology behind next basket... ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Nov 10, 2020 Personalization Recommendation Engine [![](https://cdn-images-1.medium.com/max/1000/1*dSVF4ZxwQPIaIdnmPCsmPQ.png)](https://medium.com/recombee-blog/introduction-to-personalized-search-2b70eb5fa5ae) ### [Introduction to Personalized Search](https://medium.com/recombee-blog/introduction-to-personalized-search-2b70eb5fa5ae) Personalized search should take into account user preferences and interactions of similar users. We combined search engine and recommender. ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Mar 24, 2020 [![](https://cdn-images-1.medium.com/max/1000/1*YTIhIUMgAequmZ7cmghKTA@2x.png)](https://medium.com/recombee-blog/recombee-in-2019-new-features-and-improvements-bcbfc85acb80) ### [Recombee in 2019: New Features and Improvements](https://medium.com/recombee-blog/recombee-in-2019-new-features-and-improvements-bcbfc85acb80) This year was really huge for us. We worked on new features so hard that we almost forgot to write a blog post about them :) ![](https://www.recombee.com/img/blog/authors/tomasrehorek.png) Tomas Rehorek Dec 15, 2019 New Features --- # Recombee Insights: The Next Level of Analytics in Recombee UI > Source: https://www.recombee.com/blog/insights-the-next-level-of-analytics-in-recombee-ui > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # Insights: The Next Level of Analytics in Recombee UI ![](https://www.recombee.com/img/blog/authors/ondrejfiedler.png) Ondrej Fiedler May 09, 2024 ![](https://www.recombee.com/img/blog/insights-the-next-level-of-analytics-in-recombee-ui/main.png) ## What is our new feature Insights about? * Insights elevates Recombee capabilities by incorporating **advanced real-time analytics** into its Admin UI. * It reveals how users interact with recommendations and your platform, offering **transparent and close monitoring of the recommender's behavior,** user engagement details, and organic interactions. * It enables **easy creation of various visualizations and report types,** from stacked to line charts and tables, with agile data slicing, selective filtering, and compound metrics calculation. * Our curated **library of visualizations,** crafted for the most common analytical use cases, offers immediate application with the flexibility for adjustments and further exploration. * In synergy with Recombee's robust customization capabilities, Insights provides targeted **control over your unique KPIs,** enabling informed, strategic decisions. At Recombee, we understand that a great recommender engine today extends beyond merely optimizing metrics such as click-through rate (CTR) and conversion rate (CR). While these metrics are crucial and our engine aims to uplift them, it's essential that the recommended content **aligns with our customers' product visions**and editorial teams' goals, while also adhering to their **unique KPIs.** And these KPIs can often be more complex than just CTR and CR. To help our customers follow their product strategies, we've introduced several features that allow for the **customization of recommendations** within different parts of their web or app. Among these features are the ability to customize the recommendation model's behavior, select which content to recommend or boost (either through predefined rules or our flexible query language, ReQL), and our latest addition, [Item Segmentations](https://www.recombee.com/blog/recombee-item-segmentations). ![](https://www.recombee.com/img/blog/insights-the-next-level-of-analytics-in-recombee-ui/01.png) Scenario configuration in the Recombee Admin UI However, until now, assessing the **impact of these business rules,** especially concerning custom KPIs, has been challenging. Many of our customers have had questions like: * Is the current boosting of my VoD platform’s original content sufficient to reach most users? * How has boosting by margin changed the distribution of recommended brands? * And which content pieces or products are actually getting the most clicks in individual recommendation boxes? Our newest feature called Insights has an answer to all these questions. ## Introducing Insights Insights is a new analytics section within the Recombee Admin UI. ## Library of Predefined Insights The Insights feature comes with a **library of predefined visualizations,** providing a straightforward way to understand the tool and its capabilities. ![](https://www.recombee.com/img/blog/insights-the-next-level-of-analytics-in-recombee-ui/02.png) The library of predefined Insights **These predefined Insights can answer many questions, such as:** * What are the most recommended items in the individual places where you show recommendations? * What categories do the recommended items belong to? * or How is the generated revenue from recommendations distributed among your affiliate partners? All the Insights are updated in near real-time, taking new recommendations and interactions into account in less than a second. ![](https://www.recombee.com/img/blog/insights-the-next-level-of-analytics-in-recombee-ui/03.png) The most recommended items per scenario Insight from the library We believe these predefined Insights will be of great help to many of our customers. However, at the same time, we know that your questions about the distribution of recommended and interacted items can be complex and unique, and no library of predefined visualizations can fully address them all. That's why we have spent a tremendous amount of designing and engineering time to come up with a tool that is both **user-friendly and versatile** enough to handle your unique queries. ## Custom Insights To create a **custom insight,** begin by either editing an existing Insight from the library or creating a new one from scratch. In that case, you need to first choose its type - from various chart types to tables for more complex reports - and then configure: ### Data that shall be presented You can choose from a wide variety of data sources and associated metrics, e.g.: * number of recommended items, * number of interactions based on recommendations, * number of interactions in general, * number of interacting users, * generated profit, * watched portions of videos, * …and much more. ### Filters on these data You may want to go into more detail and look at the data only for: * a specific recommendation box (Scenario), * a specific category (or any other property that you have specified in your Items catalog), * or even for a particular Item or User ### Splits You can break down the data by various fields revealing the top: * scenarios, * categories, * affiliate partners, * brands, * and much more… The combination of Data Sources, Filters, and Splits is pretty powerful. For instance, the following figure shows the profit generated by out clicks (Data Source) from the recommendations box on product detail (Filter on a Scenario) per affiliate partner (Split): ![](https://www.recombee.com/img/blog/insights-the-next-level-of-analytics-in-recombee-ui/04.png) You can also look into the ratio metrics, such as what is the share of each category on all the views: ![](https://www.recombee.com/img/blog/insights-the-next-level-of-analytics-in-recombee-ui/05.png) And if you want to examine multiple metrics simultaneously, you can create a table report such as this one: ![](https://www.recombee.com/img/blog/insights-the-next-level-of-analytics-in-recombee-ui/06.png) Have these examples piqued your interest? Check out our tutorial video to see these configurations in action. ## Conclusion Insights, the **analytics section of our Admin UI,** offers various **predefined and fully customizable reports** to track recommended items and how users interact with these recommendations. In Recombee, we are particularly excited about the configurability of Insights and believe you will love it too. The feature is now in public Beta, available to all our customers in the US and Europe. ## Want to see Insights in action? [Get your free trial here!](https://admin.recombee.com/sign-up) New Features Recommendation Engine ## Next Articles [![](https://www.recombee.com/img/blog/video-recommendations-made-easy-integrating-axinom-mosaic-with-recombee.png)](https://www.axinom.com/webinar/video-backends-with-recommendations) ### [Video Recommendations Made Easy: Integrating Axinom Mosaic with Recombee](https://www.axinom.com/webinar/video-backends-with-recommendations) In this webinar we look into building a data-driven video backend geared towards personalized video recommendations, integration with Axinom Mosaic, and how to transform user experiences on streaming platforms. ![](https://www.recombee.com/img/blog/authors/grigorygrin.png) Grigory Grin (Axinom) Sep 09, 2024 Personalization Partnerships [![](https://www.recombee.com/img/blog/recombee-partners-with-axinom-to-enhance-video-streaming-experiences.png)](https://www.recombee.com/blog/recombee-partners-with-axinom-to-enhance-video-streaming-experiences) ### [Recombee Partners With Axinom to Enhance Video Streaming Experiences](https://www.recombee.com/blog/recombee-partners-with-axinom-to-enhance-video-streaming-experiences) This collaboration is set to introduce a new era of personalized and engaging digital user experiences. ![](https://www.recombee.com/img/blog/authors/janvaluch.png) Jan Valuch Mar 20, 2024 Partnerships Personalization [![](https://www.recombee.com/img/blog/elevate-your-personalization-strategy-with-recombees-innovative-features.png)](https://www.recombee.com/blog/elevate-your-personalization-strategy-with-recombees-innovative-features) ### [Elevate Your Personalization Strategy with Recombee's Innovative Features](https://www.recombee.com/blog/elevate-your-personalization-strategy-with-recombees-innovative-features) The digital landscape and customer preferences and behavior are changing faster than ever now. To help our clients stay on top of the game, our team has focused on developing innovative features... ![](https://www.recombee.com/img/blog/authors/janvaluch.png) Jan Valuch Mar 13, 2024 New Features Recommendation Engine --- # Integrations | Blog > Source: https://www.recombee.com/blog/integrations > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # Blog ## Integrations [All](https://www.recombee.com/blog) [Recommendation Engine](https://www.recombee.com/blog/recommendation-engine) [Personalization](https://www.recombee.com/blog/personalization) [Integrations](https://www.recombee.com/blog/integrations) [New Features](https://www.recombee.com/blog/new-features) [Partnerships](https://www.recombee.com/blog/partnerships) [Company News](https://www.recombee.com/blog/company-news) [![](https://www.recombee.com/img/blog/recombee-real-time-ai-recommendations-as-the-new-destination-in-segment.png)](https://www.recombee.com/blog/recombee-real-time-ai-recommendations-as-the-new-destination-in-segment) ### [Recombee Real-Time AI Recommendations as the New Destination in Segment](https://www.recombee.com/blog/recombee-real-time-ai-recommendations-as-the-new-destination-in-segment) Segment has enabled its users to enjoy Recombee personalization services without the need to leave their platform and with minimum coding involved. With a few simple clicks, domains using Segment can upgrade their services to maximize the digital experience for their customers. ![](https://www.recombee.com/img/blog/authors/adelasloupenska.png) Adela Sloupenska Mar 05, 2024 Personalization Integrations Partnerships --- # Is this comment useful? Enhancing Personalized Recommendations by Considering User Rating Uncertainty > Source: https://www.recombee.com/blog/is-this-comment-useful-enhancing-personalized-recommendations-by-considering-user-rating-uncertainty > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # Is This Comment Useful? Enhancing Personalized Recommendations by Considering User Rating Uncertainty ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Mar 01, 2024 Picture this: you're on the hunt for the perfect new smartphone, browsing through your favourite online electronics store. The online store’s recommendation engine pops up with what it thinks could be your possible next gadget love. You narrow down your options and, to ensure you're making the right choice, you dive into the reviews. Then you find the two reviews below, **both awarded five out of five stars** for the same smartphone: ![](https://www.recombee.com/img/blog/is-this-comment-useful-enhancing-personalized-recommendations-by-considering-user-rating-uncertainty/01.png) You probably have faced this situation before. On one hand, the review of Anna is detailed, citing specific features, performance metrics, and personal experiences with the phone. It reads like the user has really put the phone through its paces and knows what they're talking about. The Bob’s review, on the other hand, is vague and non-committal, offering little beyond a general sense of satisfaction. Instinctively, you're more inclined to trust the first reviewer's opinion. Their thoroughness and confidence suggest they're well-informed, making their review a valuable piece of your decision-making puzzle. But here's the million-dollar question: will the recommendation system take this disparity in review depth and confidence into account when suggesting your next potential purchase? Unfortunately, the answer is: not frequently! Traditional methods almost never take such properties into account when evaluating users' feedback. Let’s take the example of recommender systems (RSs) based on matrix factorization (MF) methods, a very popular and efficient and accurate way of providing personalised user experiences by aggregating collective feedback. In the explicit feedback setting, where users explicitly rate part of the items set (as described in the figure above), the RSs rely on the ratings matrix, where rows (resp. columns) represent users (resp. items) and only a small proportion of the entries (user-to-item ratings) are observed. To provide customised recommendations, an MF recommender performs the following two steps: (1) factorise the ratings matrix in order to predict the large set of unrated ratings using an MF method, and (2) rank the items for each individual user based on the predictions. When new feedback arrives, the system restarts the described cycle from the beginning, resulting in a newly trained model with the potential to improve the recommendation accuracy. Although MF recommenders are often efficient and accurate, they are highly dependent on the quality of the users' feedback. On the one hand, clear and consistent feedback supports the algorithm's ability to filter collaborative behavior. Noisy or anomalous feedback, on the other hand, can induce the learning method to misinterpret users' preferences, potentially leading to suboptimal recommendations. Since users frequently perceive rating as a tedious process, certain inconsistencies in the data are unavoidable. In MF methods this type of vulnerability to noise is especially complex: the predictions for all of the users are related to each other, and strong noise in one user (or item) can influence and propagate through the predictions of the entire model. However, classic MF methods **do not adapt to the noise** in user-item ratings or factor the uncertainty into the model. Even the simple fact that some users are (or less) confident in their judgement is typically neglected. By giving the same importance to the ratings of all users (both normal and anomalous), the learning method can be notably affected if a substantial number of users provide evidently erratic feedback. Crucially, unlike other machine learning methods, inaccurate ratings and noise in recommender systems can be highly structured in their dependence on the input. **In simpler terms, these systems only look at the numerical rating a user leaves, ignoring any written comments. Even though these comments could offer valuable insights for making better recommendations in the future, they're not taken into account.** The factorization of our smartphone example is illustrated in the figure below: ![](https://www.recombee.com/img/blog/is-this-comment-useful-enhancing-personalized-recommendations-by-considering-user-rating-uncertainty/02.png) How to deal with this problem? Previous attempts at dealing with so called **natural noise management in RSs** can be classified into two types: those that rely solely on rating values and those that require extra side information for noise detection. The primary objective of the first group is to detect noisy ratings by analysing their distribution regarding a particular user and/or item. They then use a collaborative filtering strategy to predict unobserved ratings by only considering denoised and noise-free items. The second group, on the other hand, utilises item attributes to detect users' inconsistent behaviour when rating similar items and, then, correct the rating predictions. Despite their important contributions, neither trend concentrated on weighting the loss function to directly inject knowledge about the uncertainty attached to the quality of user ratings into the model. Indeed, existing methods simply ignore user, items or ratings which they deem anomalous, instead of providing a model-based directly on a continuous notion of reliability or noisiness. More importantly, the few results in this branch of research which pertain to MF approaches are entirely empirical and no theoretical guarantees were given. Recently, we present a matrix factorization method that uses side information in the form of an estimate of the user's uncertainty to deliver more robust predictions. Our paper was published in the renowned IEEE Transactions on Neural Networks and Learning Systems journal \[1\]. The two main technical improvements are the introduction of **a weighting on the loss function and a modified regularisation** strategy aimed at controlling any negative effects of this injected form of non-uniformity. How we do that? We postulate that more comprehensive reviews made by a particular user indicate that he/she/they is meticulous in his/her/their evaluations and thus may consistently generate less noise during the rating process. We reinforce the feedback from such users who appear to demonstrate a higher standard of thoroughness and trustworthiness in the assessment of the items they interact with. To achieve this aim, we construct weights for each user indicating the degree of trustworthiness of their reviews, and use them to re-weight the empirical estimate of the loss function. The weight will be higher if we estimate that their predictions are an accurate reflection of a consistent taste and lower if their rating behaviour appears erratic. However, this does not mean that users deemed the most "trustworthy" must produce ratings which agree with each other, but rather that they are not subject to outside noise not directly related to the user-item combination. In other words, we give more importance to users which consistently gives consistent comments together with their feedback. In a mathematical sense, our loss function differentiate from traditional matrix factorization as the following: ![](https://www.recombee.com/img/blog/is-this-comment-useful-enhancing-personalized-recommendations-by-considering-user-rating-uncertainty/03.png) For a deeper dive into the technicalities, specially regarding our modified regularisation procedure, we invite you to read our paper. Yet, a pivotal question remains: how do we determine the weight of each user's input in the loss function? The methodology can significantly differ based on the specific use case. Options include organizing comments with sophisticated large language models or assigning greater importance to more elaborate and comprehensive feedback. Alternatively, we could directly seek user opinions to assess the relevance of their contributions. Note that in our model, the weights are related to each user; this means that we do not need to observe reviews for all items the user interacts with, but typically just a few are enough. Investing in research is crucial here, as it enables us not only to utilize existing technologies but also to innovate and tailor solutions to meet individual client needs. ## References \[1\] Alves, R., Ledent, A., & Kloft, M. (2023). Uncertainty-adjusted recommendation via matrix factorization with weighted losses. IEEE Transactions on Neural Networks and Learning Systems. Recommendation Engine Personalization ## Next Articles [![](https://www.recombee.com/img/blog/recombee-real-time-ai-recommendations-as-the-new-destination-in-segment.png)](https://www.recombee.com/blog/recombee-real-time-ai-recommendations-as-the-new-destination-in-segment) ### [Recombee Real-Time AI Recommendations as the New Destination in Segment](https://www.recombee.com/blog/recombee-real-time-ai-recommendations-as-the-new-destination-in-segment) Segment has enabled its users to enjoy Recombee personalization services without the need to leave their platform and with minimum coding involved. With a few simple clicks, domains using Segment can upgrade their services to maximize the digital experience for their customers. ![](https://www.recombee.com/img/blog/authors/adelasloupenska.png) Adela Sloupenska Mar 05, 2024 Personalization Integrations Partnerships [![](https://www.recombee.com/img/blog/recombeelabs-2023-research-publications.png)](https://www.recombee.com/blog/recombeelabs-2023-research-publications) ### [Recombeelab's 2023 Research Publications](https://www.recombee.com/blog/recombeelabs-2023-research-publications) Recombeelab, a joint research laboratory of Recombee and the Faculty of Information Technology at the Czech Technical University in Prague, experienced a highly productive year in 2023, publishing a series of insightful and impactful papers in the field of recommendation systems. ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Jan 19, 2024 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/ai-news-and-outlook-for-2024.png)](https://www.recombee.com/blog/ai-news-and-outlook-for-2024) ### [AI News and Outlook for 2024](https://www.recombee.com/blog/ai-news-and-outlook-for-2024) We look at the most interesting research directions and assess the state of knowledge in key areas of AI. We'll also estimate future developments in 2024 so you know what to prepare for. ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Jan 16, 2024 Recommendation Engine Personalization --- # New Features | Blog > Source: https://www.recombee.com/blog/new-features > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # Blog ## New Features [All](https://www.recombee.com/blog) [Recommendation Engine](https://www.recombee.com/blog/recommendation-engine) [Personalization](https://www.recombee.com/blog/personalization) [Integrations](https://www.recombee.com/blog/integrations) [New Features](https://www.recombee.com/blog/new-features) [Partnerships](https://www.recombee.com/blog/partnerships) [Company News](https://www.recombee.com/blog/company-news) [![](https://www.recombee.com/img/blog/2025-sneak-peek.png)](https://www.recombee.com/blog/2025-sneak-peek) ### [2025 Sneak Peek](https://www.recombee.com/blog/2025-sneak-peek) This year is already off to an exciting start, and we’re rolling out new tools to improve efficiency and optimize recommendations. Here’s what’s available and what’s coming next. ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Mar 19, 2025 Recommendation Engine New Features [![](https://www.recombee.com/img/blog/2024-wrap-up.png)](https://www.recombee.com/blog/2024-wrap-up) ### [2024 Wrap-Up](https://www.recombee.com/blog/2024-wrap-up) As we wrap up 2024, let’s take a moment to celebrate the milestones we’ve achieved together. Your collaboration and feedback have been at the heart of our progress. Here’s a quick recap of last year’s highlights. ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Jan 27, 2025 New Features Personalization [![](https://www.recombee.com/img/blog/introducing-beeformer-a-framework-for-training-foundational-models-for-recommender-systems.png)](https://www.recombee.com/blog/introducing-beeformer-a-framework-for-training-foundational-models-for-recommender-systems) ### [Introducing beeFormer: A Framework for Training Foundational Models for Recommender Systems](https://www.recombee.com/blog/introducing-beeformer-a-framework-for-training-foundational-models-for-recommender-systems) In the fast-evolving world of recommender systems, understanding both how users interact with content and the actual content itself is crucial. Many existing recommender systems struggle to balance these two aspects... ![](https://www.recombee.com/img/blog/authors/vojtechvancura.png) Vojtech Vancura Oct 15, 2024 New Features Recommendation Engine [![](https://www.recombee.com/img/blog/insights-the-next-level-of-analytics-in-recombee-ui.png)](https://www.recombee.com/blog/insights-the-next-level-of-analytics-in-recombee-ui) ### [Insights: The Next Level of Analytics in Recombee UI](https://www.recombee.com/blog/insights-the-next-level-of-analytics-in-recombee-ui) Insights, the analytics section of our Admin UI, offers various predefined and fully customizable reports to track recommended items and how users interact with these recommendations. ![](https://www.recombee.com/img/blog/authors/ondrejfiedler.png) Ondrej Fiedler May 09, 2024 New Features Recommendation Engine [![](https://www.recombee.com/img/blog/elevate-your-personalization-strategy-with-recombees-innovative-features.png)](https://www.recombee.com/blog/elevate-your-personalization-strategy-with-recombees-innovative-features) ### [Elevate Your Personalization Strategy with Recombee's Innovative Features](https://www.recombee.com/blog/elevate-your-personalization-strategy-with-recombees-innovative-features) The digital landscape and customer preferences and behavior are changing faster than ever now. To help our clients stay on top of the game, our team has focused on developing innovative features... ![](https://www.recombee.com/img/blog/authors/janvaluch.png) Jan Valuch Mar 13, 2024 New Features Recommendation Engine [![](https://www.recombee.com/img/blog/innovative-personalization-features-for-2023.png)](https://www.recombee.com/blog/innovative-personalization-features-for-2023) ### [Innovative Personalization Features for 2023](https://www.recombee.com/blog/innovative-personalization-features-for-2023) The digital world is changing; users' expectations for personalization are increasing, and our Recombee features are continuously improving. One of our focuses is to support our clients in providing the best possible user experiences... ![](https://www.recombee.com/img/blog/authors/gabrielatakacova.png) Gabriela Takacova Feb 17, 2023 New Features Recommendation Engine [![](https://www.recombee.com/img/blog/recombee-item-segmentations.png)](https://www.recombee.com/blog/recombee-item-segmentations) ### [Recombee Item Segmentations](https://www.recombee.com/blog/recombee-item-segmentations) Item Segmentations are Recombee's original and elegant solution to various advanced tasks related to hierarchical and relational data. The feature provides a flexible way to group items (products or pieces of content) into segments... ![](https://www.recombee.com/img/blog/authors/tomasrehorek.png) Tomas Rehorek Jan 11, 2023 New Features Recommendation Engine [![](https://www.recombee.com/img/blog/new-features-for-a-better-personalization-experience.png)](https://www.recombee.com/blog/new-features-for-a-better-personalization-experience) ### [New Features for a Better Personalization Experience](https://www.recombee.com/blog/new-features-for-a-better-personalization-experience) Like most of the world, the majority of 2021 was spent on home office or in isolation - which left us with all the time to be invested in work (and Netflix :) ) and improving UX for our clients. We are now happy to share new features we can offer to reach new levels of personalization. ![](https://www.recombee.com/img/blog/authors/ondrejfiedler.png) Ondrej Fiedler Mar 07, 2022 New Features [![](https://www.recombee.com/img/blog/recombee-in-2020/main.png)](https://www.recombee.com/blog/recombee-in-2020) ### [Recombee in 2020: New Features and Improvements](https://www.recombee.com/blog/recombee-in-2020) We know that this year has been quite challenging for many people, including ourselves. However, today we want to focus entirely on the positive (no pun included) side of the year and the stuff we are the proudest of. ![](https://www.recombee.com/img/blog/authors/ondrejfiedler.png) Ondrej Fiedler Dec 30, 2020 New Features [![](https://cdn-images-1.medium.com/max/1000/1*YTIhIUMgAequmZ7cmghKTA@2x.png)](https://medium.com/recombee-blog/recombee-in-2019-new-features-and-improvements-bcbfc85acb80) ### [Recombee in 2019: New Features and Improvements](https://medium.com/recombee-blog/recombee-in-2019-new-features-and-improvements-bcbfc85acb80) This year was really huge for us. We worked on new features so hard that we almost forgot to write a blog post about them :) ![](https://www.recombee.com/img/blog/authors/tomasrehorek.png) Tomas Rehorek Dec 15, 2019 New Features --- # Partnerships | Blog > Source: https://www.recombee.com/blog/partnerships > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # Blog ## Partnerships [All](https://www.recombee.com/blog) [Recommendation Engine](https://www.recombee.com/blog/recommendation-engine) [Personalization](https://www.recombee.com/blog/personalization) [Integrations](https://www.recombee.com/blog/integrations) [New Features](https://www.recombee.com/blog/new-features) [Partnerships](https://www.recombee.com/blog/partnerships) [Company News](https://www.recombee.com/blog/company-news) [![](https://www.recombee.com/img/blog/celestial-tiger-entertainment-launches-new-chinese-movie-app-cmgo-with-diagnal.png)](https://www.diagnal.com/cmgo/) ### [Celestial Tiger Entertainment launches new Chinese Movie app, CMGO, with DIAGNAL](https://www.diagnal.com/cmgo/) With Recombee’s AI-powered recommendation engine working with DIAGNAL Enhance, CMGO serves up personalised experiences for each viewer, driving engagement for the service. ![](https://www.recombee.com/img/blog/authors/diagnal.png) Diagnal Nov 14, 2024 Personalization Partnerships [![](https://www.recombee.com/img/blog/video-recommendations-made-easy-integrating-axinom-mosaic-with-recombee.png)](https://www.axinom.com/webinar/video-backends-with-recommendations) ### [Video Recommendations Made Easy: Integrating Axinom Mosaic with Recombee](https://www.axinom.com/webinar/video-backends-with-recommendations) In this webinar we look into building a data-driven video backend geared towards personalized video recommendations, integration with Axinom Mosaic, and how to transform user experiences on streaming platforms. ![](https://www.recombee.com/img/blog/authors/grigorygrin.png) Grigory Grin (Axinom) Sep 09, 2024 Personalization Partnerships [![](https://www.recombee.com/img/blog/recombee-partners-with-axinom-to-enhance-video-streaming-experiences.png)](https://www.recombee.com/blog/recombee-partners-with-axinom-to-enhance-video-streaming-experiences) ### [Recombee Partners With Axinom to Enhance Video Streaming Experiences](https://www.recombee.com/blog/recombee-partners-with-axinom-to-enhance-video-streaming-experiences) This collaboration is set to introduce a new era of personalized and engaging digital user experiences. ![](https://www.recombee.com/img/blog/authors/janvaluch.png) Jan Valuch Mar 20, 2024 Partnerships Personalization [![](https://www.recombee.com/img/blog/recombee-real-time-ai-recommendations-as-the-new-destination-in-segment.png)](https://www.recombee.com/blog/recombee-real-time-ai-recommendations-as-the-new-destination-in-segment) ### [Recombee Real-Time AI Recommendations as the New Destination in Segment](https://www.recombee.com/blog/recombee-real-time-ai-recommendations-as-the-new-destination-in-segment) Segment has enabled its users to enjoy Recombee personalization services without the need to leave their platform and with minimum coding involved. With a few simple clicks, domains using Segment can upgrade their services to maximize the digital experience for their customers. ![](https://www.recombee.com/img/blog/authors/adelasloupenska.png) Adela Sloupenska Mar 05, 2024 Personalization Integrations Partnerships [![](https://www.recombee.com/img/blog/recombee-in-e-mail-marketing-a-partner-success-story-with-ryzeo.png)](https://www.recombee.com/blog/recombee-in-e-mail-marketing-a-partner-success-story-with-ryzeo) ### [Recombee in E-mail Marketing: A Partner Success Story with Ryzeo](https://www.recombee.com/blog/recombee-in-e-mail-marketing-a-partner-success-story-with-ryzeo) Do you feel there is a potential to increase your success with customers through an efficient recommender engine? You're highly likely right. Adding a recommender service to your emailing campaigns gives each client tailored product recommendations in all of their emails. ![](https://www.recombee.com/img/blog/authors/russellmiller.png) Russell Miller (Ryzeo) Oct 5, 2022 Partnerships [![](https://www.recombee.com/img/blog/how-we-are-using-ai-to-power-content-recommendations.png)](https://revium.com.au/blog/how-we-are-using-ai-to-power-content-recommendations) ### [How We Are Using AI to Power Content Recommendations](https://revium.com.au/blog/how-we-are-using-ai-to-power-content-recommendations) In this article we walk you through how we are using the AI recommendation engine Recombee embedded in our headless CMS StoryBlok to drive content recommendations throughout our own website. ![](https://www.recombee.com/img/blog/authors/revium.png) Revium Sep 30, 2022 Partnerships [![](https://www.recombee.com/img/blog/ai-powered-content-recommendations-with-a-headless-cms.png)](https://www.luminary.com/blog/ai-content-recommendations-headless-cms-recombee) ### [AI-Powered Content Recommendations With a Headless CMS](https://www.luminary.com/blog/ai-content-recommendations-headless-cms-recombee) Thanks to its API-first nature, it is quite straightforward to integrate your headless CMS with the most powerful AI-powered content recommendations available on the market. Luminary just did that with their own website, Kontent.ai and Recombee. ![](https://www.recombee.com/img/blog/authors/andythompson.png) Andy Thompson (Luminary) Aug 31, 2022 Partnerships --- # Personalization | Blog > Source: https://www.recombee.com/blog/personalization > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # Blog ## Personalization [All](https://www.recombee.com/blog) [Recommendation Engine](https://www.recombee.com/blog/recommendation-engine) [Personalization](https://www.recombee.com/blog/personalization) [Integrations](https://www.recombee.com/blog/integrations) [New Features](https://www.recombee.com/blog/new-features) [Partnerships](https://www.recombee.com/blog/partnerships) [Company News](https://www.recombee.com/blog/company-news) [![](https://www.recombee.com/img/blog/new-feature-ab-testing.png)](https://www.recombee.com/blog/new-feature-ab-testing) ### [New Feature: A/B Testing](https://www.recombee.com/blog/new-feature-ab-testing) Personalization is never finished. Every change to your recommendation strategy raises new questions. Should you boost fresh content? Change used Logic? Introduce new filters? ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Aug 14, 2026 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/mid-year-roundup-2026-so-far.png)](https://www.recombee.com/blog/mid-year-roundup-2026-so-far) ### [Mid-Year Roundup: 2026 So Far](https://www.recombee.com/blog/mid-year-roundup-2026-so-far) The digital world continues to change at an incredible pace. As technology evolves and user expectations keep rising, creating experiences that feel relevant has never been more important. ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Jul 20, 2026 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/a-2025-research-retrospective.png)](https://www.recombee.com/blog/a-2025-research-retrospective) ### [A 2025 Research Retrospective](https://www.recombee.com/blog/a-2025-research-retrospective) In 2025, I became Head of Research at Recombee. I knew it would be a challenge: and I wasn’t wrong. Recombee is a leader in recommendation-as-a-service, with thousands of clients... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Jan 29, 2026 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/looking-back-at-2025.png)](https://www.recombee.com/blog/looking-back-at-2025) ### [Looking Back at 2025](https://www.recombee.com/blog/looking-back-at-2025) 2025 marked 10 years of Recombee. A decade of building personalization from first principles, shaped by research, real-world deployments, and close collaboration with partners across industries. ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Jan 20, 2026 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/product-highlights-from-2025.png)](https://www.recombee.com/blog/product-highlights-from-2025) ### [Product Highlights from 2025](https://www.recombee.com/blog/product-highlights-from-2025) In 2025, we focused on making advanced personalization easier to implement, scale, and maintain across products and platforms. Here’s a look at the key product updates we released last year. ![](https://www.recombee.com/img/blog/authors/ondrejfiedler.png) Ondrej Fiedler Jan 08, 2026 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/ai-news-and-outlook-for-2026.png)](https://www.recombee.com/blog/ai-news-and-outlook-for-2026) ### [AI News and Outlook for 2026](https://www.recombee.com/blog/ai-news-and-outlook-for-2026) Here’s what caught my attention in AI research lately, and where things might be heading in 2026\. After 25+ years in this field, the pace has gotten hard to keep up with. I’m trying to make sense of... ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Dec 17, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/modern-recommender-systems-part-3-objectives.png)](https://www.recombee.com/blog/modern-recommender-systems-part-3-objectives) ### [Modern Recommender Systems - Part 3: Objectives](https://www.recombee.com/blog/modern-recommender-systems-part-3-objectives) Learning objectives of recommender systems and personalized search. ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Sep 03, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/the-building-blocks-of-privacy-friendly-personalization.png)](https://www.recombee.com/blog/the-building-blocks-of-privacy-friendly-personalization) ### [The Building Blocks of Privacy-Friendly Personalization](https://www.recombee.com/blog/the-building-blocks-of-privacy-friendly-personalization) Personalization can be achieved without compromising user privacy. While many personalization systems have historically relied on practices now considered intrusive, like... ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Aug 07, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/no-code-search-widget-personalized-powerful-effortless.png)](https://www.recombee.com/blog/no-code-search-widget-personalized-powerful-effortless) ### [No-Code Search Widget: Personalized, Powerful, Effortless](https://www.recombee.com/blog/no-code-search-widget-personalized-powerful-effortless) At Recombee, we don't just excel at recommendations – we provide powerful full-text search capabilities too. Our Quick, No-Code Search Widget exemplifies this, offering a seamless... ![](https://www.recombee.com/img/blog/authors/ondrejfiedler.png) Ondrej Fiedler Jul 25, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/how-regionalization-based-recommendations-can-improve-your-operations.png)](https://www.recombee.com/blog/how-regionalization-based-recommendations-can-improve-your-operations) ### [How Regionalization-Based Recommendations Can Improve Your Operations](https://www.recombee.com/blog/how-regionalization-based-recommendations-can-improve-your-operations) From ancient trade routes to modern urban planning, geography has consistently shaped human decisions and opportunities. Today, in the world of online business and personalized recommendations... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Jul 18, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/shield-the-universal-framework-making-ai-search-safer-for-everyone.png)](https://www.recombee.com/blog/shield-the-universal-framework-making-ai-search-safer-for-everyone) ### [SHIELD: The Universal Framework Making AI Search Safer for Everyone](https://www.recombee.com/blog/shield-the-universal-framework-making-ai-search-safer-for-everyone) Imagine searching for "glass tubing" and getting recommendations for drug manufacturing equipment. As AI-powered search becomes ubiquitous — from online marketplaces to social networks... ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Jun 24, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/making-recommendations-fairer-a-new-way-to-guarantee-exposure-for-all.png)](https://www.recombee.com/blog/making-recommendations-fairer-a-new-way-to-guarantee-exposure-for-all) ### [Making Recommendations Fairer: A New Way to Guarantee Exposure for All](https://www.recombee.com/blog/making-recommendations-fairer-a-new-way-to-guarantee-exposure-for-all) As recommender systems become more widespread across digital platforms, concerns around fairness are coming to the forefront. Standard relevance-based ranking techniques, while effective... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Apr 29, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/build-vs-buy-deciding-the-best-approach-for-your-recommender-system.png)](https://www.recombee.com/blog/build-vs-buy-deciding-the-best-approach-for-your-recommender-system) ### [Build vs. Buy: Deciding the Best Approach for Your Recommender System](https://www.recombee.com/blog/build-vs-buy-deciding-the-best-approach-for-your-recommender-system) When it comes to deciding between buying a recommender system and building one from scratch, the choice isn’t always straightforward. Both options come with their own set of pros and cons... ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova & Ondrej Fiedler Mar 14, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/are-you-here-to-stay-unraveling-the-dynamics-of-stable-and-curious-audiences-in-web-systems.png)](https://www.recombee.com/blog/are-you-here-to-stay-unraveling-the-dynamics-of-stable-and-curious-audiences-in-web-systems) ### [Are You Here to Stay? Unraveling the Dynamics of Stable and Curious Audiences in Web Systems](https://www.recombee.com/blog/are-you-here-to-stay-unraveling-the-dynamics-of-stable-and-curious-audiences-in-web-systems) Why do influencers frequently request their subscribers to enable all notifications for their channels? This practice stems from their awareness that not all subscribers are regular... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Mar 03, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/recombee-research-2024.png)](https://www.recombee.com/blog/recombee-research-2024) ### [Recombee Research 2024](https://www.recombee.com/blog/recombee-research-2024) Recombee has always been deeply connected to academia, with four of our six co-founders holding Ph.D. degrees. Over the years, our investment in research has grown alongside our company and the increasing demand for advanced features in the market. ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Feb 23, 2025 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/key-trends-in-news-and-media-for-2025-and-the-role-of-ai-powered-solutions.png)](https://www.recombee.com/blog/key-trends-in-news-and-media-for-2025-and-the-role-of-ai-powered-solutions) ### [Key Trends in News & Media for 2025 and the Role of AI-Powered Solutions](https://www.recombee.com/blog/key-trends-in-news-and-media-for-2025-and-the-role-of-ai-powered-solutions) The way news is produced, delivered, and consumed keeps shifting, and as we move through 2025, a few big changes have become impossible to ignore. ![](https://www.recombee.com/img/blog/authors/annapetr.png) Gabriela Takacova & Violeta Milarova Feb 20, 2025 Personalization [![](https://www.recombee.com/img/blog/2024-wrap-up.png)](https://www.recombee.com/blog/2024-wrap-up) ### [2024 Wrap-Up](https://www.recombee.com/blog/2024-wrap-up) As we wrap up 2024, let’s take a moment to celebrate the milestones we’ve achieved together. Your collaboration and feedback have been at the heart of our progress. Here’s a quick recap of last year’s highlights. ![](https://www.recombee.com/img/blog/authors/annapetr.png) Violeta Milarova Jan 27, 2025 New Features Personalization [![](https://www.recombee.com/img/blog/celestial-tiger-entertainment-launches-new-chinese-movie-app-cmgo-with-diagnal.png)](https://www.diagnal.com/cmgo/) ### [Celestial Tiger Entertainment launches new Chinese Movie app, CMGO, with DIAGNAL](https://www.diagnal.com/cmgo/) With Recombee’s AI-powered recommendation engine working with DIAGNAL Enhance, CMGO serves up personalised experiences for each viewer, driving engagement for the service. ![](https://www.recombee.com/img/blog/authors/diagnal.png) Diagnal Nov 14, 2024 Personalization Partnerships [![](https://www.recombee.com/img/blog/video-recommendations-made-easy-integrating-axinom-mosaic-with-recombee.png)](https://www.axinom.com/webinar/video-backends-with-recommendations) ### [Video Recommendations Made Easy: Integrating Axinom Mosaic with Recombee](https://www.axinom.com/webinar/video-backends-with-recommendations) In this webinar we look into building a data-driven video backend geared towards personalized video recommendations, integration with Axinom Mosaic, and how to transform user experiences on streaming platforms. ![](https://www.recombee.com/img/blog/authors/grigorygrin.png) Grigory Grin (Axinom) Sep 09, 2024 Personalization Partnerships [![](https://www.recombee.com/img/blog/recombee-partners-with-axinom-to-enhance-video-streaming-experiences.png)](https://www.recombee.com/blog/recombee-partners-with-axinom-to-enhance-video-streaming-experiences) ### [Recombee Partners With Axinom to Enhance Video Streaming Experiences](https://www.recombee.com/blog/recombee-partners-with-axinom-to-enhance-video-streaming-experiences) This collaboration is set to introduce a new era of personalized and engaging digital user experiences. ![](https://www.recombee.com/img/blog/authors/janvaluch.png) Jan Valuch Mar 20, 2024 Partnerships Personalization [![](https://www.recombee.com/img/blog/modern-recommender-systems-part-2-data.png)](https://www.recombee.com/blog/modern-recommender-systems-part-2-data) ### [Modern Recommender Systems - Part 2: Data](https://www.recombee.com/blog/modern-recommender-systems-part-2-data) Data used by modern recommenders and how we can measure progress towards goals. ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Mar 07, 2024 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/recombee-real-time-ai-recommendations-as-the-new-destination-in-segment.png)](https://www.recombee.com/blog/recombee-real-time-ai-recommendations-as-the-new-destination-in-segment) ### [Recombee Real-Time AI Recommendations as the New Destination in Segment](https://www.recombee.com/blog/recombee-real-time-ai-recommendations-as-the-new-destination-in-segment) Segment has enabled its users to enjoy Recombee personalization services without the need to leave their platform and with minimum coding involved. With a few simple clicks, domains using Segment can upgrade their services to maximize the digital experience for their customers. ![](https://www.recombee.com/img/blog/authors/adelasloupenska.png) Adela Sloupenska Mar 05, 2024 Personalization Integrations Partnerships [![](https://www.recombee.com/img/blog/is-this-comment-useful-enhancing-personalized-recommendations-by-considering-user-rating-uncertainty.png)](https://www.recombee.com/blog/is-this-comment-useful-enhancing-personalized-recommendations-by-considering-user-rating-uncertainty) ### [Is This Comment Useful? Enhancing Personalized Recommendations by Considering User Rating Uncertainty](https://www.recombee.com/blog/is-this-comment-useful-enhancing-personalized-recommendations-by-considering-user-rating-uncertainty) Picture this: you're on the hunt for the perfect new smartphone, browsing through your favourite online electronics store. The online store’s recommendation engine pops up with what it thinks could be your possible next gadget love... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Mar 01, 2024 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/recombeelabs-2023-research-publications.png)](https://www.recombee.com/blog/recombeelabs-2023-research-publications) ### [Recombeelab's 2023 Research Publications](https://www.recombee.com/blog/recombeelabs-2023-research-publications) Recombeelab, a joint research laboratory of Recombee and the Faculty of Information Technology at the Czech Technical University in Prague, experienced a highly productive year in 2023, publishing a series of insightful and impactful papers in the field of recommendation systems. ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Jan 19, 2024 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/ai-news-and-outlook-for-2024.png)](https://www.recombee.com/blog/ai-news-and-outlook-for-2024) ### [AI News and Outlook for 2024](https://www.recombee.com/blog/ai-news-and-outlook-for-2024) We look at the most interesting research directions and assess the state of knowledge in key areas of AI. We'll also estimate future developments in 2024 so you know what to prepare for. ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Jan 16, 2024 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/ai-assistants-know-your-preferences-even-better-than-you-do.png)](https://www.recombee.com/blog/ai-assistants-know-your-preferences-even-better-than-you-do) ### [AI Assistants Know Your Preferences, Even Better Than You Do](https://www.recombee.com/blog/ai-assistants-know-your-preferences-even-better-than-you-do) Recommender systems and ethical controversies ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Nov 23, 2023 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/the-ai-revolution-in-the-media-industry.png)](https://www.recombee.com/blog/the-ai-revolution-in-the-media-industry) ### [The AI (R)Evolution in the Media Industry](https://www.recombee.com/blog/the-ai-revolution-in-the-media-industry) In today's digital age, personalization has become the cornerstone of the media industry. Whether it's tailoring content recommendations, refining marketing strategies, or enhancing user experiences... ![](https://www.recombee.com/img/blog/authors/annapetr.png) Anna Dolezelova & Petr Popov Oct 23, 2023 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/modern-recommender-systems-part-1-introduction.png)](https://www.recombee.com/blog/modern-recommender-systems-part-1-introduction) ### [Modern Recommender Systems - Part 1: Introduction](https://www.recombee.com/blog/modern-recommender-systems-part-1-introduction) How machine learning methods simplify item discovery and search. ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Apr 17, 2023 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/explaining-recommender-systems-to-product-owners.png)](https://pavelkordik.substack.com/p/explaining-recommender-systems-to) ### [Explaining Recommender Systems to Product Owners](https://pavelkordik.substack.com/p/explaining-recommender-systems-to) In my presentation at the Data Technology Seminar organized by the European Broadcasting Union, I have focused on demonstrating that recommender systems can actually help public media organizations to better fulfill their role in society and reduce content distribution biases. ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Mar 27, 2023 Personalization Recommendation Engine [![](https://www.recombee.com/img/blog/inductive-matrix-completion-how-to-improve-recommendations-for-cold-start-users-and-items-by-incorporating-their-attributes.png)](https://www.recombee.com/blog/inductive-matrix-completion-how-to-improve-recommendations-for-cold-start-users-and-items-by-incorporating-their-attributes) ### [Inductive Matrix Completion: How to Improve Recommendations for Cold Start Users and Items by Incorporating Their Attributes](https://www.recombee.com/blog/inductive-matrix-completion-how-to-improve-recommendations-for-cold-start-users-and-items-by-incorporating-their-attributes) Matrix completion (MC), the problem of recovering the missing entries of a partially observed matrix, has found use in a wide range of domains. Still, its potentially most successful application is as a collaborative filtering technique for recommender systems (RSs)... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Mar 20, 2023 Personalization Recommendation Engine [![](https://www.recombee.com/img/blog/breaking-the-news-the-role-of-ai-in-modern-journalism.png)](https://www.recombee.com/blog/breaking-the-news-the-role-of-ai-in-modern-journalism) ### [Breaking the News: The Role of AI in Modern Journalism](https://www.recombee.com/blog/breaking-the-news-the-role-of-ai-in-modern-journalism) Artificial Intelligence (AI) has rapidly transformed the media industry in recent years. From automated news production to trend analysis and personalized content recommendations, AI has brought significant changes to the way media is created, distributed, and consumed. ![](https://www.recombee.com/img/blog/authors/tanalancova.png) Tana Lancova Mar 14, 2023 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/keeping-up-with-digital-media-convergence.png)](https://www.recombee.com/blog/keeping-up-with-digital-media-convergence) ### [Keeping Up With Digital Media Convergence](https://www.recombee.com/blog/keeping-up-with-digital-media-convergence) At Recombee, we felt the transition within the media industry accelerated by the pandemic. OTT and CTV consumption ballooned at a significant rate. ![](https://www.recombee.com/img/blog/authors/annadolezelova.png) Anna Dolezelova Oct 8, 2022 Personalization [![](https://www.recombee.com/img/blog/real-time-personalization-of-content-with-ai-powered-recommendations.png)](https://www.recombee.com/blog/real-time-personalization-of-content-with-ai-powered-recommendations) ### [Real-Time Personalization of Content With AI-Powered Recommendations](https://www.recombee.com/blog/real-time-personalization-of-content-with-ai-powered-recommendations) Do you manage a publishing company, online gaming platform, or a streaming site with a content-heavy catalog and are thinking about how to improve the user experience? ![](https://www.recombee.com/img/blog/authors/karenharazimova.png) Karen Harazimova Sep 16, 2022 Personalization [![](https://www.recombee.com/img/blog/recombee-xperience-kentico.png)](https://xperience.io/discover/blog/recombee-kentico-xperience-1-on-1-personalization) ### [Recombee and Kentico Xperience: Guide to One-On-One Personalization](https://xperience.io/discover/blog/recombee-kentico-xperience-1-on-1-personalization) Recombee expanded its integration options - and now is available at the Kentico Xperience platform! Analyzing different types of personalization, we look into why Kentiko chose our AI-powered recommendation engine over manual segmentation. ![](https://www.recombee.com/img/blog/authors/gabrielatakacova.png) Gabriela Takacova May 6, 2021 Personalization Recommendation Engine [![](https://www.recombee.com/img/blog/23d22w82u3d52as.png)](https://medium.com/recombee-blog/deep-learning-for-recommender-systems-next-basket-prediction-and-sequential-product-recommendation-796228b34dee) ### [Deep Learning for Recommender Systems: Next Basket Prediction and Sequential Product Recommendation](https://medium.com/recombee-blog/deep-learning-for-recommender-systems-next-basket-prediction-and-sequential-product-recommendation-796228b34dee) Accurate “next basket prediction” will be enabling next generation e-commerce — predictive shopping and logistics. In this blogpost, we will discuss the deep learning technology behind next basket... ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Nov 10, 2020 Personalization Recommendation Engine [![](https://cdn-images-1.medium.com/max/1000/1*d8rJ5EWZTOgfEp-lKYBJeA.jpeg)](https://medium.com/recombee-blog/evaluating-recommender-systems-choosing-the-best-one-for-your-business-c688ab781a35) ### [Evaluating Recommender Systems: Choosing the Best One for Your Business](https://medium.com/recombee-blog/evaluating-recommender-systems-choosing-the-best-one-for-your-business-c688ab781a35) Together with the endless expansion of E-commerce and online media in the last years, there are more and more Software-as-a-Service (SaaS)… ![](https://www.recombee.com/img/blog/authors/tomasrehorek.png) Tomas Rehorek Dec 18, 2016 Personalization Recommendation Engine [![](https://cdn-images-1.medium.com/max/1000/1*iKboSkp-zHi8ZNH8ChtP1w.png)](https://medium.com/recombee-blog/the-value-of-personalized-recommendations-for-your-business-6b2e81ce0a4d) ### [The Value of Personalized Recommendations for Your Business](https://medium.com/recombee-blog/the-value-of-personalized-recommendations-for-your-business-6b2e81ce0a4d) The e-commerce boom makes online environment more competitive. Internet retailers seek competitive advantages and a personalized experience… ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Aug 23, 2016 Personalization Recommendation Engine [![](https://miro.medium.com/max/1400/1*NARxz9O7ZX3Un_HMacPcWw.png)](https://medium.com/recombee-blog/artificial-intelligence-in-the-cloud-310da5e0325e) ### [Artificial Intelligence in the Cloud](https://medium.com/recombee-blog/artificial-intelligence-in-the-cloud-310da5e0325e) At Recombee, we “think big”, and prefer making big leaps in technology over taking small steps. Our team has been involved in data science and artificial intelligence research for many years. Beginning in 2012, we began to capitalize our knowledge and experience, developing products which… ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Mar 30, 2016 Personalization Recommendation Engine --- # Recombee in 2020: New Features and Improvements | Blog > Source: https://www.recombee.com/blog/recombee-in-2020 > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # Recombee in 2020: New Features and Improvements ![](https://www.recombee.com/img/blog/authors/ondrejfiedler.png) Ondrej Fiedler Dec 30, 2020 Welcome to our annual blog post about the things we accomplished during the year, including a little bit of shameless bragging about some important newly-added features. We know that this year has been quite challenging for many people, including ourselves. However, today we want to focus entirely on the positive (no pun included) side of the year and the stuff we are the proudest of. ![](https://www.recombee.com/img/blog/recombee-in-2020/main.png) We acquired new tech hires from the beginning of the year, and they have become irreplaceable reinforcements to the teams. The business development team expanded, enabling Recombee to improve the service we are providing to our new and existing clients. We have also doubled our AI/ML team to stay ahead of competitors not only in the performance of recommendation algorithms but also to speed up research and development of new features requested by our customers. ## No-Code Widgets Integration As much as we believe that API integration of Recombee into your sites and applications has always been straightforward for programmers, we still felt a need to introduce an integration method that is more accessible for everybody, be it a programmer or a non-programmer. Following this philosophy, we have worked on the widget feature. ![](https://www.recombee.com/img/blog/recombee-in-2020/html-widget.png) We have extended the Recombee Admin Interface by a UI based ([WYSIWYG](https://en.wikipedia.org/wiki/WYSIWYG)) editor, much like Word, where you define/click the whole widget and its behavior. Each widget can be easily used just by deploying a generated code tag to your site. Once you deploy the widget code tag, you can tweak and modify widget behavior in our editor and deploy all the changes with a single click. No additional coding is needed! You can optimize widget behavior depending on the screen or site element size so that you can use a single widget configuration on both mobile and web versions of your site. Our UI/UX team worked countless hours to bring the most polished and streamlined version of the UI widget editor that we believe you will enjoy using. We invite you to give us feedback on existing features and possibly request new features, as we care about your experience using our product. ## Personalized Full-Text Search You can probably imagine that when a programmer searches for “python”, and when a reptiles enthusiast searches for “python”, they both expect very different results. And that is why we introduced a new API endpoint for personalized full-text search! ![](https://www.recombee.com/img/blog/recombee-in-2020/search.png) Traditional search solutions take into account only the user’s search query, but our search solution also takes into account additional user data such as interaction history and properties, and can therefore return exactly what the user is looking for. Another benefit is that you can combine the search capabilities with other Recombee features, such as [Filters and Boosters](https://docs.recombee.com/scenarios#scenario-settings-in-the-admin-ui), for applying your business rules (e.g., slightly preferring new content or products with higher margin). If you like to learn more about the algorithms behind personalized search, please refer to our [recent blog post](https://medium.com/recombee-blog/introduction-to-personalized-search-2b70eb5fa5ae). ## Improvements in Scalability: We Can Now Handle Traffic of ANY Site In 2020 we started to work with clients that require a significantly higher number of recommendation requests than we used to process in the previous years. Although Recombee is designed from the very beginning to be a very scalable solution, we had to introduce several new improvements to handle dozens of thousands of recommendation requests per second and similar amounts of interactions. ![](https://www.recombee.com/img/blog/recombee-in-2020/scalability.png) What is even more challenging is that these recommendation requests need to be served immediately, providing results within tens of milliseconds. Within this limited time, complex computations to calculate the result need to be performed. Our engine is simultaneously updated in a real-time fashion with incoming user interactions, and models are being re-trained on the fly. All these computations are performed on a server infrastructure that has to be robust and self-healing because Recombee provides strict availability guarantees to many enterprise customers. We believe that it is not just bragging when we say that we are now capable of providing recommendations for any website or app in the world - of course while maintaining the quality of the recommendations and providing just the right content or products for each individual user. ## Infrastructure on the US West Coast We have a fast-growing customer base in the US, and establishing a second server cluster in North America was yet another step to boost this trend. We can now serve our US customers with minimal network latencies, as we operate infrastructures on both the East and the West Coast. Besides the two North American infrastructures, we also offer clusters in Europe and Australia. ![](https://www.recombee.com/img/blog/recombee-in-2020/infrastructure.png) ## Improvements in Recommendation Models Last year, we added several [new algorithms](https://medium.com/recombee-blog/recombee-in-2019-new-features-and-improvements-bcbfc85acb80) such as real-time deep learning models, contextual bandits or deep reinforcement learning to our portfolio. In 2020 we have been improving the existing models and work on end-to-end recommendation models. Our research of [next basket prediction](https://medium.com/recombee-blog/deep-learning-for-recommender-systems-next-basket-prediction-and-sequential-product-recommendation-796228b34dee) involves large scale evaluation of different architectures that are computationally expensive. Thanks to our partnership with NVIDIA, we can utilize advanced GPU clusters for research purposes. ![](https://www.recombee.com/img/blog/recombee-in-2020/models.png) We are also participating in an interdisciplinary research team (that includes also UX specialists and journalists) which is working on analyzing and modeling the impact of editors and the recommender engine on the audience and studying the effects in behavioral data (user interactions). We work with several customers, where Recombee personalizes everything, including the front page. Hence, including the preferences of editors, curators and studying the impact on the audience is of crucial importance. The goal is to research and develop a powerful recsys human-machine interface enabling our customers (not exclusively in media) to understand better which content makes their users happy in all niches. Also, we aim to visualize how different recommendation strategies and rules that our customers use to adjust the recommender system impact their end-users. We believe this is an important step towards explainability in recommender systems and the reduction of various biases that have negative impact on the end-users. [Let us know](mailto:research@recombee.com) if you would like to work with us. ## See You In 2021! Our team is working on new disruptive features that promise to make our customers more successful with their products. Some features, such as next basket prediction, will open new opportunities and markets for them. We also work hard to democratize our AI technology, so you do not need to be a programmer to deploy and manage Recombee to enhance your product. We wish you a Happy New Year and stay safe! ![](https://www.recombee.com/img/blog/recombee-in-2020/hny.png) New Features ## Next Articles [![](https://www.recombee.com/img/blog/recombee-xperience-kentico.png)](https://xperience.io/discover/blog/recombee-kentico-xperience-1-on-1-personalization) ### [Recombee and Kentico Xperience: Guide to One-On-One Personalization](https://xperience.io/discover/blog/recombee-kentico-xperience-1-on-1-personalization) Recombee expanded its integration options - and now is available at the Kentico Xperience platform! Analyzing different types of personalization, we look into why Kentiko chose our AI-powered recommendation engine over manual segmentation. ![](https://www.recombee.com/img/blog/authors/gabrielatakacova.png) Gabriela Takacova May 6, 2021 Personalization Recommendation Engine [![](https://www.recombee.com/img/blog/23d22w82u3d52as.png)](https://medium.com/recombee-blog/deep-learning-for-recommender-systems-next-basket-prediction-and-sequential-product-recommendation-796228b34dee) ### [Deep Learning for Recommender Systems: Next Basket Prediction and Sequential Product Recommendation](https://medium.com/recombee-blog/deep-learning-for-recommender-systems-next-basket-prediction-and-sequential-product-recommendation-796228b34dee) Accurate “next basket prediction” will be enabling next generation e-commerce — predictive shopping and logistics. In this blogpost, we will discuss the deep learning technology behind next basket... ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Nov 10, 2020 Personalization Recommendation Engine [![](https://www.recombee.com/img/blog/how-interdisciplinary-collaboration-can-accelerate-ai-innovation.png)](https://www.recombee.com/blog/how-interdisciplinary-collaboration-can-accelerate-ai-innovation) ### [How Interdisciplinary Collaboration Can Accelerate AI Innovation](https://www.recombee.com/blog/how-interdisciplinary-collaboration-can-accelerate-ai-innovation) In a world where innovation is the new standard, Recombee uses the power of interdisciplinary collaboration to stay at the cutting edge of innovation. Partnering up with the leading player in the food industry (Bofrost) and academia (FIT CTU), allowed Recombee to hold a student competition to create AI which can shape the future of the food industry. ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Sep 1, 2020 --- # Recombeelab's 2023 Research Publications | Blog > Source: https://www.recombee.com/blog/recombeelabs-2023-research-publications > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Blog**](https://www.recombee.com/blog) # Recombeelab's 2023 Research Publications ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Jan 19, 2024 ![](https://www.recombee.com/img/blog/recombeelabs-2023-research-publications/main.png) Recombeelab, a joint research laboratory of Recombee and the Faculty of Information Technology at the Czech Technical University in Prague, experienced a highly productive year in 2023, publishing a series of insightful and impactful papers in the field of recommendation systems. Our research spans various types, ranging from (A) basic research, where we aim to understand phenomena without immediate practical application, to (B) fundamental research, where we develop methods applicable to general tasks, and (C) practical research, where we address specific problems (often tied to customer needs). ![](https://www.recombee.com/img/blog/recombeelabs-2023-research-publications/01.png) In the first category (A), for example, we made a significant contribution by introducing generalization bounds for inductive matrix completion in low-noise settings \[1\]. This contribution is important to the field as it establishes a theoretical framework for comprehending the performance of matrix completion algorithms under various noise levels. In simpler terms, the authors of these papers aim to understand how well the prediction performance of matrix completion (a vital method for recommender systems) behaves in different scenarios. Similar insights are presented in \[2\], where the focus is on understanding how the model behaves given uncertainties in the observation of user-item interactions. This involves scenarios where some users may be more reliable than others when rating items in the catalog. In the realm of (B), fundamental research, our researchers have made valuable contributions in various aspects. For example, in \[3\], the authors propose a new factorization method that incorporates graph neural networks into the learning process. This algorithm proves to be more adept at handling incomplete data and is more robust to outliers compared to traditional matrix completion algorithms. Furthermore, \[4\] introduces a novel approach to addressing the cold-start problem in recommendation systems. This method utilizes ontologies and knowledge graphs to infer user preferences from incomplete data. Lastly, \[5\] presents a new algorithm designed to enhance the diversity and serendipity of recommendations in cold start environments. Employing an ontology-based approach, this algorithm identifies and recommends items likely to pique the interest of users. Finally, in the domain of (C), practical research, Recombeelab also has made significant contributions. For example, in \[6\], the authors proposed a framework to minimize biases, including gender bias, ensuring a fair ranking that balances the visibility of items across all groups. For instance, traditional recommender systems often exhibited bias towards male candidates for high job positions due to biased data. However, this method focuses on providing exposure guarantees for all involved groups, addressing biases and aligning with the principles of responsible AI in today's context. Moving on to \[7\], the authors aimed to enhance our understanding of the real-world performance of recommender systems by introducing a new metric for evaluation. This metric is time-dependent and free from popularity bias, making it more accurate for real-life scenarios. Additionally, \[8\] extends a shallow autoencoder (also developed by our group) algorithm for collaborative filtering. This scalable and easily explainable algorithm proves to be a practical choice for real-world recommender systems. Moreover, \[9\] introduces a new algorithm designed to improve the diversity and serendipity of recommendations in cold start environments. This algorithm employs an ontology-based approach to identify and recommend items likely to be of interest to users. ## Overall Impact These publications represent a significant contribution to the field of recommendation systems. They provide new theoretical insights, propose novel algorithms, and address important theoretical and practical problems. Recombeelab is one of the leading research laboratories in this field, and our work has the potential to significantly improve the performance and usability of recommender systems. Regarding the venue of publication, RecombeeLab has submitted to high-impact journals such as the IEEE Transactions on Neural Networks and Learning Systems and Expert Systems with Applications. As for conferences, in the year 2023, we have published papers in premier recommender systems conferences such as ACM RecSys and the prestigious AAAI Conference on Artificial Intelligence (considered one of the top-5 best conferences in the AI field). These journals and conferences are widely recognized by the research community for their impactful contributions, and their high rejection rates indicate the rigorous selection process ensuring only significant research is accepted. One highlight is the Best Student Paper award received by Petr Kasalicky in the Recommender Systems Evaluation Workshop held at the prestigious Conference on Knowledge Discovery and Data Mining (KDD). Kasalicky, an industrial PhD student in RecombeeLab, is actively involved in important projects crucial to Recombee. ![](https://www.recombee.com/img/blog/recombeelabs-2023-research-publications/02.png) We strongly believe in the necessity of building an international, inclusive, accessible, multicultural, and prejudice-free environment for research. In 2023, our international research collaborations were co-authored by individuals from seven different nationalities and affiliated with eight institutions located in four countries. We remain committed to exploring new collaborations and making our lab accessible to researchers from diverse backgrounds. If you are interested in learning more about our research, please feel free to [contact us](mailto:research@recombee.com). ## References \[1\] Ledent, Antoine, Rodrigo Alves, Yunwen Lei, Yann Guermeur, and Marius Kloft. "Generalization bounds for inductive matrix completion in low-noise settings." In Proceedings of the AAAI Conference on Artificial Intelligence, vol. 37, no. 7, pp. 8447-8455\. 2023. \[2\] Alves, Rodrigo, Antoine Ledent, and Marius Kloft. "Uncertainty-Adjusted Recommendation via Matrix Factorization With Weighted Losses." IEEE Transactions on Neural Networks and Learning Systems (2023). \[3\] Kasalicky, Petr, Antoine Ledent, and Rodrigo Alves. "Uncertainty-adjusted Inductive Matrix Completion with Graph Neural Networks." In Proceedings of the 17th ACM Conference on Recommender Systems, pp. 1169-1174\. 2023. \[4\] Kuznetsov, Stanislav, and Pavel Kordik. "Overcoming the cold-start problem in recommendation systems with ontologies and knowledge graphs." In European Conference on Advances in Databases and Information Systems, pp. 591-603\. Cham: Springer Nature Switzerland, 2023. \[5\] Kuznetsov, Stanislav, and Pavel Kordik. "Improving recommendation diversity and serendipity with an Ontology-based algorithm for cold start environments." International Journal of Data Science and Analytics (2023): 1-13. \[6\] Lopes, Ramon, Rodrigo Alves, Antoine Ledent, Rodrygo LT Santos, and Marius Kloft. "Recommendations with minimum exposure guarantees: A post-processing framework." Expert Systems with Applications 236 (2024): 121164. \[7\] Kasalicky, Petr, Rodrigo Alves, and Pavel Kordik. "Bridging Offline-Online Evaluation with a Time-dependent and Popularity Bias-free Offline Metric for Recommenders." arXiv preprint arXiv:2308.06885 (2023). \[8\] Vancura, Vojtech. "Scalable and Explainable Linear Shallow Autoencoders for Collaborative Filtering from Industrial Perspective." In Proceedings of the 31st ACM Conference on User Modeling, Adaptation and Personalization, pp. 290-295\. 2023. \[9\] Zid, Cenek, Pavel Kordik, and Stanislav Kuznetsov. "Personalised Recommendations and Profile Based Re-ranking Improve Distribution of Student Opportunities." In Computational Intelligence in Security for Information Systems Conference, pp. 217-227\. Cham: Springer Nature Switzerland, 2023. Recommendation Engine Personalization ## Next Articles [![](https://www.recombee.com/img/blog/is-this-comment-useful-enhancing-personalized-recommendations-by-considering-user-rating-uncertainty.png)](https://www.recombee.com/blog/is-this-comment-useful-enhancing-personalized-recommendations-by-considering-user-rating-uncertainty) ### [Is This Comment Useful? Enhancing Personalized Recommendations by Considering User Rating Uncertainty](https://www.recombee.com/blog/is-this-comment-useful-enhancing-personalized-recommendations-by-considering-user-rating-uncertainty) Picture this: you're on the hunt for the perfect new smartphone, browsing through your favourite online electronics store. The online store’s recommendation engine pops up with what it thinks could be your possible next gadget love... ![](https://www.recombee.com/img/blog/authors/rodrigoalves.png) Rodrigo Alves Mar 01, 2024 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/ai-news-and-outlook-for-2024.png)](https://www.recombee.com/blog/ai-news-and-outlook-for-2024) ### [AI News and Outlook for 2024](https://www.recombee.com/blog/ai-news-and-outlook-for-2024) We look at the most interesting research directions and assess the state of knowledge in key areas of AI. We'll also estimate future developments in 2024 so you know what to prepare for. ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Jan 16, 2024 Recommendation Engine Personalization [![](https://www.recombee.com/img/blog/ai-assistants-know-your-preferences-even-better-than-you-do.png)](https://www.recombee.com/blog/ai-assistants-know-your-preferences-even-better-than-you-do) ### [AI Assistants Know Your Preferences, Even Better Than You Do](https://www.recombee.com/blog/ai-assistants-know-your-preferences-even-better-than-you-do) Recommender systems and ethical controversies ![](https://www.recombee.com/img/blog/authors/pavelkordik.png) Pavel Kordik Nov 23, 2023 Recommendation Engine Personalization --- # FAQ > Source: https://www.recombee.com/faq > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # FAQ [All->](https://www.recombee.com/faq) [How It Works->](https://www.recombee.com/faq/how-it-works) [Recommendations->](https://www.recombee.com/faq/recommendations) [Personalized Search->](https://www.recombee.com/faq/personalized-search) [Features->](https://www.recombee.com/faq/features) FAQ * [We cache recommendations to reduce server load. Could that actually be hurting our engagement numbers over time?](https://www.recombee.com/faq/does-caching-recommendations-hurt-engagement) [How It Works](https://www.recombee.com/faq/how-it-works) [Explore->](https://www.recombee.com/faq/does-caching-recommendations-hurt-engagement) * [We're launching new products every week. How quickly can they actually get recommended to the right users if they have no clicks yet?](https://www.recombee.com/faq/how-quickly-can-new-products-get-recommended) [Features](https://www.recombee.com/faq/features) [Explore->](https://www.recombee.com/faq/how-quickly-can-new-products-get-recommended) * [Beyond clicks, what interaction data should we actually be sending in to get meaningfully better recommendations?](https://www.recombee.com/faq/what-interaction-data-should-we-send) [How It Works](https://www.recombee.com/faq/how-it-works) [Explore->](https://www.recombee.com/faq/what-interaction-data-should-we-send) * [A lot of our users aren't logged in. Do recommendations for them just default to "most popular" and stop there?](https://www.recombee.com/faq/how-do-you-handle-recommendations-for-anonymous-users) [How It Works](https://www.recombee.com/faq/how-it-works) [Explore->](https://www.recombee.com/faq/how-do-you-handle-recommendations-for-anonymous-users) * [My users don't behave the same way across a session - sometimes they're in research mode, sometimes just browsing. Can a recommender actually handle that, or does it just pick one mode and stick with it?](https://www.recombee.com/faq/how-do-you-handle-different-user-behaviors-in-a-session) [Recommendations](https://www.recombee.com/faq/recommendations) [Explore->](https://www.recombee.com/faq/how-do-you-handle-different-user-behaviors-in-a-session) * [We have freemium users I want to convert to paid subscribers. Can I actually tune the recommender to push them toward subscription - or is that too manual to set up?](https://www.recombee.com/faq/how-do-you-handle-freemium-user-conversion) [Recommendations](https://www.recombee.com/faq/recommendations) [Explore->](https://www.recombee.com/faq/how-do-you-handle-freemium-user-conversion) * [When I hear "targeted ads", I assume the recommender is behind it. Is that actually how it works?](https://www.recombee.com/faq/how-do-you-handle-targeted-ads) [How It Works](https://www.recombee.com/faq/how-it-works) [Explore->](https://www.recombee.com/faq/how-do-you-handle-targeted-ads) * [Is the growth in recommender system adoption a trend we need to take seriously, or is it already plateaued?](https://www.recombee.com/faq/is-the-growth-in-recommender-system-adoption-a-trend-we-need-to-take-seriously) [How It Works](https://www.recombee.com/faq/how-it-works) [Explore->](https://www.recombee.com/faq/is-the-growth-in-recommender-system-adoption-a-trend-we-need-to-take-seriously) * [What types of data sources does a modern recommender system rely on to generate personalized recommendations?](https://www.recombee.com/faq/what-data-sources-does-a-modern-recommender-system-rely-on) [How It Works](https://www.recombee.com/faq/how-it-works) [Explore->](https://www.recombee.com/faq/what-data-sources-does-a-modern-recommender-system-rely-on) * [Why are historical items that are no longer available to users still stored in the item catalog?](https://www.recombee.com/faq/why-are-historical-items-still-stored-in-the-item-catalog) [How It Works](https://www.recombee.com/faq/how-it-works) [Explore->](https://www.recombee.com/faq/why-are-historical-items-still-stored-in-the-item-catalog) * [How can item categories be used to control which recommendations a user sees?](https://www.recombee.com/faq/how-can-item-categories-be-used-to-control-which-recommendations-a-user-sees) [Features](https://www.recombee.com/faq/features) [Explore->](https://www.recombee.com/faq/how-can-item-categories-be-used-to-control-which-recommendations-a-user-sees) * [What role do neural text embeddings play in recommending items that have few or no user interactions?](https://www.recombee.com/faq/how-do-neural-text-embeddings-help-with-cold-start-recommendations) [Recommendations](https://www.recombee.com/faq/recommendations) [Explore->](https://www.recombee.com/faq/how-do-neural-text-embeddings-help-with-cold-start-recommendations) * [In what recommendation scenarios does a user's geographic location become a critical input?](https://www.recombee.com/faq/how-does-geographic-location-affect-recommendation-quality) [Recommendations](https://www.recombee.com/faq/recommendations) [Explore->](https://www.recombee.com/faq/how-does-geographic-location-affect-recommendation-quality) * [What strategic risk does neglecting cold-start item coverage create for a product catalog-driven business?](https://www.recombee.com/faq/how-does-neglecting-cold-start-coverage-affect-business-outcomes) [How It Works](https://www.recombee.com/faq/how-it-works) [Explore->](https://www.recombee.com/faq/how-does-neglecting-cold-start-coverage-affect-business-outcomes) * [How should a business weigh GDPR and data privacy requirements against the need for rich user data to drive personalization?](https://www.recombee.com/faq/how-should-a-business-weigh-gdpr-and-data-privacy-requirements) [How It Works](https://www.recombee.com/faq/how-it-works) [Explore->](https://www.recombee.com/faq/how-should-a-business-weigh-gdpr-and-data-privacy-requirements) * [What is the strategic value of investing in image-based neural embeddings for a marketplace where sellers upload their own product photos?](https://www.recombee.com/faq/what-is-the-strategic-value-of-investing-in-image-based-neural-embeddings) [Recommendations](https://www.recombee.com/faq/recommendations) [Explore->](https://www.recombee.com/faq/what-is-the-strategic-value-of-investing-in-image-based-neural-embeddings) * [How does incorporating user background attributes such as skills or interests affect recommendation quality in domains with sparse interaction data?](https://www.recombee.com/faq/how-does-incorporating-user-background-attributes-affect-recommendation-quality) [How It Works](https://www.recombee.com/faq/how-it-works) [Explore->](https://www.recombee.com/faq/how-does-incorporating-user-background-attributes-affect-recommendation-quality) * [How do recommendation objectives get defined for a specific platform and its individual use cases?](https://www.recombee.com/faq/how-do-recommendation-objectives-get-defined-for-a-specific-platform) [How It Works](https://www.recombee.com/faq/how-it-works) [Explore->](https://www.recombee.com/faq/how-do-recommendation-objectives-get-defined-for-a-specific-platform) * [Can recommendation systems be used to convert free users into paying subscribers?](https://www.recombee.com/faq/can-recommendation-systems-be-used-to-convert-free-users-into-paying) [Recommendations](https://www.recombee.com/faq/recommendations) [Explore->](https://www.recombee.com/faq/can-recommendation-systems-be-used-to-convert-free-users-into-paying) * [What makes recommendation objective design particularly complex for platforms like job boards or dating sites?](https://www.recombee.com/faq/what-makes-recommendation-design-complex-for-job-boards-dating-sites) [How It Works](https://www.recombee.com/faq/how-it-works) [Explore->](https://www.recombee.com/faq/what-makes-recommendation-design-complex-for-job-boards-dating-sites) * [How do content streaming platforms balance supporting niche creators with optimizing for mainstream user engagement?](https://www.recombee.com/faq/how-do-content-streaming-platforms-balance-niche-creators-and-engagement) [Recommendations](https://www.recombee.com/faq/recommendations) [Explore->](https://www.recombee.com/faq/how-do-content-streaming-platforms-balance-niche-creators-and-engagement) * [Why is content discovery a standalone recommendation objective rather than a byproduct of relevance optimization?](https://www.recombee.com/faq/why-is-content-discovery-a-standalone-recommendation-objective) [Personalized Search](https://www.recombee.com/faq/personalized-search) [Explore->](https://www.recombee.com/faq/why-is-content-discovery-a-standalone-recommendation-objective) * [How did recommender systems originate, and what distinguished early systems from modern personalized ones?](https://www.recombee.com/faq/how-did-recommender-systems-originate) [How It Works](https://www.recombee.com/faq/how-it-works) [Explore->](https://www.recombee.com/faq/how-did-recommender-systems-originate) * [What is driving the continued growth in the volume of recommendations served to online users?](https://www.recombee.com/faq/what-is-driving-the-continued-growth-in-the-volume-of-recommendations) [How It Works](https://www.recombee.com/faq/how-it-works) [Explore->](https://www.recombee.com/faq/what-is-driving-the-continued-growth-in-the-volume-of-recommendations) * [Are targeted advertisements the same thing as AI-powered recommender systems?](https://www.recombee.com/faq/are-targeted-advertisements-the-same-thing-as-ai-powered-recommender-systems) [How It Works](https://www.recombee.com/faq/how-it-works) [Explore->](https://www.recombee.com/faq/are-targeted-advertisements-the-same-thing-as-ai-powered-recommender-systems) * [What is the strategic significance of recommendation systems becoming pervasive across virtually every major online platform?](https://www.recombee.com/faq/what-is-the-strategic-significance-of-recommendation-systems-becoming-pervasive) [How It Works](https://www.recombee.com/faq/how-it-works) [Explore->](https://www.recombee.com/faq/what-is-the-strategic-significance-of-recommendation-systems-becoming-pervasive) * [What was the significance of the GroupLens system in the history of personalized recommendations?](https://www.recombee.com/faq/what-was-the-significance-of-the-group-lens-system) [How It Works](https://www.recombee.com/faq/how-it-works) [Explore->](https://www.recombee.com/faq/what-was-the-significance-of-the-group-lens-system) Haven’t Found The Information You Are Looking For? --- # Features | FAQ > Source: https://www.recombee.com/faq/features > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # FAQ [All->](https://www.recombee.com/faq) [How It Works->](https://www.recombee.com/faq/how-it-works) [Recommendations->](https://www.recombee.com/faq/recommendations) [Personalized Search->](https://www.recombee.com/faq/personalized-search) [Features->](https://www.recombee.com/faq/features) FAQ * [We're launching new products every week. How quickly can they actually get recommended to the right users if they have no clicks yet?](https://www.recombee.com/faq/how-quickly-can-new-products-get-recommended) [Features](https://www.recombee.com/faq/features) [Explore->](https://www.recombee.com/faq/how-quickly-can-new-products-get-recommended) * [How can item categories be used to control which recommendations a user sees?](https://www.recombee.com/faq/how-can-item-categories-be-used-to-control-which-recommendations-a-user-sees) [Features](https://www.recombee.com/faq/features) [Explore->](https://www.recombee.com/faq/how-can-item-categories-be-used-to-control-which-recommendations-a-user-sees) Haven’t Found The Information You Are Looking For? --- # How It Works | FAQ > Source: https://www.recombee.com/faq/how-it-works > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # FAQ [All->](https://www.recombee.com/faq) [How It Works->](https://www.recombee.com/faq/how-it-works) [Recommendations->](https://www.recombee.com/faq/recommendations) [Personalized Search->](https://www.recombee.com/faq/personalized-search) [Features->](https://www.recombee.com/faq/features) FAQ * [We cache recommendations to reduce server load. Could that actually be hurting our engagement numbers over time?](https://www.recombee.com/faq/does-caching-recommendations-hurt-engagement) [How It Works](https://www.recombee.com/faq/how-it-works) [Explore->](https://www.recombee.com/faq/does-caching-recommendations-hurt-engagement) * [Beyond clicks, what interaction data should we actually be sending in to get meaningfully better recommendations?](https://www.recombee.com/faq/what-interaction-data-should-we-send) [How It Works](https://www.recombee.com/faq/how-it-works) [Explore->](https://www.recombee.com/faq/what-interaction-data-should-we-send) * [A lot of our users aren't logged in. Do recommendations for them just default to "most popular" and stop there?](https://www.recombee.com/faq/how-do-you-handle-recommendations-for-anonymous-users) [How It Works](https://www.recombee.com/faq/how-it-works) [Explore->](https://www.recombee.com/faq/how-do-you-handle-recommendations-for-anonymous-users) * [When I hear "targeted ads", I assume the recommender is behind it. Is that actually how it works?](https://www.recombee.com/faq/how-do-you-handle-targeted-ads) [How It Works](https://www.recombee.com/faq/how-it-works) [Explore->](https://www.recombee.com/faq/how-do-you-handle-targeted-ads) * [Is the growth in recommender system adoption a trend we need to take seriously, or is it already plateaued?](https://www.recombee.com/faq/is-the-growth-in-recommender-system-adoption-a-trend-we-need-to-take-seriously) [How It Works](https://www.recombee.com/faq/how-it-works) [Explore->](https://www.recombee.com/faq/is-the-growth-in-recommender-system-adoption-a-trend-we-need-to-take-seriously) * [What types of data sources does a modern recommender system rely on to generate personalized recommendations?](https://www.recombee.com/faq/what-data-sources-does-a-modern-recommender-system-rely-on) [How It Works](https://www.recombee.com/faq/how-it-works) [Explore->](https://www.recombee.com/faq/what-data-sources-does-a-modern-recommender-system-rely-on) * [Why are historical items that are no longer available to users still stored in the item catalog?](https://www.recombee.com/faq/why-are-historical-items-still-stored-in-the-item-catalog) [How It Works](https://www.recombee.com/faq/how-it-works) [Explore->](https://www.recombee.com/faq/why-are-historical-items-still-stored-in-the-item-catalog) * [What strategic risk does neglecting cold-start item coverage create for a product catalog-driven business?](https://www.recombee.com/faq/how-does-neglecting-cold-start-coverage-affect-business-outcomes) [How It Works](https://www.recombee.com/faq/how-it-works) [Explore->](https://www.recombee.com/faq/how-does-neglecting-cold-start-coverage-affect-business-outcomes) * [How should a business weigh GDPR and data privacy requirements against the need for rich user data to drive personalization?](https://www.recombee.com/faq/how-should-a-business-weigh-gdpr-and-data-privacy-requirements) [How It Works](https://www.recombee.com/faq/how-it-works) [Explore->](https://www.recombee.com/faq/how-should-a-business-weigh-gdpr-and-data-privacy-requirements) * [How does incorporating user background attributes such as skills or interests affect recommendation quality in domains with sparse interaction data?](https://www.recombee.com/faq/how-does-incorporating-user-background-attributes-affect-recommendation-quality) [How It Works](https://www.recombee.com/faq/how-it-works) [Explore->](https://www.recombee.com/faq/how-does-incorporating-user-background-attributes-affect-recommendation-quality) * [How do recommendation objectives get defined for a specific platform and its individual use cases?](https://www.recombee.com/faq/how-do-recommendation-objectives-get-defined-for-a-specific-platform) [How It Works](https://www.recombee.com/faq/how-it-works) [Explore->](https://www.recombee.com/faq/how-do-recommendation-objectives-get-defined-for-a-specific-platform) * [What makes recommendation objective design particularly complex for platforms like job boards or dating sites?](https://www.recombee.com/faq/what-makes-recommendation-design-complex-for-job-boards-dating-sites) [How It Works](https://www.recombee.com/faq/how-it-works) [Explore->](https://www.recombee.com/faq/what-makes-recommendation-design-complex-for-job-boards-dating-sites) * [How did recommender systems originate, and what distinguished early systems from modern personalized ones?](https://www.recombee.com/faq/how-did-recommender-systems-originate) [How It Works](https://www.recombee.com/faq/how-it-works) [Explore->](https://www.recombee.com/faq/how-did-recommender-systems-originate) * [What is driving the continued growth in the volume of recommendations served to online users?](https://www.recombee.com/faq/what-is-driving-the-continued-growth-in-the-volume-of-recommendations) [How It Works](https://www.recombee.com/faq/how-it-works) [Explore->](https://www.recombee.com/faq/what-is-driving-the-continued-growth-in-the-volume-of-recommendations) * [Are targeted advertisements the same thing as AI-powered recommender systems?](https://www.recombee.com/faq/are-targeted-advertisements-the-same-thing-as-ai-powered-recommender-systems) [How It Works](https://www.recombee.com/faq/how-it-works) [Explore->](https://www.recombee.com/faq/are-targeted-advertisements-the-same-thing-as-ai-powered-recommender-systems) * [What is the strategic significance of recommendation systems becoming pervasive across virtually every major online platform?](https://www.recombee.com/faq/what-is-the-strategic-significance-of-recommendation-systems-becoming-pervasive) [How It Works](https://www.recombee.com/faq/how-it-works) [Explore->](https://www.recombee.com/faq/what-is-the-strategic-significance-of-recommendation-systems-becoming-pervasive) * [What was the significance of the GroupLens system in the history of personalized recommendations?](https://www.recombee.com/faq/what-was-the-significance-of-the-group-lens-system) [How It Works](https://www.recombee.com/faq/how-it-works) [Explore->](https://www.recombee.com/faq/what-was-the-significance-of-the-group-lens-system) Haven’t Found The Information You Are Looking For? --- # Personalized Search | FAQ > Source: https://www.recombee.com/faq/personalized-search > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # FAQ [All->](https://www.recombee.com/faq) [How It Works->](https://www.recombee.com/faq/how-it-works) [Recommendations->](https://www.recombee.com/faq/recommendations) [Personalized Search->](https://www.recombee.com/faq/personalized-search) [Features->](https://www.recombee.com/faq/features) FAQ * [Why is content discovery a standalone recommendation objective rather than a byproduct of relevance optimization?](https://www.recombee.com/faq/why-is-content-discovery-a-standalone-recommendation-objective) [Personalized Search](https://www.recombee.com/faq/personalized-search) [Explore->](https://www.recombee.com/faq/why-is-content-discovery-a-standalone-recommendation-objective) Haven’t Found The Information You Are Looking For? --- # Recommendations | FAQ > Source: https://www.recombee.com/faq/recommendations > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # FAQ [All->](https://www.recombee.com/faq) [How It Works->](https://www.recombee.com/faq/how-it-works) [Recommendations->](https://www.recombee.com/faq/recommendations) [Personalized Search->](https://www.recombee.com/faq/personalized-search) [Features->](https://www.recombee.com/faq/features) FAQ * [My users don't behave the same way across a session - sometimes they're in research mode, sometimes just browsing. Can a recommender actually handle that, or does it just pick one mode and stick with it?](https://www.recombee.com/faq/how-do-you-handle-different-user-behaviors-in-a-session) [Recommendations](https://www.recombee.com/faq/recommendations) [Explore->](https://www.recombee.com/faq/how-do-you-handle-different-user-behaviors-in-a-session) * [We have freemium users I want to convert to paid subscribers. Can I actually tune the recommender to push them toward subscription - or is that too manual to set up?](https://www.recombee.com/faq/how-do-you-handle-freemium-user-conversion) [Recommendations](https://www.recombee.com/faq/recommendations) [Explore->](https://www.recombee.com/faq/how-do-you-handle-freemium-user-conversion) * [What role do neural text embeddings play in recommending items that have few or no user interactions?](https://www.recombee.com/faq/how-do-neural-text-embeddings-help-with-cold-start-recommendations) [Recommendations](https://www.recombee.com/faq/recommendations) [Explore->](https://www.recombee.com/faq/how-do-neural-text-embeddings-help-with-cold-start-recommendations) * [In what recommendation scenarios does a user's geographic location become a critical input?](https://www.recombee.com/faq/how-does-geographic-location-affect-recommendation-quality) [Recommendations](https://www.recombee.com/faq/recommendations) [Explore->](https://www.recombee.com/faq/how-does-geographic-location-affect-recommendation-quality) * [What is the strategic value of investing in image-based neural embeddings for a marketplace where sellers upload their own product photos?](https://www.recombee.com/faq/what-is-the-strategic-value-of-investing-in-image-based-neural-embeddings) [Recommendations](https://www.recombee.com/faq/recommendations) [Explore->](https://www.recombee.com/faq/what-is-the-strategic-value-of-investing-in-image-based-neural-embeddings) * [Can recommendation systems be used to convert free users into paying subscribers?](https://www.recombee.com/faq/can-recommendation-systems-be-used-to-convert-free-users-into-paying) [Recommendations](https://www.recombee.com/faq/recommendations) [Explore->](https://www.recombee.com/faq/can-recommendation-systems-be-used-to-convert-free-users-into-paying) * [How do content streaming platforms balance supporting niche creators with optimizing for mainstream user engagement?](https://www.recombee.com/faq/how-do-content-streaming-platforms-balance-niche-creators-and-engagement) [Recommendations](https://www.recombee.com/faq/recommendations) [Explore->](https://www.recombee.com/faq/how-do-content-streaming-platforms-balance-niche-creators-and-engagement) Haven’t Found The Information You Are Looking For? --- # Are targeted advertisements the same thing as AI-powered recommender systems? | FAQ > Source: https://www.recombee.com/faq/are-targeted-advertisements-the-same-thing-as-ai-powered-recommender-systems > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # [FAQ](https://www.recombee.com/faq) [All->](https://www.recombee.com/faq) [How It Works->](https://www.recombee.com/faq/how-it-works) [Recommendations->](https://www.recombee.com/faq/recommendations) [Personalized Search->](https://www.recombee.com/faq/personalized-search) [Features->](https://www.recombee.com/faq/features) [FAQ](https://www.recombee.com/faq) [<- Back to List](https://www.recombee.com/faq) # Are targeted advertisements the same thing as AI-powered recommender systems? No. Targeted advertisements and AI-powered recommender systems are distinct technologies that are frequently conflated. **Many common ad formats - such as abandoned cart retargeting - rely on simple rule-based heuristics rather than machine learning.** Ads displayed on news and media sites are typically auctioned through AdTech platforms based on context and user profiles, not generated by a recommender system. For product teams, this distinction is operationally important: deploying a recommendation system is a different technical and ethical undertaking than running targeted ad campaigns, and the two should not be governed by the same assumptions. [How It Works](https://www.recombee.com/faq/how-it-works) Haven’t Found The Information You Are Looking For? ## Recommended Topics for You --- # Can recommendation systems be used to convert free users into paying subscribers? | FAQ > Source: https://www.recombee.com/faq/can-recommendation-systems-be-used-to-convert-free-users-into-paying > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # [FAQ](https://www.recombee.com/faq) [All->](https://www.recombee.com/faq) [How It Works->](https://www.recombee.com/faq/how-it-works) [Recommendations->](https://www.recombee.com/faq/recommendations) [Personalized Search->](https://www.recombee.com/faq/personalized-search) [Features->](https://www.recombee.com/faq/features) [FAQ](https://www.recombee.com/faq) [<- Back to List](https://www.recombee.com/faq) # Can recommendation systems be used to convert free users into paying subscribers? Yes. Recommendation systems can be explicitly optimized for subscription conversion, not just engagement. For free-tier users on subscription-based platforms, **recommending highly relevant content that exists beyond the paywall** is a documented strategy to drive conversion. This makes the recommendation engine a direct revenue tool rather than purely a retention mechanism. Product teams building freemium content platforms should treat conversion rate for free-to-paid users as a first-class objective when configuring or evaluating their recommendation system, alongside traditional engagement metrics like session duration or content completion. [Recommendations](https://www.recombee.com/faq/recommendations) Haven’t Found The Information You Are Looking For? ## Recommended Topics for You --- # We cache recommendations to reduce server load. Could that actually be hurting our engagement numbers over time? | FAQ > Source: https://www.recombee.com/faq/does-caching-recommendations-hurt-engagement > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # [FAQ](https://www.recombee.com/faq) [All->](https://www.recombee.com/faq) [How It Works->](https://www.recombee.com/faq/how-it-works) [Recommendations->](https://www.recombee.com/faq/recommendations) [Personalized Search->](https://www.recombee.com/faq/personalized-search) [Features->](https://www.recombee.com/faq/features) [FAQ](https://www.recombee.com/faq) [<- Back to List](https://www.recombee.com/faq) # We cache recommendations to reduce server load. Could that actually be hurting our engagement numbers over time? Yes, caching recommendations can degrade engagement over time if the system is not notified of repeated exposures. When a recommender receives no feedback on items it has already shown, it loses the signal that non-engagement carries - treating a skipped item identically to an unseen one. The result is a recommendation loop where items continue appearing despite repeated non-engagement. The fix is ensuring every cached impression is logged back to the recommender so it can factor repeated exposure into future ranking decisions. [How It Works](https://www.recombee.com/faq/how-it-works) Haven’t Found The Information You Are Looking For? ## Recommended Topics for You --- # How can item categories be used to control which recommendations a user sees? | FAQ > Source: https://www.recombee.com/faq/how-can-item-categories-be-used-to-control-which-recommendations-a-user-sees > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # [FAQ](https://www.recombee.com/faq) [All->](https://www.recombee.com/faq) [How It Works->](https://www.recombee.com/faq/how-it-works) [Recommendations->](https://www.recombee.com/faq/recommendations) [Personalized Search->](https://www.recombee.com/faq/personalized-search) [Features->](https://www.recombee.com/faq/features) [FAQ](https://www.recombee.com/faq) [<- Back to List](https://www.recombee.com/faq) # How can item categories be used to control which recommendations a user sees? Item categories give teams direct levers for shaping recommendation output. **Categories can be used to filter items out of results entirely, boost the probability that items from a specific category appear, or target particular category segments to a given user.** A hierarchical category structure is also supported, allowing one item to belong to multiple categories simultaneously. This means product teams can enforce business rules - such as promoting a seasonal category - without retraining the underlying model. Planning a well-structured category taxonomy at the start of an integration pays dividends in configuration flexibility later. [Features](https://www.recombee.com/faq/features) Haven’t Found The Information You Are Looking For? ## Recommended Topics for You --- # How did recommender systems originate, and what distinguished early systems from modern personalized ones? | FAQ > Source: https://www.recombee.com/faq/how-did-recommender-systems-originate > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # [FAQ](https://www.recombee.com/faq) [All->](https://www.recombee.com/faq) [How It Works->](https://www.recombee.com/faq/how-it-works) [Recommendations->](https://www.recombee.com/faq/recommendations) [Personalized Search->](https://www.recombee.com/faq/personalized-search) [Features->](https://www.recombee.com/faq/features) [FAQ](https://www.recombee.com/faq) [<- Back to List](https://www.recombee.com/faq) # How did recommender systems originate, and what distinguished early systems from modern personalized ones? Early recommender systems grew out of information retrieval (IR) systems in the early 1970s, and their defining limitation was that they produced the same output for every user. **The shift to personalization came when personal computers and widespread internet access made it possible to factor in individual user interaction histories.** One of the first systems to rely exclusively on user historical interactions was GroupLens in 1992, which used explicit article ratings. This historical progression matters practically: understanding that modern systems layer multiple techniques on top of that foundation helps teams set realistic expectations about what the technology requires to function well. [How It Works](https://www.recombee.com/faq/how-it-works) Haven’t Found The Information You Are Looking For? ## Recommended Topics for You --- # How do content streaming platforms balance supporting niche creators with optimizing for mainstream user engagement? | FAQ > Source: https://www.recombee.com/faq/how-do-content-streaming-platforms-balance-niche-creators-and-engagement > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # [FAQ](https://www.recombee.com/faq) [All->](https://www.recombee.com/faq) [How It Works->](https://www.recombee.com/faq/how-it-works) [Recommendations->](https://www.recombee.com/faq/recommendations) [Personalized Search->](https://www.recombee.com/faq/personalized-search) [Features->](https://www.recombee.com/faq/features) [FAQ](https://www.recombee.com/faq) [<- Back to List](https://www.recombee.com/faq) # How do content streaming platforms balance supporting niche creators with optimizing for mainstream user engagement? Content streaming platforms treat the balance between mainstream and niche content exposure as an explicit recommendation objective, promoting diverse content to serve varied user tastes while also supporting a healthy creator ecosystem. **Fair exposure and monetization for content creators is listed as a distinct optimization goal**, separate from pure engagement maximization. This means a well-designed streaming recommender is not solely optimizing for what the majority of users click on - it incorporates creator ecosystem health as a parallel objective. Platforms that ignore this balance risk concentrating traffic on a small slice of their catalog, which can reduce long-term content diversity and creator participation. [Recommendations](https://www.recombee.com/faq/recommendations) Haven’t Found The Information You Are Looking For? ## Recommended Topics for You --- # What role do neural text embeddings play in recommending items that have few or no user interactions? | FAQ > Source: https://www.recombee.com/faq/how-do-neural-text-embeddings-help-with-cold-start-recommendations > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # [FAQ](https://www.recombee.com/faq) [All->](https://www.recombee.com/faq) [How It Works->](https://www.recombee.com/faq/how-it-works) [Recommendations->](https://www.recombee.com/faq/recommendations) [Personalized Search->](https://www.recombee.com/faq/personalized-search) [Features->](https://www.recombee.com/faq/features) [FAQ](https://www.recombee.com/faq) [<- Back to List](https://www.recombee.com/faq) # What role do neural text embeddings play in recommending items that have few or no user interactions? Neural text embeddings allow recommender systems to surface relevant cold-start items - those with few or no interaction history - by computing similarity based on item descriptions rather than behavioral data. **When a new item lacks interaction signals, text-based neural embeddings can substitute as a proxy for relevance**, enabling the system to recommend it alongside established items without waiting for engagement to accumulate. This is particularly valuable in content-heavy domains such as news or article platforms where new items are published continuously. Teams launching new catalog items can therefore expect meaningful recommendation coverage from day one if rich text descriptions are provided. [Recommendations](https://www.recombee.com/faq/recommendations) Haven’t Found The Information You Are Looking For? ## Recommended Topics for You --- # How do recommendation objectives get defined for a specific platform and its individual use cases? | FAQ > Source: https://www.recombee.com/faq/how-do-recommendation-objectives-get-defined-for-a-specific-platform > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # [FAQ](https://www.recombee.com/faq) [All->](https://www.recombee.com/faq) [How It Works->](https://www.recombee.com/faq/how-it-works) [Recommendations->](https://www.recombee.com/faq/recommendations) [Personalized Search->](https://www.recombee.com/faq/personalized-search) [Features->](https://www.recombee.com/faq/features) [FAQ](https://www.recombee.com/faq) [<- Back to List](https://www.recombee.com/faq) # How do recommendation objectives get defined for a specific platform and its individual use cases? Recommendation objectives for a specific platform are typically defined through careful analysis of user needs, business requirements, and strategic goals, emerging from stakeholder discussions, user research, and business strategy sessions. **Multiple stakeholders with potentially conflicting interests - including users, content creators, editors, and the business - must be identified and balanced.** Effective optimization seeks to align these competing interests rather than serve any single party exclusively. This means objective-setting is fundamentally a cross-functional process, not a purely technical one, and should involve product, strategy, and busienss teams. [How It Works](https://www.recombee.com/faq/how-it-works) Haven’t Found The Information You Are Looking For? ## Recommended Topics for You --- # My users don't behave the same way across a session - sometimes they're in research mode, sometimes just browsing. Can a recommender actually handle that, or does it just pick one mode and stick with it? | FAQ > Source: https://www.recombee.com/faq/how-do-you-handle-different-user-behaviors-in-a-session > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # [FAQ](https://www.recombee.com/faq) [All->](https://www.recombee.com/faq) [How It Works->](https://www.recombee.com/faq/how-it-works) [Recommendations->](https://www.recombee.com/faq/recommendations) [Personalized Search->](https://www.recombee.com/faq/personalized-search) [Features->](https://www.recombee.com/faq/features) [FAQ](https://www.recombee.com/faq) [<- Back to List](https://www.recombee.com/faq) # My users don't behave the same way across a session - sometimes they're in research mode, sometimes just browsing. Can a recommender actually handle that, or does it just pick one mode and stick with it? A single recommendation scenario will not cover both modes well. Running multiple recommendation scenarios in parallel - one optimized for discovery, one for purchase intent - allows different parts of the interface to serve different user states simultaneously. Rather than attempting to detect which mode a user is in and switching between them, the standard approach is to design distinct scenarios for each context and place them in the appropriate locations. This separates the optimization objectives so neither goal is compromised by the other. [Recommendations](https://www.recombee.com/faq/recommendations) Haven’t Found The Information You Are Looking For? ## Recommended Topics for You --- # We have freemium users I want to convert to paid subscribers. Can I actually tune the recommender to push them toward subscription - or is that too manual to set up? | FAQ > Source: https://www.recombee.com/faq/how-do-you-handle-freemium-user-conversion > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # [FAQ](https://www.recombee.com/faq) [All->](https://www.recombee.com/faq) [How It Works->](https://www.recombee.com/faq/how-it-works) [Recommendations->](https://www.recombee.com/faq/recommendations) [Personalized Search->](https://www.recombee.com/faq/personalized-search) [Features->](https://www.recombee.com/faq/features) [FAQ](https://www.recombee.com/faq) [<- Back to List](https://www.recombee.com/faq) # We have freemium users I want to convert to paid subscribers. Can I actually tune the recommender to push them toward subscription - or is that too manual to set up? Yes. Recommendation systems can be configured to serve business objectives beyond engagement, including subscription conversion. For freemium users, the system can prioritize content behind the paywall or content that historically preceded subscription in similar users' journeys. Both approaches use existing data - paywall boundaries and past conversion sequences - rather than requiring manual curation. This makes conversion a measurable recommendation objective: it becomes testable whether users shown paywall-adjacent content convert at a higher rate than those shown standard recommendations. [Recommendations](https://www.recombee.com/faq/recommendations) Haven’t Found The Information You Are Looking For? ## Recommended Topics for You --- # A lot of our users aren't logged in. Do recommendations for them just default to "most popular" and stop there? | FAQ > Source: https://www.recombee.com/faq/how-do-you-handle-recommendations-for-anonymous-users > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # [FAQ](https://www.recombee.com/faq) [All->](https://www.recombee.com/faq) [How It Works->](https://www.recombee.com/faq/how-it-works) [Recommendations->](https://www.recombee.com/faq/recommendations) [Personalized Search->](https://www.recombee.com/faq/personalized-search) [Features->](https://www.recombee.com/faq/features) [FAQ](https://www.recombee.com/faq) [<- Back to List](https://www.recombee.com/faq) # A lot of our users aren't logged in. Do recommendations for them just default to "most popular" and stop there? No, anonymous users receive session-based recommendations, not just popularity lists. Multi-armed bandit algorithms can personalize within a session using signals a user generates during that visit alone, without any historical profile. The critical infrastructure decision is what happens at login: the system should merge the anonymous session history with the authenticated user's existing profile rather than discarding it. On platforms with high anonymous traffic - up to 70% in ad-supported models - handling this merge correctly has a direct impact on recommendation quality at scale. [How It Works](https://www.recombee.com/faq/how-it-works) Haven’t Found The Information You Are Looking For? ## Recommended Topics for You --- # When I hear "targeted ads", I assume the recommender is behind it. Is that actually how it works? | FAQ > Source: https://www.recombee.com/faq/how-do-you-handle-targeted-ads > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # [FAQ](https://www.recombee.com/faq) [All->](https://www.recombee.com/faq) [How It Works->](https://www.recombee.com/faq/how-it-works) [Recommendations->](https://www.recombee.com/faq/recommendations) [Personalized Search->](https://www.recombee.com/faq/personalized-search) [Features->](https://www.recombee.com/faq/features) [FAQ](https://www.recombee.com/faq) [<- Back to List](https://www.recombee.com/faq) # When I hear "targeted ads", I assume the recommender is behind it. Is that actually how it works? No, targeted advertising and recommender systems are distinct systems with different data inputs. Recommender systems work only with anonymized interaction data from within the platform itself, without accessing cross-site behavioral profiles or third-party user attributes that ad-targeting platforms rely on. A recommender's personalization is based entirely on what a user has done on the same website. This distinction matters for data architecture and privacy compliance: the data scope of a recommendation system is narrower and more contained than that of an ad-targeting stack. [How It Works](https://www.recombee.com/faq/how-it-works) Haven’t Found The Information You Are Looking For? ## Recommended Topics for You --- # In what recommendation scenarios does a user's geographic location become a critical input? | FAQ > Source: https://www.recombee.com/faq/how-does-geographic-location-affect-recommendation-quality > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # [FAQ](https://www.recombee.com/faq) [All->](https://www.recombee.com/faq) [How It Works->](https://www.recombee.com/faq/how-it-works) [Recommendations->](https://www.recombee.com/faq/recommendations) [Personalized Search->](https://www.recombee.com/faq/personalized-search) [Features->](https://www.recombee.com/faq/features) [FAQ](https://www.recombee.com/faq) [<- Back to List](https://www.recombee.com/faq) # In what recommendation scenarios does a user's geographic location become a critical input? Geographic location is critical in scenarios where users are interested in items physically tied to a place, such as real estate listings, job postings, or local events. **Crucially, location data enables relevant recommendations even for users with no interaction history**, because the system can fall back on items that are popular within the user's region. This makes location one of the most actionable cold-start signals available. [Recommendations](https://www.recombee.com/faq/recommendations) Haven’t Found The Information You Are Looking For? ## Recommended Topics for You --- # How does incorporating user background attributes such as skills or interests affect recommendation quality in domains with sparse interaction data? | FAQ > Source: https://www.recombee.com/faq/how-does-incorporating-user-background-attributes-affect-recommendation-quality > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # [FAQ](https://www.recombee.com/faq) [All->](https://www.recombee.com/faq) [How It Works->](https://www.recombee.com/faq/how-it-works) [Recommendations->](https://www.recombee.com/faq/recommendations) [Personalized Search->](https://www.recombee.com/faq/personalized-search) [Features->](https://www.recombee.com/faq/features) [FAQ](https://www.recombee.com/faq) [<- Back to List](https://www.recombee.com/faq) # How does incorporating user background attributes such as skills or interests affect recommendation quality in domains with sparse interaction data? In domains where users interact infrequently - such as job platforms or professional networks - interaction history alone is insufficient to build a reliable preference model. **Background attributes like skills, interests, or professional bio provide a non-behavioral signal that compensates for sparse interaction data**, enabling relevant recommendations from a user's first session. This is especially consequential for new user retention: a platform that can deliver relevant results before a user has clicked or purchased is more likely to establish a habit. Organizations in low-frequency domains should prioritize collecting structured background attributes during onboarding rather than relying solely on behavioral data accumulation. [How It Works](https://www.recombee.com/faq/how-it-works) Haven’t Found The Information You Are Looking For? ## Recommended Topics for You --- # What strategic risk does neglecting cold-start item coverage create for a product catalog-driven business? | FAQ > Source: https://www.recombee.com/faq/how-does-neglecting-cold-start-coverage-affect-business-outcomes > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # [FAQ](https://www.recombee.com/faq) [All->](https://www.recombee.com/faq) [How It Works->](https://www.recombee.com/faq/how-it-works) [Recommendations->](https://www.recombee.com/faq/recommendations) [Personalized Search->](https://www.recombee.com/faq/personalized-search) [Features->](https://www.recombee.com/faq/features) [FAQ](https://www.recombee.com/faq) [<- Back to List](https://www.recombee.com/faq) # What strategic risk does neglecting cold-start item coverage create for a product catalog-driven business? Neglecting cold-start coverage means newly added items receive no recommendation exposure until they accumulate interactions, creating a self-reinforcing cycle where only established items get traffic. **Image and text neural embeddings directly mitigate this risk by providing similarity signals independent of interaction history.** For businesses with rapidly rotating catalogs - such as online marketplaces and deal aggregators - this gap translates directly into lost revenue on new inventory. [How It Works](https://www.recombee.com/faq/how-it-works) Haven’t Found The Information You Are Looking For? ## Recommended Topics for You --- # We're launching new products every week. How quickly can they actually get recommended to the right users if they have no clicks yet? | FAQ > Source: https://www.recombee.com/faq/how-quickly-can-new-products-get-recommended > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # [FAQ](https://www.recombee.com/faq) [All->](https://www.recombee.com/faq) [How It Works->](https://www.recombee.com/faq/how-it-works) [Recommendations->](https://www.recombee.com/faq/recommendations) [Personalized Search->](https://www.recombee.com/faq/personalized-search) [Features->](https://www.recombee.com/faq/features) [FAQ](https://www.recombee.com/faq) [<- Back to List](https://www.recombee.com/faq) # We're launching new products every week. How quickly can they actually get recommended to the right users if they have no clicks yet? New items don't have to wait for their first clicks before entering recommendations. Modern recommender systems generate neural embeddings from item text and images, which allow a new product to be positioned relative to existing catalog items based on content similarity alone. A product with a complete description and quality images can be surfaced to relevant users from the moment it is added. The practical implication is that metadata quality and completeness at upload time directly determines how quickly new products reach the right audience. [Features](https://www.recombee.com/faq/features) Haven’t Found The Information You Are Looking For? ## Recommended Topics for You --- # How should a business weigh GDPR and data privacy requirements against the need for rich user data to drive personalization? | FAQ > Source: https://www.recombee.com/faq/how-should-a-business-weigh-gdpr-and-data-privacy-requirements > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # [FAQ](https://www.recombee.com/faq) [All->](https://www.recombee.com/faq) [How It Works->](https://www.recombee.com/faq/how-it-works) [Recommendations->](https://www.recombee.com/faq/recommendations) [Personalized Search->](https://www.recombee.com/faq/personalized-search) [Features->](https://www.recombee.com/faq/features) [FAQ](https://www.recombee.com/faq) [<- Back to List](https://www.recombee.com/faq) # How should a business weigh GDPR and data privacy requirements against the need for rich user data to drive personalization? Regulatory frameworks like GDPR can be treated as a structural advantage rather than a constraint. **Data minimization strategies and pseudonymization allow recommender systems to deliver highly personalized experiences without storing unnecessary personal data**, which simultaneously reduces compliance risk and builds user trust. A system designed with privacy by default is more defensible to regulators and more credible to users. For executives, this reframes privacy investment not as a cost center but as a mechanism for sustainable personalization - one that reduces the legal and reputational exposure that comes with over-collecting user data. [How It Works](https://www.recombee.com/faq/how-it-works) Haven’t Found The Information You Are Looking For? ## Recommended Topics for You --- # Is the growth in recommender system adoption a trend we need to take seriously, or is it already plateaued? | FAQ > Source: https://www.recombee.com/faq/is-the-growth-in-recommender-system-adoption-a-trend-we-need-to-take-seriously > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # [FAQ](https://www.recombee.com/faq) [All->](https://www.recombee.com/faq) [How It Works->](https://www.recombee.com/faq/how-it-works) [Recommendations->](https://www.recombee.com/faq/recommendations) [Personalized Search->](https://www.recombee.com/faq/personalized-search) [Features->](https://www.recombee.com/faq/features) [FAQ](https://www.recombee.com/faq) [<- Back to List](https://www.recombee.com/faq) # Is the growth in recommender system adoption a trend we need to take seriously, or is it already plateaued? Recommender system usage has been growing consistently and shows no signs of slowing. The number of recommendations served to an average active online user has grown exponentially over the past decade, with no indication the rate is approaching saturation. For teams evaluating investment in recommendation infrastructure, this trajectory suggests that personalization is becoming standard across a widening range of digital products rather than remaining a capability limited to large platforms. [How It Works](https://www.recombee.com/faq/how-it-works) Haven’t Found The Information You Are Looking For? ## Recommended Topics for You --- # What types of data sources does a modern recommender system rely on to generate personalized recommendations? | FAQ > Source: https://www.recombee.com/faq/what-data-sources-does-a-modern-recommender-system-rely-on > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # [FAQ](https://www.recombee.com/faq) [All->](https://www.recombee.com/faq) [How It Works->](https://www.recombee.com/faq/how-it-works) [Recommendations->](https://www.recombee.com/faq/recommendations) [Personalized Search->](https://www.recombee.com/faq/personalized-search) [Features->](https://www.recombee.com/faq/features) [FAQ](https://www.recombee.com/faq) [<- Back to List](https://www.recombee.com/faq) # What types of data sources does a modern recommender system rely on to generate personalized recommendations? Modern recommender systems draw on three primary data categories: an item catalog, a user catalog, and a history of user-item interactions. **The item catalog stores both active and historical items**, which matters because historical items help measure similarity between users who interacted with them in the past. The user catalog holds (often optional) attributes such as location, subscription status, and user bio. Together, these sources allow a recommender to build accurate, personalized outputs even before a user has accumulated a long interaction history. Teams integrating a recommender should plan data pipelines for all three categories from the start. [How It Works](https://www.recombee.com/faq/how-it-works) Haven’t Found The Information You Are Looking For? ## Recommended Topics for You --- # Beyond clicks, what interaction data should we actually be sending in to get meaningfully better recommendations? | FAQ > Source: https://www.recombee.com/faq/what-interaction-data-should-we-send > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # [FAQ](https://www.recombee.com/faq) [All->](https://www.recombee.com/faq) [How It Works->](https://www.recombee.com/faq/how-it-works) [Recommendations->](https://www.recombee.com/faq/recommendations) [Personalized Search->](https://www.recombee.com/faq/personalized-search) [Features->](https://www.recombee.com/faq/features) [FAQ](https://www.recombee.com/faq) [<- Back to List](https://www.recombee.com/faq) # Beyond clicks, what interaction data should we actually be sending in to get meaningfully better recommendations? Click data alone is a weak signal - partial consumption data is significantly more informative. Tracking what fraction of content a user actually consumed - the percentage of a video watched, the portion of an article read, the segment of a song played - gives the system a direct measure of satisfaction rather than intent alone. A click followed by immediate exit carries very different meaning from a click followed by full consumption. Sending engagement depth signals allows the recommender to distinguish genuine interest from accidental or disappointed clicks. [How It Works](https://www.recombee.com/faq/how-it-works) Haven’t Found The Information You Are Looking For? ## Recommended Topics for You --- # What is driving the continued growth in the volume of recommendations served to online users? | FAQ > Source: https://www.recombee.com/faq/what-is-driving-the-continued-growth-in-the-volume-of-recommendations > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # [FAQ](https://www.recombee.com/faq) [All->](https://www.recombee.com/faq) [How It Works->](https://www.recombee.com/faq/how-it-works) [Recommendations->](https://www.recombee.com/faq/recommendations) [Personalized Search->](https://www.recombee.com/faq/personalized-search) [Features->](https://www.recombee.com/faq/features) [FAQ](https://www.recombee.com/faq) [<- Back to List](https://www.recombee.com/faq) # What is driving the continued growth in the volume of recommendations served to online users? The growth in recommendation volume is driven by three compounding factors: **more internet users globally, more time each person spends online, and a rising number of websites and services adopting recommender systems.** The article notes this acceleration has been exponential over the last decade and is far from saturation. For organizations evaluating whether to invest in recommendation infrastructure now or later, the data suggests the competitive baseline is rising continuously - waiting means catching up against platforms that are already benefiting from this compounding growth. [How It Works](https://www.recombee.com/faq/how-it-works) Haven’t Found The Information You Are Looking For? ## Recommended Topics for You --- # What is the strategic significance of recommendation systems becoming pervasive across virtually every major online platform? | FAQ > Source: https://www.recombee.com/faq/what-is-the-strategic-significance-of-recommendation-systems-becoming-pervasive > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # [FAQ](https://www.recombee.com/faq) [All->](https://www.recombee.com/faq) [How It Works->](https://www.recombee.com/faq/how-it-works) [Recommendations->](https://www.recombee.com/faq/recommendations) [Personalized Search->](https://www.recombee.com/faq/personalized-search) [Features->](https://www.recombee.com/faq/features) [FAQ](https://www.recombee.com/faq) [<- Back to List](https://www.recombee.com/faq) # What is the strategic significance of recommendation systems becoming pervasive across virtually every major online platform? Recommender systems have become the most influential machine learning technology in consumer-facing products, with the average active online user receiving hundreds of recommendations daily across news, music, video, e-commerce, and social media. **The strategic implication is that recommendation capability is no longer a differentiator for large platforms - it is a baseline expectation.** For executives evaluating whether to invest in AI-driven personalization, the relevant benchmark is not whether competitors are using recommenders, but how well-tuned those systems are. The growth trajectory described suggests that platforms without effective recommendation infrastructure will face increasing disadvantage as user expectations continue to rise. [How It Works](https://www.recombee.com/faq/how-it-works) Haven’t Found The Information You Are Looking For? ## Recommended Topics for You --- # What is the strategic value of investing in image-based neural embeddings for a marketplace where sellers upload their own product photos? | FAQ > Source: https://www.recombee.com/faq/what-is-the-strategic-value-of-investing-in-image-based-neural-embeddings > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # [FAQ](https://www.recombee.com/faq) [All->](https://www.recombee.com/faq) [How It Works->](https://www.recombee.com/faq/how-it-works) [Recommendations->](https://www.recombee.com/faq/recommendations) [Personalized Search->](https://www.recombee.com/faq/personalized-search) [Features->](https://www.recombee.com/faq/features) [FAQ](https://www.recombee.com/faq) [<- Back to List](https://www.recombee.com/faq) # What is the strategic value of investing in image-based neural embeddings for a marketplace where sellers upload their own product photos? In user-generated marketplaces, sellers are unlikely to provide structured text descriptions, making image embeddings the primary available signal for item similarity. **Visual neural embeddings enable the recommender to identify related items and surface relevant alternatives even when textual metadata is sparse or absent.** This directly supports recommendation coverage across the full catalog rather than only well-described listings. For a marketplace operator, this translates into broader monetizable surface area - more items receiving recommendation-driven impressions - without requiring sellers to change their listing behavior. The business case is strongest where catalog quality is variable and text-based signals are unreliable. [Recommendations](https://www.recombee.com/faq/recommendations) Haven’t Found The Information You Are Looking For? ## Recommended Topics for You --- # What makes recommendation objective design particularly complex for platforms like job boards or dating sites? | FAQ > Source: https://www.recombee.com/faq/what-makes-recommendation-design-complex-for-job-boards-dating-sites > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # [FAQ](https://www.recombee.com/faq) [All->](https://www.recombee.com/faq) [How It Works->](https://www.recombee.com/faq/how-it-works) [Recommendations->](https://www.recombee.com/faq/recommendations) [Personalized Search->](https://www.recombee.com/faq/personalized-search) [Features->](https://www.recombee.com/faq/features) [FAQ](https://www.recombee.com/faq) [<- Back to List](https://www.recombee.com/faq) # What makes recommendation objective design particularly complex for platforms like job boards or dating sites? Job boards and dating sites face recommendation objectives that are more complex than single-sided platforms because they must optimize for the satisfaction of multiple parties simultaneously under constraints. **Unlike e-commerce or streaming, success requires a match between two parties**, meaning a recommendation that satisfies one side may not satisfy the other. This multi-stakeholder optimization problem is structurally distinct from maximizing a single metric like watch time or conversion rate. Any platform operating a two-sided or multi-party marketplace should account for this asymmetry when defining success metrics and evaluating recommendation quality. [How It Works](https://www.recombee.com/faq/how-it-works) Haven’t Found The Information You Are Looking For? ## Recommended Topics for You --- # What was the significance of the GroupLens system in the history of personalized recommendations? | FAQ > Source: https://www.recombee.com/faq/what-was-the-significance-of-the-group-lens-system > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # [FAQ](https://www.recombee.com/faq) [All->](https://www.recombee.com/faq) [How It Works->](https://www.recombee.com/faq/how-it-works) [Recommendations->](https://www.recombee.com/faq/recommendations) [Personalized Search->](https://www.recombee.com/faq/personalized-search) [Features->](https://www.recombee.com/faq/features) [FAQ](https://www.recombee.com/faq) [<- Back to List](https://www.recombee.com/faq) # What was the significance of the GroupLens system in the history of personalized recommendations? GroupLens, introduced in 1992, was one of the first systems to base recommendations exclusively on user historical interactions - specifically, explicit ratings of news articles. **Its significance is that it established the foundational principle of collaborative filtering: using recorded user behavior rather than item attributes alone to drive recommendations.** This approach marked the transition from generic IR outputs to genuinely personalized results. For product teams evaluating recommender infrastructure today, understanding this origin clarifies why behavioral data collection - interaction history, explicit ratings, implicit signals - remains central to recommendation quality. [How It Works](https://www.recombee.com/faq/how-it-works) Haven’t Found The Information You Are Looking For? ## Recommended Topics for You --- # Why are historical items that are no longer available to users still stored in the item catalog? | FAQ > Source: https://www.recombee.com/faq/why-are-historical-items-still-stored-in-the-item-catalog > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # [FAQ](https://www.recombee.com/faq) [All->](https://www.recombee.com/faq) [How It Works->](https://www.recombee.com/faq/how-it-works) [Recommendations->](https://www.recombee.com/faq/recommendations) [Personalized Search->](https://www.recombee.com/faq/personalized-search) [Features->](https://www.recombee.com/faq/features) [FAQ](https://www.recombee.com/faq) [<- Back to List](https://www.recombee.com/faq) # Why are historical items that are no longer available to users still stored in the item catalog? Historical items are retained in the catalog because **they are essential for measuring similarity between users who interacted with those items in the past**. Even if an item is no longer active or purchasable, the interaction signal it generated remains a meaningful data point for understanding user preferences and computing user-to-user or item-to-item similarity. Removing historical items from the catalog would degrade the quality of those similarity calculations. Organizations should therefore treat catalog management as a long-term data asset strategy rather than a simple housekeeping task. [How It Works](https://www.recombee.com/faq/how-it-works) Haven’t Found The Information You Are Looking For? ## Recommended Topics for You --- # Why is content discovery a standalone recommendation objective rather than a byproduct of relevance optimization? | FAQ > Source: https://www.recombee.com/faq/why-is-content-discovery-a-standalone-recommendation-objective > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # [FAQ](https://www.recombee.com/faq) [All->](https://www.recombee.com/faq) [How It Works->](https://www.recombee.com/faq/how-it-works) [Recommendations->](https://www.recombee.com/faq/recommendations) [Personalized Search->](https://www.recombee.com/faq/personalized-search) [Features->](https://www.recombee.com/faq/features) [FAQ](https://www.recombee.com/faq) [<- Back to List](https://www.recombee.com/faq) # Why is content discovery a standalone recommendation objective rather than a byproduct of relevance optimization? Content discovery is treated as an independent objective because relevance optimization alone tends to surface familiar or already-popular content, which does not necessarily help users find new items they would enjoy. **Accelerating content discovery specifically targets catalog breadth**, helping users encounter content beyond their established preferences or the platform's most-trafficked titles. This distinction matters because a system optimizing purely for predicted relevance can inadvertently narrow the user's exposure over time. Platforms should track content discovery metrics separately from relevance scores to ensure their recommendation system is genuinely expanding user awareness of available catalog rather than reinforcing existing patterns. [Personalized Search](https://www.recombee.com/faq/personalized-search) Haven’t Found The Information You Are Looking For? ## Recommended Topics for You --- # Why Recombee Is the Best Alternative to Amazon Personalize > Source: https://www.recombee.com/vs/amazon-personalize > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. Recombee vs. Amazon Personalize # Why Recombee Is the Best Alternative to Amazon Personalize Recombee is a flexible solution for companies that want to have full control over their recommendations and deliver personalized user experience real time. Discover why Recombee is the go-to recommendation solution for over 1000 companies around the world. ![](https://www.recombee.com/img/vs/amazon-personalize.png) ![](https://www.recombee.com/img/customers-domains/9gag.svg)![](https://www.recombee.com/img/homepage/showmax.svg)![](https://www.recombee.com/img/customers-domains/audiomack.svg)![](https://www.recombee.com/img/customers-domains/zumper.svg)![](https://www.recombee.com/img/customers-domains/tripadvisor.svg) ## 5 Reasons to Choose Recombee Over Amazon Personalize ### 1 Industry-Leading Model Training #### Recombee * Incrementally trained in **real-time** * **Effective** with newly added content * **Ideal for fast-changing** sectors #### Amazon Personalize * Requires retraining **every 7 days** * **Less responsive** to new content * **Slower to adapt** to dynamic environments ### 2 Advanced Customization of Recommendations #### Recombee * **Advanced** filters and fine-tuning options for unique needs * Out-of-the-box tailored solutions for a **variety of industries** * **Supports flexible boosters** for promoting specific content, brands, affiliate partners, higher-margin items, categories, and more #### Amazon Personalize * **Restricted** customization and logic variations * **Limited** domain-specific solutions * **Lacks booster** support ### 3 Broad Recommendation Scope #### Recombee * Recognizes and recommends not only specific movies or products but also categories, brands, or genres * Provides a fully personalized homepage, enabling you to tailor both recommendation rows and their order for each user #### Amazon Personalize * Limited to basic title recommendations, focusing on individual items like movies or products * Homepage personalization lacks advanced features, offering less control over content and order customization ### 4 User-Friendly Interface with Built-in Analytics #### Recombee * Offers **built-in analytics** for seamless tracking of product strategy * **Intuitive interface** designed for easy navigation * **Faster time to insight** with integrated tools, reducing reliance on external platforms #### Amazon Personalize * **Requires external tools** like Amazon CloudWatch for tracking metrics * User interface can be **more complex and less intuitive for non-technical users** * **Increased time to insight** due to reliance on multiple external tools ### 5 Seamless Integration #### Recombee * Offers **10 diverse SDKs** for various platforms * **Flexible integration** with your existing infrastructure #### Amazon Personalize * Provides **only 3 SDKs** * **Limited flexibility for specialized tech stacks**, potentially leading to integration challenges with non-standard platforms ## Don’t just take our word for it... ![Logo - DAZN](https://www.recombee.com/img/testimonials/dazn-logo.svg) [Explore Success Stories](https://www.recombee.com/case-studies) ![Christoph Haas, DAZN](https://www.recombee.com/img/testimonials/dazn.png) Christoph Haas EVP Product & Platform Engineering at DAZN At DAZN, being the Global Home of Sports means delivering the right matches, highlights, and moments to audiences in 200+ markets - bringing fans even closer to the live game. That’s why we’ve teamed up with Recombee to personalize experiences at scale. Their tech enables us to connect each viewer on any device with the right game or clip in real time through flexible solution built for growth. This partnership sets the pace for a smarter, more connected global sports experience. ![Logo - The Telegraph](https://www.recombee.com/img/testimonials/the-telegraph-logo.svg) [Read Case Study](https://www.recombee.com/case-studies/the-telegraph) ![Tom Kelleher, The Telegraph](https://www.recombee.com/img/testimonials/the-telegraph.png) Tom Kelleher Director of Emerging Technology – AI & Personalisation at The Telegraph We use Recombee to power our AI personalization & Search at The Telegraph. It immediately proved its value, securing a 35% CTR uplift in an A/B test against a competing solution while simultaneously enhancing our editors' ability to manage and analyse content performance. Beyond the numbers, the collaboration with the Recombee team has been excellent, they helped us push our thinking and we were delighted to jointly present at the RecSys conference and be recognized as an INMA '26 finalist for "Best Use of Generative AI". ![Logo - 9GAG](https://www.recombee.com/img/testimonials/9gag-logo.png) [Read Case Study](https://www.recombee.com/case-studies/9gag) ![Kristie Chen, 9GAG](https://www.recombee.com/img/testimonials/9gag.png) Kristie Chen Product Head at 9GAG Thanks to the Recombee solution, we managed to increase, among other KPIs, post views by 37%, overall interactions by 22%, and users’ session duration by 7.5%. With Recombee, we've taken engagement with user-generated content to the next level. ![Logo - Audiomack](https://www.recombee.com/img/testimonials/audiomack-logo.svg) [Read Case Study](https://www.recombee.com/case-studies/audiomack) ![Christopher Dalla Riva, Audiomack](https://www.recombee.com/img/testimonials/audiomack.png) Christopher Dalla Riva Senior Product Manager at Audiomack Thanks to the Recombee recommender engine, our monthly plays increased by 206% and weekly follows by 67%. Because the recommendations performed so well, we moved them from our Search page to the top of our main Discover tab. They are now the best-performing module within that tab, accounting for 46% of all plays. ![Logo - Crexi](https://www.recombee.com/img/testimonials/crexi-logo.svg) [Read Case Study](https://www.recombee.com/case-studies/crexi) ![Larkin Magner, Crexi](https://www.recombee.com/img/testimonials/crexi.png) Larkin Magner Director of Product Management at Crexi Our customers now receive highly relevant property recommendations that cater to their specific needs, with one notable email campaign seeing a 178% uplift in CTOR. Recombee commitment to excellence is evident in the 40% increase in listing engagements on our platform, contributing to our growth in the competitive real estate market. ![Logo - Slickdeals](https://www.recombee.com/img/testimonials/slickdeals-logo.svg) [Read Case Study](https://www.recombee.com/case-studies/slickdeals) ![Daniel Uhm, Slickdeals](https://www.recombee.com/img/testimonials/slickdeals.png) Daniel Uhm Product Manager at Slickdeals Placing recommendations on our homepage was a huge success — 70%+ higher product detail page views and 30%+ higher clickthroughs. The Recombee team is a great partner in helping solve our unique use cases, and we look forward to continue working with them. ![Logo - Pepper](https://www.recombee.com/img/testimonials/pepper-logo.svg) [Read Case Study](https://www.recombee.com/case-studies/pepper) ![Heike Guertler, Pepper](https://www.recombee.com/img/testimonials/pepper.png) Heike Guertler Head Of Product at Pepper Recombee problem-solving skills have proven invaluable, helping us overcome various business challenges while allowing us to consistently increase our click-outs and deliver a better user experience. Thanks to their solution we've seen our click-outs increase by up to 21%. ## The Recombee Personalization Powerhouse With Recombee’s all-in-one recommendations solution, you can take your business to new heights. Whether you need content recommendations, product recommendations, or personalized search, we can do it all. Integrate within minutes, customize with ease in our intuitive Admin UI, and delve into our Insights Analytics section where you can monitor your recommendation performance in real-time. ![Recombee Recommendations](https://www.recombee.com/img/homepage/wwd-recommendations.png) ### Recommend Recombee offers real-time personalized recommendations across all platforms: website, app, or email, ensuring each user experience is engaging and perfectly tailored. By leveraging AI to understand individual preferences and behaviors, Recombee turns every interaction into a customized journey, enhancing satisfaction and boosting conversions across the digital landscape. ![Recombee Search](https://www.recombee.com/img/homepage/wwd-search.png) ### Search Our robust system supports multilingual, typo-tolerant, and fully personalized searches, covering everything from titles to categories to ensure users always find what they seek. When two users type the same query, they won’t necessarily see the same results. For instance, after typing 'Ha,' User A might see 'Hannibal' first if they watch thrillers, while User B might see 'Harry Potter' if they prefer fantasy. Our versatile search also allows you to filter content for specific target audiences, such as those with different subscription plans. ![Recombee Insights](https://www.recombee.com/img/homepage/wwd-insights.png) ### Analyze Insights is a transformative analytics feature within Recombee’s Admin UI, designed to provide advanced, real-time insights into user interactions and recommendation performance. It offers customizable visualizations and reports, enabling targeted analysis and strategic decision-making to enhance user engagement and meet unique KPIs. --- # A Step-By-Step Guide to Integrate Recombee Recommendation Engine > Source: https://www.recombee.com/handbook/download-a-step-by-step-guide-to-the-fastest-integration-eyw34h56k2hu56sh > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. **Handbook** # A Step-By-Step Guide to the Fastest Integration This guide shows the simplest form of integration using a No-Code widget and minimum coding involved. For integration through API, please [see our documentation.](https://docs.recombee.com/) [Explore Case Studies](https://www.recombee.com/case-studies) ## Handbooks & Booklets [![](https://www.recombee.com/img/booklets/modern-recommender-systems-part-1.png)Modern Recommender Systems](https://www.recombee.com/handbook/modern-recommender-systems) [![](https://www.recombee.com/img/handbooks/content-recommendations.png)Scenario Setup Guide for Content Recommendations](https://www.recombee.com/handbook/download-content-recommendations-35kl362g0946239g0er) [![](https://www.recombee.com/img/handbooks/product-recommendations.png)Scenario Setup Guide for Product Recommendations](https://www.recombee.com/handbook/download-product-recommendations-dfh09362my2ld4kl32l) --- # Guide: Advanced Media Content Setup for Recombee Recommendation Engine > Source: https://www.recombee.com/handbook/download-content-recommendations-35kl362g0946239g0er > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. **Handbook** # Scenario Setup Guide for Content Recommendations This guide names the most popular scenarios, there are multiple advanced scenarios available - contact [support@recombee.com](mailto:support@recombee.com) for more info. [Explore Case Studies](https://www.recombee.com/case-studies) ## Handbooks & Booklets [![](https://www.recombee.com/img/booklets/modern-recommender-systems-part-1.png)Modern Recommender Systems](https://www.recombee.com/handbook/modern-recommender-systems) [![](https://www.recombee.com/img/handbooks/a-step-by-step-guide-to-the-fastest-integration.png)A Step-By-Step Guide to the Fastest Integration](https://www.recombee.com/handbook/download-a-step-by-step-guide-to-the-fastest-integration-eyw34h56k2hu56sh) [![](https://www.recombee.com/img/handbooks/product-recommendations.png)Scenario Setup Guide for Product Recommendations](https://www.recombee.com/handbook/download-product-recommendations-dfh09362my2ld4kl32l) --- # Guide: Advanced Product Personalization Engine Recombee > Source: https://www.recombee.com/handbook/download-product-recommendations-dfh09362my2ld4kl32l > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. **Handbook** # Scenario Setup Guide for Product Recommendations This guide names the most popular scenarios, there are multiple advanced scenarios available - contact [support@recombee.com](mailto:support@recombee.com) for more info. [Explore Case Studies](https://www.recombee.com/case-studies) ## Handbooks & Booklets [![](https://www.recombee.com/img/booklets/modern-recommender-systems-part-1.png)Modern Recommender Systems](https://www.recombee.com/handbook/modern-recommender-systems) [![](https://www.recombee.com/img/handbooks/a-step-by-step-guide-to-the-fastest-integration.png)A Step-By-Step Guide to the Fastest Integration](https://www.recombee.com/handbook/download-a-step-by-step-guide-to-the-fastest-integration-eyw34h56k2hu56sh) [![](https://www.recombee.com/img/handbooks/content-recommendations.png)Scenario Setup Guide for Content Recommendations](https://www.recombee.com/handbook/download-content-recommendations-35kl362g0946239g0er) --- # Modern Recommender Systems > Source: https://www.recombee.com/handbook/modern-recommender-systems > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. **Booklet** # Modern Recommender Systems Recommender systems have become key to user experience and business growth. Over the years, Recombee has worked with many companies, learning the ups and downs of using this tech at scale. This booklet shares hands-on advice for experts, product leads, and tech decision-makers. [Read on the Blog](#read-on-the-blog) ## Handbooks & Booklets [![](https://www.recombee.com/img/handbooks/a-step-by-step-guide-to-the-fastest-integration.png)A Step-By-Step Guide to the Fastest Integration](https://www.recombee.com/handbook/download-a-step-by-step-guide-to-the-fastest-integration-eyw34h56k2hu56sh) [![](https://www.recombee.com/img/handbooks/content-recommendations.png)Scenario Setup Guide for Content Recommendations](https://www.recombee.com/handbook/download-content-recommendations-35kl362g0946239g0er) [![](https://www.recombee.com/img/handbooks/product-recommendations.png)Scenario Setup Guide for Product Recommendations](https://www.recombee.com/handbook/download-product-recommendations-dfh09362my2ld4kl32l) ## Read on the Blog [![Modern Recommender Systems - Part 1: Introduction](https://www.recombee.com/img/blog/modern-recommender-systems-part-1-introduction.png)](https://www.recombee.com/blog/modern-recommender-systems-part-1-introduction) ### [Modern Recommender Systems - Part 1: Introduction](https://www.recombee.com/blog/modern-recommender-systems-part-1-introduction) Pavel Kordik, Apr 17, 2023 [![Modern Recommender Systems - Part 2: Data](https://www.recombee.com/img/blog/modern-recommender-systems-part-2-data.png)](https://www.recombee.com/blog/modern-recommender-systems-part-2-data) ### [Modern Recommender Systems - Part 2: Data](https://www.recombee.com/blog/modern-recommender-systems-part-2-data) Pavel Kordik, Mar 07, 2024 [![Modern Recommender Systems - Part 3: Objectives](https://www.recombee.com/img/blog/modern-recommender-systems-part-3-objectives.png)](https://www.recombee.com/blog/modern-recommender-systems-part-3-objectives) ### [Modern Recommender Systems - Part 3: Objectives](https://www.recombee.com/blog/modern-recommender-systems-part-3-objectives) Pavel Kordik, Sep 03, 2025 --- # Research > Source: https://www.recombee.com/research > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. Community Contributions # Advancing AI Through Research & Community ![](https://www.recombee.com/img/research/main.png) At Recombee, we believe in open innovation and collaborative advancement of AI technologies. Through RecombeeLab, our research division, we actively contribute to the scientific community by publishing state-of-the-art research, supporting PhD students, and openly sharing our methods. Our partnerships with top universities and our commitment to open-source development reflect our mission to democratize advanced AI technologies. ## Research Excellence Through RecombeeLab, our joint research laboratory with the Faculty of Information Technology at the Czech Technical University in Prague, we're pushing the boundaries of recommendation systems and machine learning. We provide financial support to PhD students and actively collaborate on groundbreaking research projects. ### Publications [See all ->](https://www.recombee.com/research-publications) [![Cover - Segment-Aware Analytics for Real-Time Editorial Support in Media Groups: Lessons from The Telegraph](https://www.recombee.com/img/publications/segment-aware-analytics-for-real-time-editorial-support-in-media-groups-lessons-from-the-telegraph.png)](https://ceur-ws.org/Vol-4056/) ## [Segment-Aware Analytics for Real-Time Editorial Support in Media Groups: Lessons from The Telegraph](https://ceur-ws.org/Vol-4056/) Analytics for Real-Time Editorial Support: Lessons from The Telegraph. 13th International Workshop on News Recommendation and Analytics co-located with the 19th ACM Conference on Recommender Systems 2025 [![Cover - The Future is Sparse: Embedding Compression for Scalable Retrieval in Recommender Systems](https://www.recombee.com/img/publications/the-future-is-sparse-embedding-compression-for-scalable-retrieval-in-recommender-systems.png)](https://dl.acm.org/doi/full/10.1145/3705328.3748147) ## [The Future is Sparse: Embedding Compression for Scalable Retrieval in Recommender Systems](https://dl.acm.org/doi/full/10.1145/3705328.3748147) 90% Slimmer Production Embeddings. Proceedings of the 19th ACM Conference on Recommender Systems 2025 [![Cover - Recurrent Autoregressive Linear Model for Next-Basket Recommendation](https://www.recombee.com/img/publications/recurrent-autoregressive-linear-model-for-next-basket-recommendation.png)](https://dl.acm.org/doi/full/10.1145/3705328.3759313) ## [Recurrent Autoregressive Linear Model for Next-Basket Recommendation](https://dl.acm.org/doi/full/10.1145/3705328.3759313) Simplicity Wins: Linear Beats Deep in Next-Basket Recommendation. Proceedings of the 19th ACM Conference on Recommender Systems 2025 [![Cover - Evaluating Linear Shallow Autoencoders on Large Scale Datasets](https://www.recombee.com/img/publications/evaluating-linear-shallow-autoencoders-on-large-scale-datasets.png)](https://dl.acm.org/doi/10.1145/3748335) ## [Evaluating Linear Shallow Autoencoders on Large Scale Datasets](https://dl.acm.org/doi/10.1145/3748335) Scalable Recommendation in Industrial Scale. ACM Transactions on Recommender Systems 2025 ### Research Posters [See all ->](https://www.recombee.com/research-posters) [![Poster - Active Recommendation for Email Outreach Dynamics](https://www.recombee.com/img/research-posters/active-recommendation-for-email-outreach-dynamics.png)](https://www.recombee.com/img/research-posters/active-recommendation-for-email-outreach-dynamics.png) [![Poster - SAGEA: Sparse Autoencoder-based Group Embeddings Aggregation for Fairness-Preserving Group Recommendations](https://www.recombee.com/img/research-posters/sagea-sparse-autoencoder-based-group-embeddings-aggregation-for-fairness-preserving-group-recommendations.png)](https://www.recombee.com/img/research-posters/sagea-sparse-autoencoder-based-group-embeddings-aggregation-for-fairness-preserving-group-recommendations.png) [![Poster - Recurrent Autoregressive Linear Model for Next-Basket Recommendation](https://www.recombee.com/img/research-posters/recurrent-autoregressive-linear-model-for-next-basket-recommendation.png)](https://www.recombee.com/img/research-posters/recurrent-autoregressive-linear-model-for-next-basket-recommendation.png) ## Open Source Projects ![](https://www.recombee.com/img/research/elsa.png) ### ELSA Scalable linear shallow autoencoder for collaborative filtering. [Github](https://github.com/recombee/ELSA) ![](https://www.recombee.com/img/research/repsys.png) ### Repsys Open-source framework for building and evaluating recommendation systems. [Github](https://github.com/cowjen01/repsys) [Demo](https://repsys.recombee.net/) ![](https://www.recombee.com/img/research/beeformer.png) ### beeFormer Advanced transformer architecture optimized for recommendation tasks. [Github](https://github.com/recombee/beeformer) ![](https://www.recombee.com/img/research/compressae.png) ### CompresSAE Sparse compression of embeddings for scalable retrieval. [Github](https://github.com/recombee/CompresSAE) ## Community Engagement We contribute as journal reviewers and serve on the program committees of major conferences. ### Conference Support Proud sponsor and co-organizer of major industry conferences including RecSys, contributing to the global advancement of recommendation systems research. ### Non-Profit Collaboration Supporting organizations like prg.ai and aidetem.cz in their mission to enhance education through AI-assisted personalization. ## Making an Impact Our commitment to open innovation and community support helps advance the field of AI while making cutting-edge technology accessible to researchers and developers worldwide. ### Research Support Funding PhD research and academic collaborations. ### Open Source Sharing advanced AI tools with the community. ### Education Supporting AI-driven educational initiatives. ## Collaboration Would you like to collaborate with us to push the boundaries of recommender systems? Are you interested in doing an industrial master’s or PhD thesis? Contact us at [research@recombee.com](mailto:research@recombee.com) or check our [Research Opportunities](https://experts.ai/widgets/organizations/193179?opportunity=true) --- # Research Posters > Source: https://www.recombee.com/research-posters > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [Community Contributions](https://www.recombee.com/research) / Research Posters # Research Posters [![Poster - Active Recommendation for Email Outreach Dynamics](https://www.recombee.com/img/research-posters/active-recommendation-for-email-outreach-dynamics.png)](https://www.recombee.com/img/research-posters/active-recommendation-for-email-outreach-dynamics.png) [![Poster - SAGEA: Sparse Autoencoder-based Group Embeddings Aggregation for Fairness-Preserving Group Recommendations](https://www.recombee.com/img/research-posters/sagea-sparse-autoencoder-based-group-embeddings-aggregation-for-fairness-preserving-group-recommendations.png)](https://www.recombee.com/img/research-posters/sagea-sparse-autoencoder-based-group-embeddings-aggregation-for-fairness-preserving-group-recommendations.png) [![Poster - Recurrent Autoregressive Linear Model for Next-Basket Recommendation](https://www.recombee.com/img/research-posters/recurrent-autoregressive-linear-model-for-next-basket-recommendation.png)](https://www.recombee.com/img/research-posters/recurrent-autoregressive-linear-model-for-next-basket-recommendation.png) [![Poster - Probabilistic Modeling, Learnability and Uncertainty Estimation for Interaction Prediction in Movie Rating Datasets](https://www.recombee.com/img/research-posters/probabilistic-modeling-learnability-and-uncertainty-estimation-for-interaction-prediction-in-movie-rating-datasets.png)](https://www.recombee.com/img/research-posters/probabilistic-modeling-learnability-and-uncertainty-estimation-for-interaction-prediction-in-movie-rating-datasets.png) [![Poster - The Future is Sparse: Embedding Compression for Scalable Retrieval in Recommender Systems](https://www.recombee.com/img/research-posters/the-future-is-sparse-embedding-compression-for-scalable-retrieval-in-recommender-systems.png)](https://www.recombee.com/img/research-posters/the-future-is-sparse-embedding-compression-for-scalable-retrieval-in-recommender-systems.png) [![Poster - Segment-Aware Analytics for Real-Time Editorial Support in Media Groups: Lessons from The Telegraph](https://www.recombee.com/img/research-posters/segment-aware-analytics-for-real-time-editorial-support-in-media-groups-lessons-from-the-telegraph.png)](https://www.recombee.com/img/research-posters/segment-aware-analytics-for-real-time-editorial-support-in-media-groups-lessons-from-the-telegraph.png) --- # Research Publications > Source: https://www.recombee.com/research-publications > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [Community Contributions](https://www.recombee.com/research) / Publications # Publications [![Cover - Segment-Aware Analytics for Real-Time Editorial Support in Media Groups: Lessons from The Telegraph](https://www.recombee.com/img/publications/segment-aware-analytics-for-real-time-editorial-support-in-media-groups-lessons-from-the-telegraph.png)](https://ceur-ws.org/Vol-4056/) ## [Segment-Aware Analytics for Real-Time Editorial Support in Media Groups: Lessons from The Telegraph](https://ceur-ws.org/Vol-4056/) Analytics for Real-Time Editorial Support: Lessons from The Telegraph. 2025 [![Cover - SAGEA: Sparse Autoencoder-based Group Embeddings Aggregation for Fairness-Preserving Group Recommendations](https://www.recombee.com/img/publications/sagea-sparse-autoencoder-based-group-embeddings-aggregation-for-fairness-preserving-group-recommendations.png)](https://dl.acm.org/doi/10.1145/3705328.3759322) ## [SAGEA: Sparse Autoencoder-based Group Embeddings Aggregation for Fairness-Preserving Group Recommendations](https://dl.acm.org/doi/10.1145/3705328.3759322) Improving Group Recommendations with Sparse Autoencoders by Balancing Accuracy, Fairness, and Efficiency. 2025 [![Cover - Recurrent Autoregressive Linear Model for Next-Basket Recommendation](https://www.recombee.com/img/publications/recurrent-autoregressive-linear-model-for-next-basket-recommendation.png)](https://dl.acm.org/doi/full/10.1145/3705328.3759313) ## [Recurrent Autoregressive Linear Model for Next-Basket Recommendation](https://dl.acm.org/doi/full/10.1145/3705328.3759313) Simplicity Wins: Linear Beats Deep in Next-Basket Recommendation. 2025 [![Cover - The Future is Sparse: Embedding Compression for Scalable Retrieval in Recommender Systems](https://www.recombee.com/img/publications/the-future-is-sparse-embedding-compression-for-scalable-retrieval-in-recommender-systems.png)](https://dl.acm.org/doi/full/10.1145/3705328.3748147) ## [The Future is Sparse: Embedding Compression for Scalable Retrieval in Recommender Systems](https://dl.acm.org/doi/full/10.1145/3705328.3748147) 90% Slimmer Production Embeddings. 2025 [![Cover - Conv4Rec: A 1-by-1 Convolutional Autoencoder for User Profiling Through Joint Analysis of Implicit and Explicit Feedback](https://www.recombee.com/img/publications/conv4rec-a-1-by-1-convolutional-autoencoder-for-user-profiling-through-joint-analysis-of-implicit-and-explicit-feedback.png)](https://ieeexplore.ieee.org/abstract/document/11159277) ## [Conv4Rec: A 1-by-1 Convolutional Autoencoder for User Profiling Through Joint Analysis of Implicit and Explicit Feedback](https://ieeexplore.ieee.org/abstract/document/11159277) Jointly Learning Implicit and Explicit feedback. 2025 [![Cover - Evaluating Linear Shallow Autoencoders on Large Scale Datasets](https://www.recombee.com/img/publications/evaluating-linear-shallow-autoencoders-on-large-scale-datasets.png)](https://dl.acm.org/doi/10.1145/3748335) ## [Evaluating Linear Shallow Autoencoders on Large Scale Datasets](https://dl.acm.org/doi/10.1145/3748335) Scalable Recommendation in Industrial Scale. 2025 [![Cover - Probabilistic Modeling, Learnability and Uncertainty Estimation for Interaction Prediction in Movie Rating Datasets](https://www.recombee.com/img/publications/probabilistic-modeling-learnability-and-uncertainty-estimation-for-interaction-prediction-in-movie-rating-datasets.png)](https://dl.acm.org/doi/full/10.1145/3705328.3759332) ## [Probabilistic Modeling, Learnability and Uncertainty Estimation for Interaction Prediction in Movie Rating Datasets](https://dl.acm.org/doi/full/10.1145/3705328.3759332) Towards Accurate Uncertainty and Test-set Retrieval Performance Estimation. 2025 [![Cover - Mitigating Risks in Online Semantic Search](https://www.recombee.com/img/publications/mitigating-risks-in-online-semantic-search.png)](https://dl.acm.org/doi/10.1145/3699682.3728329) ## [Mitigating Risks in Online Semantic Search](https://dl.acm.org/doi/10.1145/3699682.3728329) Open Dataset for Harmful and Sensitive Query Alignment. 2025 [![Cover - Multitask Learning for Triplet Analysis](https://www.recombee.com/img/publications/multitask-learning-for-triplet-analysis.png)](https://www.sciencedirect.com/science/article/abs/pii/S0957417424030549) ## [Multitask Learning for Triplet Analysis](https://www.sciencedirect.com/science/article/abs/pii/S0957417424030549) Proposition of a multitask learning approach for the triple odd-one-out problem in cognitive sciences. 2025 [![Cover - beeFormer: transformer for recommender systems](https://www.recombee.com/img/publications/beeformer-transformer-for-recommender-systems.png)](https://doi.org/10.1145/3640457.3691707) ## [beeFormer: transformer for recommender systems](https://doi.org/10.1145/3640457.3691707) Improve recommendation of cold start items by training transformers on interactions. 2024 [![Cover - Advanced popularity models for curiosity detection](https://www.recombee.com/img/publications/advanced-popularity-models-for-curiosity-detection.png)](https://doi.org/10.1145/3589334.3645473) ## [Advanced popularity models for curiosity detection](https://doi.org/10.1145/3589334.3645473) Detecting and measuring popularity rates among loyal and curious audiences for online items. 2024 [![Cover - Enhancing local and regional recommendations](https://www.recombee.com/img/publications/enhancing-local-and-regional-recommendations.png)](https://doi.org/10.1145/3656641) ## [Enhancing local and regional recommendations](https://doi.org/10.1145/3656641) Enhance recommendations by aligning them more closely with local preferences and region-specific tastes. 2024 [![Cover - Constrained matrix completion](https://www.recombee.com/img/publications/constrained-matrix-completion.png)](https://proceedings.mlr.press/v235/ledent24a.html) ## [Constrained matrix completion](https://proceedings.mlr.press/v235/ledent24a.html) Enhanced matrix completion methods with new constraints, improving prediction accuracy and efficiency through theoretical analysis and practical experiments. 2024 [![Cover - Context aware recommendation](https://www.recombee.com/img/publications/context-aware-recommendation.png)](https://ieeexplore.ieee.org/document/10496217) ## [Context aware recommendation](https://ieeexplore.ieee.org/document/10496217) Proposing a cognitive modeling approach that predicts selections from item triplets while providing interpretable context and item representations. 2024 [![Cover - LLM alignment with cognitive processes](https://www.recombee.com/img/publications/llm-alignment-with-cognitive-processes.png)](https://doi.org/10.1145/3709148) ## [LLM alignment with cognitive processes](https://doi.org/10.1145/3709148) Proposition and analysis of a methodology for assessing alignment of large language models with cognitive processes. 2024 [![Cover - Minimum item exposure guarantees](https://www.recombee.com/img/publications/minimum-item-exposure-guarantees.png)](https://www.sciencedirect.com/science/article/pii/S0957417423016664) ## [Minimum item exposure guarantees](https://www.sciencedirect.com/science/article/pii/S0957417423016664) Proposing a method to enhance the fairness (by dealing with bias) of item exposure in recommendation lists. 2024 [![Cover - Improved inductive matrix factorization](https://www.recombee.com/img/publications/improved-inductive-matrix-factorization.png)](https://ojs.aaai.org/index.php/AAAI/article/view/26018) ## [Improved inductive matrix factorization](https://ojs.aaai.org/index.php/AAAI/article/view/26018) Proposition of a method for improving inductive matrix factorization, achieving better accuracy, especially in noisy or incomplete data scenarios. 2023 [![Cover - Improving matrix factorization for recommendation](https://www.recombee.com/img/publications/improving-matrix-factorization-for-recommendation.png)](https://ieeexplore.ieee.org/document/10177891) ## [Improving matrix factorization for recommendation](https://ieeexplore.ieee.org/document/10177891) Development of a method for improving matrix-factorization-based recommendations by adjusting uncertainty in the feedback process by using side information. 2023 [![Cover - GNN enhanced matrix factorization](https://www.recombee.com/img/publications/gnn-enhanced-matrix-factorization.png)](https://doi.org/10.1145/3604915.3610654) ## [GNN enhanced matrix factorization](https://doi.org/10.1145/3604915.3610654) Proposition of a technique to enhance recommendations from matrix factorization by incorporating uncertainty adjustments in the feedback mechanism through the use of graph neural networks. 2023 [![Cover - Bridging Offline-Online Evaluation](https://www.recombee.com/img/publications/bridging-offline-online-evaluation.png)](https://www.sciencedirect.com/science/article/pii/S0957417423016664) ## [Bridging Offline-Online Evaluation](https://www.sciencedirect.com/science/article/pii/S0957417423016664) Proposing a method to bridge offline-online evaluation in time-dependent and popularity contexts. 2023 --- # Contact > Source: https://www.recombee.com/contact > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. Contact # Let’s Personalize Your Recommendations or any additional information you need to know about Recombee at [business@recombee.com](mailto:business@recombee.com) or call on +420 604 499 078 ## Product & Business [business@recombee.com](mailto:business@recombee.com) +420 604 499 078 ## Customer Support [support@recombee.com](mailto:support@recombee.com) ## Contact Address Recombee Vaclavske namesti 1 110 00 Praha [Show on map](https://www.google.com/maps/place/Václavské+nám.+1,+110+00+Můstek,+Česko/@50.0841919,14.4216469,782m/data=!3m2!1e3!4b1!4m6!3m5!1s0x470b94ec630fac27:0x1bdb29c896981c41!8m2!3d50.0841919!4d14.4242272!16s%2Fg%2F11bw3fgykg!5m1!1e2) ## Invoicing Address Recombee s.r.o. Rybna 716/24 110 00 Stare Mesto [Show on map](https://www.google.com/maps/place/Rybn%C3%A1+716%2F24,+110+00+Praha+1-Star%C3%A9+M%C4%9Bsto/@50.0902861,14.4238861,17z/data=!3m1!4b1!4m5!3m4!1s0x470b94ea0da96c41:0xd926809755e5bf9a!8m2!3d50.0902861!4d14.4260748?hl=en) #### Get in Touch on Social Networks [LinkedIn](https://www.linkedin.com/company/recombee/) [YouTube](https://www.youtube.com/@recombee) [GitHub](https://github.com/recombee) --- # About Us > Source: https://www.recombee.com/about-us > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. ![](https://www.recombee.com/img/bg/polygon-green-xs.svg) ![](https://www.recombee.com/img/bg/polygon-green-2-md.svg) ![](https://www.recombee.com/img/bg/polygon-green-sm.svg) ![](https://www.recombee.com/img/bg/polygon-yellow-sm.svg) About us # Building the Future of Personalization Recombee was founded in 2015 by machine learning and AI experts with a simple mission: to help businesses make digital experiences more relevant to their audiences. Today, our recommendation engine powers 10,000+ sites and apps worldwide, from DAZN and The Telegraph to Tripadvisor and Apify, across media, publishing, e-commerce, streaming, and more. ## Behind the technology is a team of passionate engineers and product innovators who enjoy solving complex problems, challenging the status quo, and pushing the boundaries of recommender systems. ### Co-founders ![](https://www.recombee.com/img/team/pavel-kordik.png) Pavel Kordik, Ph.D. CEO, Co-Founder [](https://www.linkedin.com/in/kordik/) [](#memberpavel-kordik) Data scientist with 16 years of experience in the field of predictive modeling and data mining. Pavel is also a lecturer of several machine learning courses, such as Algorithms of Data Mining and Knowledge Engineering Methods at Czech Technical University in Prague. ![](https://www.recombee.com/img/team/tomas-rehorek.png) Tomas Rehorek, Ph.D. CTO, Co-Founder [](https://www.linkedin.com/in/tomas-rehorek/) [](#membertomas-rehorek) Runs a team of data scientists responsible for scalable algorithms and infrastructure. Tomas specializes in recommendation algorithms and has been lecturing AI for 5 years at FIT CTU in Prague. He overlooks product research and innovations. ![](https://www.recombee.com/img/team/ondrej-fiedler.png) Ondrej Fiedler CPO, Co-Founder [](https://www.linkedin.com/in/ond%C5%99ej-fiedler-7568a899/) [](#memberondrej-fiedler) Overseeing development and deployment of new features to the production and working closely with client’s development teams to ensure smooth and high quality product delivery. ![](https://www.recombee.com/img/team/gabriela-takacova.png) Gabriela Takacova Head of Business, Co-Founder [](https://www.linkedin.com/in/gabriela-takacova/) [](#membergabriela-takacova) With rich international experience from working and studying in 6 countries, responsible for development and implementation of a growth strategy, market expansion, new customer acquisition and brand awareness. ![](https://www.recombee.com/img/team/antonin-kral.png) Antonin Kral, Ph.D. Data Protection, Co-Founder [](https://www.linkedin.com/in/kralant/) [](#memberantonin-kral) Antonin helps with designing and building the global server infrastructure. He has tremendous experience in distributed systems, data streaming and processing, services orchestration, and machine learning. ![](https://www.recombee.com/img/team/tomas-barton.png) Tomas Barton, Ph.D. Head of Infrastructure, Co-Founder [](https://www.linkedin.com/in/bartontomas/) [](#membertomas-barton) Is the architect of Recombee private cloud infrastructure. Tomas manages infrastructure advancements and adoption of cutting-edge technologies, as well as maintenance, monitoring, and data security. ### Machine Learning ![](https://www.recombee.com/img/team/josef-malik.png) Josef Malik Head of Machine Learning [](https://www.linkedin.com/in/josefmalik/) [](#memberjosef-malik) ![](https://www.recombee.com/img/team/ladislav-martinek.png) Ladislav Martinek Machine Learning Developer [](https://www.linkedin.com/in/ladislav-mart%C3%ADnek-aab44b135/) [](#memberladislav-martinek) Combines profound knowledge of machine learning and theoretical computer science while implementing optimized and highly efficient real-time models. ![](https://www.recombee.com/img/team/michael-kolinsky.png) Michael Kolinsky Machine Learning Developer [](https://www.linkedin.com/in/michael-k-72696610a/) [](#membermichael-kolinsky) ![](https://www.recombee.com/img/team/tana-lancova.png) Tana Lancova Machine Learning Specialist [](https://www.linkedin.com/in/t%C3%A1%C5%88a-lan%C4%8Dov%C3%A1-9aa1981b6/) [](#membertana-lancova) ![](https://www.recombee.com/img/team/vojtech-rozhon.png) Vojtech Rozhon Machine Learning Developer [](https://www.linkedin.com/in/vojt%C4%9Bch-rozho%C5%88-63b7231b5/) [](#membervojtech-rozhon) ![](https://www.recombee.com/img/team/jan-machacek.png) Jan Machacek Machine Learning Developer [](https://www.linkedin.com/in/jan-machacek1/) [](#memberjan-machacek) ![](https://www.recombee.com/img/team/jachym-stanek.png) Jachym Stanek Machine Learning Developer [](https://www.linkedin.com/in/j%C3%A1chym-stan%C4%9Bk-204790270/) [](#memberjachym-stanek) ![](https://www.recombee.com/img/team/daniel-kral.png) Daniel Kral Machine Learning Developer [](https://www.linkedin.com/in/daniel-kr%C3%A1l-7a51992a4/) [](#memberdaniel-kral) ![](https://www.recombee.com/img/team/hana-cassi.png) Hana Cassi Support Specialist [](https://www.linkedin.com/in/hana-cassi-pelikan-04910414/) [](#memberhana-cassi) ![](https://www.recombee.com/img/team/jan-feber.png) Jan Feber Solution Delivery Specialist [](https://cz.linkedin.com/in/jan-feber-497991104) [](#memberjan-feber) ### Research ![](https://www.recombee.com/img/team/pavel-kordik.png) Pavel Kordik, Ph.D. CEO, Co-Founder [](https://www.linkedin.com/in/kordik/) [](#memberpavel-kordik-2) Data scientist with 16 years of experience in the field of predictive modeling and data mining. Pavel is also a lecturer of several machine learning courses, such as Algorithms of Data Mining and Knowledge Engineering Methods at Czech Technical University in Prague. ![](https://www.recombee.com/img/team/rodrigo-alves.png) Dr. Rodrigo Alves Head of Research [](https://www.linkedin.com/in/rodrigo-alves-54333524) [](#memberrodrigo-alves) ![](https://www.recombee.com/img/team/vojtech-vancura.png) Vojtech Vancura, Ph.D. Machine Learning Researcher [](https://www.linkedin.com/in/vojtech-vancura/) [](#membervojtech-vancura) ![](https://www.recombee.com/img/team/petr-kasalicky.png) Petr Kasalicky Head of Applied Science [](https://www.linkedin.com/in/petr-kasalick%C3%BD-2b17b8b8/) [](#memberpetr-kasalicky) ![](https://www.recombee.com/img/team/dan-bohunek.png) Dan Bohunek Applied Scientist [](https://www.linkedin.com/in/daniel-bohunek) [](#memberdan-bohunek) ![](https://www.recombee.com/img/team/jaroslav-hradil.png) Jaroslav Hradil Applied Scientist [](https://www.linkedin.com/in/jaroslav-hradil/) [](#memberjaroslav-hradil) ![](https://www.recombee.com/img/team/vojtech-nekl.png) Vojtech Nekl Applied Scientist [](https://www.linkedin.com/in/vojtěch-nekl-564a17385/) [](#membervojtech-nekl) ![](https://www.recombee.com/img/team/david-lapunik.png) David Lapunik Machine Learning Researcher [](https://www.linkedin.com/in/david-lapuník-0a346a2a6/) [](#memberdavid-lapunik) ### Business ![](https://www.recombee.com/img/team/gabriela-takacova.png) Gabriela Takacova Head of Business, Co-Founder [](https://www.linkedin.com/in/gabriela-takacova/) [](#membergabriela-takacova-2) With rich international experience from working and studying in 6 countries, responsible for development and implementation of a growth strategy, market expansion, new customer acquisition and brand awareness. ![](https://www.recombee.com/img/team/filip-hanus.png) Filip Hanus Customer Growth Manager [](https://www.linkedin.com/in/filiphanus/) [](#memberfilip-hanus) ![](https://www.recombee.com/img/team/petr-popov.png) Petr Popov Business Development and Partnership Manager [](https://www.linkedin.com/in/petrpopov/) [](#memberpetr-popov) ![](https://www.recombee.com/img/team/ergys-kosovrasti.png) Ergys Kosovrasti Customer Growth Manager [](https://www.linkedin.com/in/ergys-kosovrasti-28571b19a/) [](#memberergys-kosovrasti) ![](https://www.recombee.com/img/team/lauryn-nigro.png) Lauryn Nigro Business Development Representative [](https://www.linkedin.com/in/lauryn-nigro/) [](#memberlauryn-nigro) ![](https://www.recombee.com/img/team/violeta-milarova.png) Violeta Milarova Copywriter [](https://www.linkedin.com/in/violetadmilarova/) [](#membervioleta-milarova) ![](https://www.recombee.com/img/team/anna-rehorkova.png) Anna Rehorkova Office Manager [](https://www.linkedin.com/in/anna-%C5%99eho%C5%99kov%C3%A1-1029434b/) [](#memberanna-rehorkova) ### Back-end ![](https://www.recombee.com/img/team/ondrej-cvacho.png) Ondrej Cvacho Head of Back-end [](https://www.linkedin.com/in/ond%C5%99ej-cvacho-9a5975152/) [](#memberondrej-cvacho) Taking care of Recombee back-end services. ![](https://www.recombee.com/img/team/michal-demko.png) Michal Demko Back-end Developer [](https://www.linkedin.com/in/michal-demko-6ab843180/) [](#membermichal-demko) ![](https://www.recombee.com/img/team/jan-fiala.png) Jan Fiala Back-end Developer [](https://www.linkedin.com/in/jan-fiala-026863246/) [](#memberjan-fiala) ![](https://www.recombee.com/img/team/michal-bilansky.png) Michal Bilansky Back-end Developer [](https://www.linkedin.com/in/michalbilansky/) [](#membermichal-bilansky) ![](https://www.recombee.com/img/team/peter-grocky.png) Peter Grocky Back-end Developer [](https://www.linkedin.com/in/peter-grocky/) [](#memberpeter-grocky) ![](https://www.recombee.com/img/team/jan-gombos.png) Jan Gombos Back-end Developer [](https://www.linkedin.com/in/jan-gombos01/) [](#memberjan-gombos) ### Front-end ![](https://www.recombee.com/img/team/jiri-brabec.png) Jiri Brabec Lead Front-end Developer [](https://www.linkedin.com/in/brabecjiri/) [](#memberjiri-brabec) Taking care of Recombee admin front-end app. ![](https://www.recombee.com/img/team/daniel-breiner.png) Daniel Breiner Front-end Developer [](https://www.linkedin.com/in/daniel-breiner/) [](#memberdaniel-breiner) ![](https://www.recombee.com/img/team/matej-stieranka.png) Matej Stieranka Front-end Developer [](https://www.linkedin.com/in/mstieranka/) [](#membermatej-stieranka) ![](https://www.recombee.com/img/team/daniel-forrai.png) Daniel Forrai Front-end Developer [](https://www.linkedin.com/in/df-/) [](#memberdaniel-forrai) ### Infrastructure ![](https://www.recombee.com/img/team/antonin-kral.png) Antonin Kral, Ph.D. Data Protection, Co-Founder [](https://www.linkedin.com/in/kralant/) [](#memberantonin-kral-2) Antonin helps with designing and building the global server infrastructure. He has tremendous experience in distributed systems, data streaming and processing, services orchestration, and machine learning. ![](https://www.recombee.com/img/team/tomas-barton.png) Tomas Barton, Ph.D. Head of Infrastructure, Co-Founder [](https://www.linkedin.com/in/bartontomas/) [](#membertomas-barton-2) Is the architect of Recombee private cloud infrastructure. Tomas manages infrastructure advancements and adoption of cutting-edge technologies, as well as maintenance, monitoring, and data security. ![](https://www.recombee.com/img/team/jan-safarik.png) Jan Safarik Infrastructure Engineer [](https://www.linkedin.com/in/jan-safarik/) [](#memberjan-safarik) ![](https://www.recombee.com/img/team/arne-rusek.png) Arne Rusek Infrastructure Engineer [](https://www.linkedin.com/in/arne-rusek/) [](#memberarne-rusek) ### Information Security ![](https://www.recombee.com/img/team/josef-vancura.png) Josef Vancura Chief Information Security Officer [](#memberjosef-vancura) ### Design & Brand ![](https://www.recombee.com/img/team/katerina-kynclova.png) Katerina Kynclova Graphic Designer [](https://www.linkedin.com/in/kate%C5%99ina-kynclov%C3%A1-934b8b160/) [](#memberkaterina-kynclova) [We are hiring! See open positions and join us](https://www.recombee.com/jobs) "Currently, we are able to provide recommender as a service to companies with dozens of millions of users and products with the plan to scale up. We work hard to make our service even more useful and profitable not only for large enterprises but also for small and medium-sized companies." ![](https://www.recombee.com/img/team/pavel-kordik.png) **Pavel Kordik** CEO, Co-founder ## Our Technology Partners [![Kentico Xperience](https://www.recombee.com/img/partners/kentico-xperience.svg)](https://xperience.io) [![Kontent.ai](https://www.recombee.com/img/partners/kontent.svg)](https://kontent.ai) [![Complex](https://www.recombee.com/img/partners/complex.png)](https://www.complex-it.de/) [![Keboola](https://www.recombee.com/img/partners/keboola.svg)](http://www.keboola.com/) [![Segment](https://www.recombee.com/img/partners/segment.svg)](http://www.segment.com/) [![Geneea](https://www.recombee.com/img/partners/geneea.png)](http://www.geneea.com/) [![CloudSailor](https://www.recombee.com/img/partners/cloudsailor.png)](https://www.cloudsailor.eu/) [![Ryzeo](https://www.recombee.com/img/partners/ryzeo.png)](http://www.ryzeo.com/) [![Revium](https://www.recombee.com/img/partners/revium.svg)](https://revium.com.au/) ## Our Academic Partners [![ČVUT](https://www.recombee.com/img/partners/cvut.png)](http://www.cvut.cz/en/) [![Data Science Laboratory](https://www.recombee.com/img/partners/datalab.png)](http://datalab.fit.cvut.cz/) ## Recently Published [![New Feature: A/B Testing](https://www.recombee.com/img/blog/new-feature-ab-testing.png)](https://www.recombee.com/blog/new-feature-ab-testing) ### [New Feature: A/B Testing](https://www.recombee.com/blog/new-feature-ab-testing) Violeta Milarova, Aug 14, 2026 [![Mid-Year Roundup: 2026 So Far](https://www.recombee.com/img/blog/mid-year-roundup-2026-so-far.png)](https://www.recombee.com/blog/mid-year-roundup-2026-so-far) ### [Mid-Year Roundup: 2026 So Far](https://www.recombee.com/blog/mid-year-roundup-2026-so-far) Violeta Milarova, Jul 20, 2026 [![A 2025 Research Retrospective](https://www.recombee.com/img/blog/a-2025-research-retrospective.png)](https://www.recombee.com/blog/a-2025-research-retrospective) ### [A 2025 Research Retrospective](https://www.recombee.com/blog/a-2025-research-retrospective) Rodrigo Alves, Jan 29, 2026 [![Looking Back at 2025](https://www.recombee.com/img/blog/looking-back-at-2025.png)](https://www.recombee.com/blog/looking-back-at-2025) ### [Looking Back at 2025](https://www.recombee.com/blog/looking-back-at-2025) Violeta Milarova, Jan 20, 2026 [![Product Highlights from 2025](https://www.recombee.com/img/blog/product-highlights-from-2025.png)](https://www.recombee.com/blog/product-highlights-from-2025) ### [Product Highlights from 2025](https://www.recombee.com/blog/product-highlights-from-2025) Ondrej Fiedler, Jan 08, 2026 [![AI News and Outlook for 2026](https://www.recombee.com/img/blog/ai-news-and-outlook-for-2026.png)](https://www.recombee.com/blog/ai-news-and-outlook-for-2026) ### [AI News and Outlook for 2026](https://www.recombee.com/blog/ai-news-and-outlook-for-2026) Pavel Kordik, Dec 17, 2025 Read more articles on [Blog ->](https://www.recombee.com/blog) --- # Explore Our Partnership Opportunities > Source: https://www.recombee.com/partnerships > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. Partnerships # Become a Recombee Partner Grow your business together with the most advanced recommendation engine on the market. Boost your client’s satisfaction and KPIs with real-time content, product and search personalization. ![](https://www.recombee.com/img/partnership.svg) ### Partnership Program Benefits #### Unlock New Revenue Stream Earn additional revenue in the form of commissions or exclusive discounts to provide best-in-class personalization experiences to your customers. #### Education and Materials Increase your knowledge of AI & Personalization by accessing Recombee’s marketing and educational materials. #### 1:1 Dedicated Support Enjoy ongoing 1:1 guidance and support from our team, ensuring you have the tools and resources needed to succeed. ### Are We a Good Fit? You’re ready to level up your business and lead the way in tech innovation by offering clients the latest AI solutions. With the tech know-how already in place, you’re all set to implement and start seeing results right away. Help clients unlock the full potential of AI personalization by integrating Recombee into their systems, delivering custom-tailored experiences across digital platforms. Boost engagement and conversions by incorporating personalized product recommendations into email campaigns, driving higher click-through rates and customer interaction. Supercharge customer data by integrating Recombee for personalized product and content recommendations, enabling more precise and effective marketing strategies. Elevate the user experience with dynamic, personalized recommendations on content management systems, enhancing engagement and driving more conversions. Deliver hyper-targeted content recommendations based on viewer preferences and behavior, increasing satisfaction and boosting overall viewership. Enhance the buyer’s journey with personalized product recommendations that drive higher cart values and conversion rates at every touchpoint. ### Types of Partnerships We offer various partnership opportunities, tailored to different business needs. As a partner, you will earn a share of yearly subscription fees based on the Monthly Recurring Revenue (MRR) of the client and your involvement in the process. If you love connecting people with great solutions, this one's for you! Simply introduce new clients to Recombee, and we’ll handle the tech. You’ll earn rewards for bringing in new business without having to dive into the implementation. Ideal for tech-savvy agencies and studios looking to take a hands-on approach. You’ll not only refer clients but also guide them in integrating Recombee’s recommendations into their platforms, giving them a truly personalized service. For platforms like CDPs, CMSs, OTT services, and email marketing tools, integrating Recombee’s powerful engine into your offering means your clients get instant access to AI-driven personalization, making your platform even more valuable. ### Get on Board, You’ll Be in Good Company [![](https://www.recombee.com/img/partners/kentico-xperience.svg)](https://www.kentico.com/platforms/xperience-by-kentico) [![](https://www.recombee.com/img/partners/kontent.svg)](https://kontent.ai) [![](https://www.recombee.com/img/partners/complex.png)](https://www.complex-it.de) [![](https://www.recombee.com/img/partners/keboola.svg)](https://www.keboola.com) [![](https://www.recombee.com/img/partners/segment.svg)](https://segment.com) [![](https://www.recombee.com/img/partners/geneea.png)](https://geneea.com) [![](https://www.recombee.com/img/partners/ryzeo.png)](https://ryzeo.com) [![](https://www.recombee.com/img/partners/revium-new.svg)](https://revium.com.au) [![](https://www.recombee.com/img/partners/expertsender.svg)](https://expertsender.com) [![](https://www.recombee.com/img/partners/axinom.svg)](https://www.axinom.com/) [![](https://www.recombee.com/img/partners/diagnal.svg)](https://www.diagnal.com/) [![](https://www.recombee.com/img/partners/polcode.svg)](https://polcode.com/) [![](https://www.recombee.com/img/partners/prg-ai.svg)](https://prg.ai/) [![](https://www.recombee.com/img/partners/infobip.svg)](https://www.infobip.com/) [![](https://www.recombee.com/img/partners/hdl.svg)](https://hbgdesignlab.se/) [![](https://www.recombee.com/img/partners/targito.svg)](https://www.targito.com/) [![](https://www.recombee.com/img/partners/modern-tv.svg)](https://www.moderntv.eu/) [![](https://www.recombee.com/img/partners/morpht.svg)](https://www.morpht.com/) [![](https://www.recombee.com/img/partners/treefort.svg)](https://treefortsystems.com/) [![](https://www.recombee.com/img/partners/umbraco.svg)](https://umbraco.com/) [![](https://www.recombee.com/img/partners/vianeos.svg)](https://www.vianeos.com/) [![](https://www.recombee.com/img/partners/bounce-commerce.svg)](https://www.bounce-commerce.de/) [![](https://www.recombee.com/img/partners/riesenia.svg)](https://www.riesenia.com/) "It has become very difficult for companies to handle all their customer journeys and patterns, so if you’d like to provide visitors with the best content and customers with the best products, you need to start thinking AI." ![David Komarek](https://www.recombee.com/img/customers/kentico.png) David KomarekProduct Owner at Kentico Software "At Complex, we choose our tools very carefully - invest into intensive analysis, test phases and target specific KPIs. With Recombee, we found a compatible match that positively surprised us - easy integration, the API is comparatively simple and easy to understand, great personal support, and any flexibility in the front end that you could wish for." ![Pascal Pischel](https://www.recombee.com/img/customers/complex.png) Pascal PischelBusiness Development at Complex GmbH & Co. KG ### How It Works #### Intro Let’s start by exploring your business goals and how a partnership with Recombee can support your growth. We’ll follow up promptly with a tailored proposal, crafted to meet your specific needs. #### Setup Setting up your partnership account is fast and free, usually taking just one day. To ensure a smooth start, our team will equip you with all the resources and documentation needed for easy integration. Plus, you’ll enjoy a 1-month free trial for each client. #### Collaboration Once the partnership is active, you’ll have full access to both technical and marketing support. Whether you’re referring clients or embedding our solution into your services, we’re here to help you succeed every step of the way. --- # Careers at Recombee | Be part of AI revolution > Source: https://www.recombee.com/jobs > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. ![](https://www.recombee.com/img/bg/polygon-green-xs.svg) ![](https://www.recombee.com/img/bg/polygon-green-2-md.svg) Jobs # Join AI Visionaries **Recombee** is a fast-growing hi-tech scaleup delivering next-gen recommendation and personalization solutions. We help our clients reach their business goals by applying advanced machine learning and AI to large-scale data, helping users discover products and content they will enjoy. Our recommendation technology is highly versatile, serving digital platforms across multiple industries including news, media, video, e-commerce, music, P2P marketplaces, sports, and more. We’re based in Prague and serve 10,000+ sites and apps worldwide, powering personalization for leading brands such as DAZN, The Telegraph, and Apify. **What We Build** * Real-time personalization at scale * Advanced machine learning and recommendation technology * High-performance infrastructure built for millions of interactions We’re a team of hi-tech enthusiasts, always looking for open-minded, creative, and passionate people to join us in the vibrant city of Prague. [Backend Developer — Python](https://www.recombee.com/jobs/backend-developer) [Sales Manager / Strategic Account Executive](https://www.recombee.com/jobs/sales-manager) [Growth Marketing Specialist / Growth Engineer](https://www.recombee.com/jobs/growth-marketing-specialist) [Business & Communications Specialist](https://www.recombee.com/jobs/business-and-communications-specialist) --- # Backend Developer — Python | Careers at Recombee | Be part of AI revolution > Source: https://www.recombee.com/jobs/backend-developer > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Jobs**](https://www.recombee.com/jobs) # Backend Developer — Python Recombee · Prague, Czech Republic · Full-time · Prague-first, Home-office friendly #### About Us **Recombee** is a fast-growing hi-tech scaleup delivering next-gen recommendation and personalization solutions. We help our clients reach their business goals by applying advanced machine learning and AI to large-scale data, helping users discover products and content they will enjoy. Our recommendation technology is highly versatile, serving digital platforms across multiple industries including news, media, video, e-commerce, music, P2P marketplaces, sports, and more. We’re based in Prague and serve 10,000+ sites and apps worldwide, powering personalization for leading brands such as DAZN, The Telegraph, and Apify. #### Our Tech Stack * Primary language: Python · modern C++ * Architecture: Distributed microservices * Databases: PostgreSQL · ClickHouse · Aerospike * Messaging: Kafka * API layer: GraphQL · REST * Orchestration: Kubernetes (expanding) * CI/CD: GitLab CI * AI tooling: LLM APIs · AI-assisted coding You don't need experience with every item above — we value curiosity and willingness to learn over a perfect match to any specific tool. #### What You'll Work On * Design and evolve components in our distributed microservice architecture — things that process serious data volumes with high availability requirements * Build and maintain APIs consumed by client integrations worldwide * Work across multiple database systems and contribute to decisions on when and how to use each * Help shape our growing Kubernetes footprint as we expand our infrastructure * Use AI tools and LLM APIs where they genuinely accelerate development — we're pragmatic about it #### What We're Looking For * Experience with distributed systems * Comfort building or consuming REST or GraphQL APIs * Working knowledge of SQL and an understanding of when relational storage fits (and when it doesn't) * Familiarity with LLM APIs (OpenAI, Anthropic, or similar) or AI-assisted development tools — or genuine eagerness to get there * A habit of thinking through operational consequences: latency, reliability, observability Kubernetes experience is a nice-to-have, not a requirement — we'll grow together on that front. #### What We Offer * 5 weeks of vacation + 5 sick days * Flexible hours — office-first with unlimited remote when you need it * Experienced team with deep technical knowledge to share * Office in the city center of Prague * Mobile tariff with unlimited data * MultiSport card #### We offer a dynamic and international working environment, flat hierarchies, fascinating challenges, and great personal responsibilities. Are you ready to push boundaries, inspire, and most importantly, be yourself? Bring a smile and become part of the Recombee family. ### Apply Today! We Are Looking Forward to Your Application! Please contact us at [career@recombee.com](mailto:career@recombee.com) [Apply for Backend Developer — Python](mailto:career@recombee.com) --- # Business & Communications Specialist | Careers at Recombee | Be part of AI revolution > Source: https://www.recombee.com/jobs/business-and-communications-specialist > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Jobs**](https://www.recombee.com/jobs) # Business & Communications Specialist Recombee · Prague, Czech Republic · Full-time #### About Us **Recombee** is a fast-growing hi-tech scaleup delivering next-gen recommendation and personalization solutions. We help our clients reach their business goals by applying advanced machine learning and AI to large-scale data, helping users discover products and content they will enjoy. Our recommendation technology is highly versatile, serving digital platforms across multiple industries including news, media, video, e-commerce, music, P2P marketplaces, sports, and more. We’re based in Prague and serve 10,000+ sites and apps worldwide, powering personalization for leading brands such as DAZN, The Telegraph, and Apify. #### About the Role We are looking for a Business & Communications Specialist to work closely with our CBO and help coordinate the dynamic day-to-day life of our business team. This is a great role for a smart, flexible generalist who enjoys variety, likes making things happen, and is excited to learn how an international AI SaaS company works. You will help keep initiatives moving, coordinate people and follow-ups, support events and content activities, and bring structure to new processes. You will not be doing this alone. You will get support from our content and design colleagues, office manager and other colleagues across Recombee. #### What You’ll Do * Support the CBO with coordination and execution of business activities. * Keep track of priorities, follow-ups, responsibilities, and next steps. * Coordinate cross-team initiatives such as trainings, hackathons, strategic projects, and events. * Help introduce new processes and make sure they are followed. * Support PR, brand-building, and content activities. * Coordinate content production related to CBO’s PR activities and conferences with our copywriter and design team (e.g. coordinate interviews and recordings with Recombee experts). * Help capture videos, and other content at conferences and speaking engagements. * Support the CBO’s online presence and relationship-building activities. * Work with our Office Manager on conference materials, logistics, and event preparation. * Learn to use AI and automation tools to make recurring work more efficient. * Occasionally travel with the team to international conferences and events. #### What We’re Looking For * Around 2–5 years of professional experience, ideally in a technology, startup, SaaS, marketing, consulting, or similarly dynamic environment. * Excellent written and spoken English. * Strong organizational and coordination skills. * Ability to take broad direction and turn it into concrete actions. * Reliability, ownership, and strong attention to detail. * Comfort working across several activities and teams at the same time. * Interest in business, technology, marketing, PR, content, and events. * Enjoyment of social media and willingness to work with basic photo and video content. * Curiosity and willingness to learn new AI, automation, and productivity tools. * A flexible mindset and genuine enjoyment of a fast-moving environment. #### Nice to Have * Previous experience in a tech company. * Experience in business operations, project coordination, marketing, communications, or events. * Experience coordinating cross-functional projects. * Familiarity with AI or automation tools. * Basic experience with video, photography, or social media content. * Czech language is welcome, but not required. #### What We Offer * Direct collaboration with Recombee’s CBO. * A broad role with exposure to business, marketing, PR, events, and company operations. * Plenty of opportunities to learn, experiment, and take ownership. * Opportunities to join international conferences and events across Europe (Amsterdam, London, Madrid). * Flat hierarchy and access to company leadership. * Office in the city center of Prague. * 5 weeks of vacation and 5 sick days. * Mobile tariff with unlimited data. * MultiSport card. #### We offer a dynamic and international working environment, flat hierarchies, fascinating challenges, and great personal responsibilities. Are you ready to push boundaries, inspire, and most importantly, be yourself? Bring a smile and become part of the Recombee family. ### Apply Today! We Are Looking Forward to Your Application! Please contact us at [career@recombee.com](mailto:career@recombee.com) [Apply for Business & Communications Specialist](mailto:career@recombee.com) --- # Growth Marketing Specialist / Growth Engineer | Careers at Recombee | Be part of AI revolution > Source: https://www.recombee.com/jobs/growth-marketing-specialist > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Jobs**](https://www.recombee.com/jobs) # Growth Marketing Specialist / Growth Engineer Recombee · Prague, Czech Republic · Full-time #### About Us **Recombee** is a fast-growing hi-tech scaleup delivering next-gen recommendation and personalization solutions. We help our clients reach their business goals by applying advanced machine learning and AI to large-scale data, helping users discover products and content they will enjoy. Our recommendation technology is highly versatile, serving digital platforms across multiple industries including news, media, video, e-commerce, music, P2P marketplaces, sports, and more. We’re based in Prague and serve 10,000+ sites and apps worldwide, powering personalization for leading brands such as DAZN, The Telegraph, and Apify. #### About the Role We are looking for an analytical Growth Marketing Specialist to **discover, test, and scale new ways of generating qualified leads for Recombee.** You will analyze our current growth performance, identify promising opportunities and bottlenecks, and run focused experiments across acquisition, website conversion, content, campaigns, data, and automation. This is a role for someone who is curious, self-managed, and biased toward execution. You should be comfortable using LLMs, AI tools, automation, and data enrichment to test ideas faster, improve existing processes, and create scalable growth systems. You will report to the Chief Business Officer and work closely with sales and other colleagues. Qualified leads will be your primary outcome, but you will have significant independence in deciding how to improve that number. #### What You’ll Do * Identify the most important growth opportunities and bottlenecks, then turn them into clear, testable hypotheses. * Monitor growth performance across PPC spend, website traffic, conversion, qualified leads, demo calls, content, events and other relevant channels. * Maintain a prioritized pipeline of experiments, define success criteria, execute tests, evaluate results, and scale what works. * Manage and optimize paid acquisition campaigns, budgets, targeting, messaging, and landing pages with a focus on lead quality and commercial impact. * Improve website conversion by testing positioning, content, calls to action, landing pages, forms, and other parts of the visitor journey. * Identify content opportunities, coordinate production and distribution with the content marketing team, and measure how individual content pieces contribute to traffic, engagement, and qualified leads. * Test new growth tactics instead of relying only on existing activities. * Build automation for research, data enrichment, lead qualification, reporting, campaign execution, and other repetitive business processes. * Work closely with sales to understand lead quality, improve targeting, and connect marketing activity with commercial outcomes. #### What We’re Looking For * Experience in B2B SaaS, AI, data, personalization, martech, media tech, e-commerce tech, or another technical product environment. * Experience running and evaluating growth experiments across multiple channels or stages of the funnel. * Good understanding of website analytics, conversion funnels, landing pages, attribution, and conversion-rate optimization. * Practical experience using LLMs, AI tools, automation, and data enrichment to improve the speed or quality of marketing and business processes. * A versatile, hands-on working style: comfortable executing independently, learning unfamiliar tools, and coordinating external specialists when necessary. * Strong prioritization. * Ownership mindset and ability to work independently. * Ability to collaborate with technical colleagues. * Clear communication and professional English. #### Nice to Have * Experience generating demand for enterprise or larger mid-market B2B clients, particularly across longer and more complex sales cycles. * Experience with paid acquisition, particularly Google Ads and LinkedIn Ads, including campaign optimization and budget management. * Familiarity with modern GTM, prospecting, enrichment, and outreach tools such as Clay, Instantly, Apollo, or similar platforms. * Familiarity with Google Analytics 4, Google Tag Manager, Looker Studio, HubSpot or Pipedrive, and website content management systems. * Experience connecting CRM, enrichment, outreach, and reporting tools using Make, n8n, Zapier, APIs, or webhooks. * Understanding of SEO, AI search visibility, lifecycle marketing, lead scoring, or marketing attribution. * Understanding of personalization, recommendations, search, machine learning, or AI products. * Czech language is welcome, but not required. #### What We Offer * Work with proven AI personalization technology used by clients worldwide. * Strategic B2B sales with international clients. * Direct collaboration with experienced technical, research, and business colleagues. * Meaningful responsibility and ownership. * Flat hierarchy and access to company leadership. * Office in the city center of Prague. * 5 weeks of vacation and 5 sick days. * Mobile tariff with unlimited data. * MultiSport card. #### We offer a dynamic and international working environment, flat hierarchies, fascinating challenges, and great personal responsibilities. Are you ready to push boundaries, inspire, and most importantly, be yourself? Bring a smile and become part of the Recombee family. ### Apply Today! We Are Looking Forward to Your Application! Please contact us at [career@recombee.com](mailto:career@recombee.com) [Apply for Growth Marketing Specialist / Growth Engineer](mailto:career@recombee.com) --- # Sales Manager / Strategic Account Executive | Careers at Recombee | Be part of AI revolution > Source: https://www.recombee.com/jobs/sales-manager > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. [**Jobs**](https://www.recombee.com/jobs) # Sales Manager / Strategic Account Executive Recombee · Prague, Czech Republic · Full-time #### About Us **Recombee** is a fast-growing hi-tech scaleup delivering next-gen recommendation and personalization solutions. We help our clients reach their business goals by applying advanced machine learning and AI to large-scale data, helping users discover products and content they will enjoy. Our recommendation technology is highly versatile, serving digital platforms across multiple industries including news, media, video, e-commerce, music, P2P marketplaces, sports, and more. We’re based in Prague and serve 10,000+ sites and apps worldwide, powering personalization for leading brands such as DAZN, The Telegraph, and Apify. #### About the Role We are looking for a Sales Manager to help us win strategic B2B opportunities with high-value clients. This is a consultative sales role. You will work with companies that are not only buying software, but evaluating Recombee as a long-term AI personalization partner. You will own the commercial process from qualification through discovery, POC, negotiation, and contract signature. Your role is to understand the client’s business, identify where Recombee can create measurable value, coordinate internal experts, and move opportunities forward with ownership and clarity. #### What You’ll Do * Qualify new business opportunities and identify strategic leads. * Manage B2B sales opportunities from first contact to contract signature. * Run discovery calls and understand the client’s business model, KPIs, stakeholders, and decision process. * Help define POC scope, success criteria, timeline, responsibilities, and next steps. * Prepare pricing inputs and support pricing strategy with the CBO. * Lead or support commercial negotiations. * Keep internal stakeholders aligned so the client experiences Recombee as one coordinated team. #### What We’re Looking For * Experience in B2B sales or account management, ideally in SaaS, AI, data, personalization, martech, media tech, e-commerce tech, or another technical product environment. * Ability to sell complex solutions where business value, technical feasibility, and stakeholder management all matter. * Strong discovery skills and commercial thinking. * Confidence speaking with senior stakeholders. * Ability to manage longer sales cycles, POCs, negotiations, and multiple stakeholders. * Ownership mindset and ability to work independently. * Ability to collaborate with technical colleagues. * Clear communication and professional English. #### Nice to Have * Experience with enterprise or larger mid-market clients. * Experience with POCs, pilots, A/B tests, or technical evaluation processes. * Understanding of personalization, recommendations, search, machine learning, or AI products. * Czech language is welcome, but not required. #### What We Offer * Work with proven AI personalization technology used by clients worldwide. * Strategic B2B sales with international clients. * Direct collaboration with experienced technical, research, and business colleagues. * Meaningful responsibility and ownership. * Flat hierarchy and access to company leadership. * Office in the city center of Prague. * 5 weeks of vacation and 5 sick days. * Mobile tariff with unlimited data. * MultiSport card. #### We offer a dynamic and international working environment, flat hierarchies, fascinating challenges, and great personal responsibilities. Are you ready to push boundaries, inspire, and most importantly, be yourself? Bring a smile and become part of the Recombee family. ### Apply Today! We Are Looking Forward to Your Application! Please contact us at [career@recombee.com](mailto:career@recombee.com) [Apply for Sales Manager / Strategic Account Executive](mailto:career@recombee.com) --- # Terms of Service > Source: https://www.recombee.com/terms-of-use > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # Recombee Terms of Service LAST UPDATED: March 26th 2021 EFFECTIVE DATE: March 26th 2021 These Terms of Service (“Terms”) govern subscription to and use of Recombee’s services. These Terms are entered into by and between Recombee, s.r.o. with a principal place of business at Rybna 716/24, Stare Mesto, 110 00 Praha 1, Czech Republic („Recombee“), and you ("Subscriber"), (each a „Party“ and collectively the „Parties“). If you register for a free trial of Recombee’s services, the applicable provisions of these Terms also govern that free trial and are effective as of the date you click to accept the Terms (the "Effective Date"). If you are acting on behalf of a legal entity you represent (e.g. a company), you represent and warrant that: I. you have full legal authority to bind Subscriber to these Terms; II. you have read and understand these Terms; and III. you agree, on behalf of Subscriber, to these Terms. If you do not have the legal authority to bind Subscriber, please do not click to accept, access or use Recombee’s services. For an offline (printed) version of Terms please contact Recombee. ## 1\. GENERAL TERMS OF SERVICE These Terms are made for the purpose of controlling the delivery of Recombee’s Services and granting a subscription to use Recombee’s Recommendation Service, support and other services (as defined below) to Subscriber. Recombee shall provide recommendation services (hereinafter “the Services”) to Subscriber. The recommendations are made by a set of algorithms that run upon Subscriber’s request on Recombee’s infrastructure (hereinafter “the Recommender”). Instructions on how to use the Recommender correctly are listed on docs.recombee.com (hereinafter “the Documentation”). In order to provide the Services, Subscriber may be required to connect to Recombee’s systems or network („Recombee Network“). Recombee shall provide to Subscriber the API to upload data (such as users, items and interactions) into the Recombee Network and access recommendations. Recombee also provides software clients and applications simplifying access to the Services; these clients are provided without any guarantee. Recombee shall maintain the Documentation in a state that allows for a correct and effective use of the Recommender. Subscriber shall follow the Documentation; Recombee shall not be responsible for any loss or troubles resulting from Subscriber’s failure to follow the Documentation. Recombee shall provide Subscriber with an access to the Admin User Interface (located at admin.recombee.com) through which Subscriber may manage the Services. Upon the access to the Admin User Interface an Organization is created. At Recombee, an Organization refers to an established method for connecting a Subscriber (user) and the Services. Recombee has no obligation to provide Subscriber with multiple Organizations.There is a possibility to invite multiple Collaborators to one Organization. Subscriber agrees not to use the Services for processing of personal data. Should Subscriber be interested in personal data processing, a separate agreement would be negotiated. ### 1.1\. USE OF SERVICES The Services are to be used solely for Subscriber’s internal business purposes and are not for resale to any third party or use on a service bureau basis. Subscriber shall only use the Recombee Network for lawful business purposes. Subscriber shall not use or allow the use of the Recombee Network in a manner that interferes with the use of the Recombee Network by Recombee or by any other authorized, third party user. Subscriber shall use the Services only for recommendation services. The use of the Services for other purposes will be considered as an abuse of Terms and may lead to an account suspension. Unless explicitly agreed otherwise, Subscriber shall have sole responsibility for the expenses associated with deployment of any hardware or software necessary to access the Recombee Network. Each Party retains all right, title and interest in and to its intellectual property. No licenses will be deemed to have been granted by either Party to any of its intellectual property. Subscriber agrees not to modify, copy, or reverse engineer the Recommender or any other software used by Recombee in the provision of the Services. ## 2\. FREE SERVICES ### 2.1 FREE TRIAL If Subscriber registers at Recombee’s websites for a free trial ("Trial Plan") of the Services, it will automatically start a Trial Plan period of the Services on the day of registration. Recombee will make the Services available to Subscriber on a trial basis free of charge until (a) the end of a thirty-day trial period (the “Trial Period”), (b) the start of any paid subscription for the Services, or (c) termination by Recombee at its sole discretion. If, at the end of the Trial Period, Subscriber does not sign up for a paid subscription of the Services, their account will be automatically downgraded into a Free Plan mode. The Trial Period lasts 30 days unless Recombee agrees, in its sole discretion, to extend it. All Trial Plans are provided without warranty of any kind. Recombee may terminate or immediately suspend a Trial Plan at any time and for any reason (or no reason). Recombee does not hold any obligation or liability for the damage that could arise from using a Trial Plan, along with any obligation or accountability with respect to Subscriber’s data (including limited liability of Free Services mentioned in section 12.1). ### 2.2 FREE PLAN In case of Free Plan the traffic is limited to: 100,000 monthly recommendation requests; 20,000 monthly active users and 2 databases. Subscriber will be notified about reaching the limit and additional requests will result in an error message (HTTP 429). There are no SLA guarantees for the Free Plan Subscribers, i.e. Recombee does not guarantee the Services’ availability, uptime or response time, nor Subscriber support response time. All Free Plans are provided without warranty of any kind. Recombee may terminate or immediately suspend a Free Plan at any time and for any reason (or no reason). Recombee does not hold any obligation or liability for the damage that could arise from using the Free Plan, along with any obligation or accountability with respect to Subscribers’s data (including limited liability of Free Services mentioned in section 14.1). ## 3\. PAID SERVICES In case of paid subscription to Recombee services, Recombee guarantees that the Services will meet the following standards (calculated per calendar year): I. 99.5% uptime for the Recommender; II. Average response time less than 500 milliseconds; III. 95% Admin User Interface availability; Recombee shall use commercially reasonable efforts to provide the Services without any service outage. Should the Services not meet the guarantees, Recombee shall provide Subscriber with an appropriate discount. ### 3.1 USAGE LIMITS The Services ecompase specific usage limits, e.g., every subscription needs to be bound to a single domain name (Organization) and specified usage limits that are agreed in a relevant Order Form (Simplified usage limits information can be found at: https://www.recombee.com/pricing). If the Services are accessed in a way that exceeds the specified contractual traffic limits or storage or bandwidth limit a new offer will be automatically generated for Subscriber’s approval. ## 4\. TERM & TERMINATION ### 4.1 FREE SERVICES The contractual relationship shall commence on the Effective Date and shall continue until Recombee or Subscriber decides to terminate it. Recombee may immediately terminate it upon written notice, e.g. if Subscriber breaches the Terms by abusing the Services. If Subscriber wishes to terminate the contractual relationship, he shall send Recombee a termination request and Recombee will close Subscriber’s account and delete Subscriber’s data within 30 days. Subscriber can terminate Recombee Service subscription by email to support@recombee.com. After Recombee receives the termination notice, the Services will be provided to Subscribers throughout the remainder of the then-current term. Subscriber’s Organization will be suspended effective as of the final day of the then-current term. ### 4.2 PAID SERVICES The contract is entered into for an indefinite term. If not stated otherwise in the customized offer, either Party may terminate this contract with one month notice without giving any reason. The notice shall be delivered by email to support@recombee.com and the notice period shall commence on the first day of the calendar month following the month in which the notice is delivered. If Subscriber attempts to breach Terms by performing activities abusing the Services, all suspicious accounts will be deleted immediately. ## 5\. TERMINATION FOR INACTIVITY ### 5.1 FREE SERVICES Recombee reserves the right to terminate the Services for inactivity, if, for a period exceeding 180 days Recombee has not served any Recommendation requests for Subscriber. ## 6\. FEES & PAYMENT TERMS ### 6.1 FREE SERVICES Recombee’s Free Services are provided to Subscriber without charge.. ### 6.2 PAID SERVICES If not agreed otherwise, the Services are available on a subscription basis (for a simplified version of Recombee pricing please visit: https://www.recombee.com/pricing). Subscriber agrees to cover the relevant subscription fees based on the agreement with the customized offer in consideration for the access rights created to Subscriber and the services provided to Subscriber by Recombee. When Subscriber upgrades their Organization to one of Recombee’s paid subscriptions, the Subscriber fully agrees to pay for Recombee’s services. This applies even when Subscriber does not use Recombee’s services for the full month or period. When an Organization is upgraded, it will receive service standards mentioned in Section 3. If not agreed otherwise, Recombee shall issue an invoice on monthly basis with the issue date falling on the same date each month. The first invoice will be issued one month from the day of upgrade for the Services. The issued invoice covers the paid subscription services provided by Recombee for the past month/period. Subscriber shall pay the invoiced amount to Recombee’s bank account within 30 days upon its receipt. The invoice shall contain all prescribed requirements of a tax document set forth by relevant legal regulations. In case the invoice does not contain the prescribed requirements or Subscriber does not agree with the invoiced items, he is obliged to communicate its reservations to Recombee within the invoice’s maturity period. Failing to do so or paying the invoice is considered acceptance of the invoiced amount. In case of legitimate reservations, a new maturity period commences as of the day of delivery of a new or corrected invoice. In cases in which the relevant fees are past the due date, Recombee may charge 10% p.a. Interest on late payment and will be entitled to terminate it’s Services provided to you and suspend your Organization, along with the Authorised User Access towards the Service, up until the amount that is needed to be compensated is fully paid. By providing your credit card details to Recombee, you approve Recombee to give authorisation to our third party full-stack payment platform Braintree (https://www.braintreepayments.com/) to store and keep you card details and charge your credit card for the subscription fees and overage fees that are associated with the basic terms and future renewal terms. These credit card charges will be done in accordance to billing frequencies which are set in the applicable Order Form. In the case Subscriber chooses payment other than credit card in the Order Form, Recombee sends the invoice or PayPal payment request in accordance to billing frequencies which are set in the applicable (usually monthly, in the beginning of the month for the traffic used in the previous month). All fees are exclusive of any tax. Should the Value Added Tax (VAT) or any other tax or duty be applicable to the services, Recombee shall calculate it on the invoice according to the applicable laws and it shall be paid by Subscriber. Recombee reserves the right to change the fees for the Services and to introduce new charges and fees, with a thirty (30) days prior notice to you (which can be sent through an email). If you (Subscriber) would like to make any enquiries, you should get in touch with Recombee’s Subscriber support department by email to support@recombee.com. Third Party Fees. Your credit card issuer may charge you a foreign transaction fee or other charges for the payments. Subscriber is responsible to check with the credit card issuer regarding these details. ## 7\. SERVICE CHANGES ### 7.1 FREE SERVICES Recombee may make upgrades or changes to the Services without prior notice to Subscriber. ### 7.2 PAID SERVICES Recombee may make upgrades or changes to the Services which will not materially diminish the functionality of the Services without prior notice to Subscriber. In the event that a change to the Services would, in Recombee’s reasonable discretion, materially affect Subscriber, Recombee shall provide Subscriber with an advanced notice. ## 8\. USE OF SUBSCRIBER DATA ### FREE & PAID SERVICES Recombee will not access or use Subscriber Data, except those necessary to provide the Services to Subscriber. The Parties have agreed that personal data processing is not part of the Services; should Subscriber want Recombee to process any personal data, a separate agreement needs to be negotiated. ## 9\. CONFIDENTIALITY ### 9.1 FREE SERVICES Each party ("receiving party") agrees to keep and maintain the confidentiality of the other party’s ("disclosing party") confidential information and to disclose it only to its personnel who: I. Have a need to know (and then only to the extent that each such person has a need to know); II. Are aware that the confidential information should be kept confidential; III. Are aware of the receiving party’s undertaking in relation to such information in terms of this agreement; and IV. Have been directed by the receiving party to keep the confidential information and have undertaken to keep and maintain the confidentiality of the confidential information or have signed an appropriate confidentiality and non-disclosure agreements. The receiving party undertakes that if it becomes aware that there has been, as a result of or in the course of the performance of this agreement, unauthorised disclosure or use of the disclosing party’s confidential information, it shall promptly bring the matter to the attention of the disclosing party in writing. The obligations of the parties in relation to the maintenance and non-disclosure of confidential information in terms of this Terms of Service does not extend to information that: Is disclosed to the receiving party in terms of or pursuant to the implementation of this Terms of Service but at the time of such disclosure, such information is known to be in lawful possession or control of the receiving party and not subject to an obligation of confidentiality; Is or becomes public knowledge otherwise than pursuant to a breach of this Terms of Service by the receiving party; Becomes available to the receiving party from a source other than the disclosing party or personnel of the disclosing party; Is required by the provisions of any law, statute or regulation, or during any court proceedings, or by the rules or regulations of any recognised stock exchange to be disclosed and the receiving party required to make the disclosure has taken all reasonable steps to oppose or prevent the disclosure of or to limit, as far as reasonably possible, the extent of such disclosure and has consulted with the disclosing party prior to making such disclosure; or Is, at the time of disclosure, in the public domain. The provisions of this clause shall survive the termination or expiration of this Terms of Service. ### 9.2 PAID SERVICES Whereas one Party may in the course of mutual collaboration disclose to the other Party trade secrets or other confidential information, the receiving Party agrees to keep all disclosed information confidential and use it only for the agreed purpose. The duty of confidentiality does not apply to any information that I. is publicly available; II. is published or made publicly available by a third party after having been disclosed to the receiving Party, provided that the third party did not obtain the information through any breach of confidentiality on the part of the receiving Party; III. was already known to the receiving Party before the execution of the Agreement, provided that the receiving Party was not at that time bound by a duty of confidentiality; IV. the receiving Party obtained from a third party that had acquired it in compliance with the law and without breaching any duty of confidentiality; V. is required by public authorities according to the law, provided that the receiving Party has taken all reasonable steps to oppose, prevent or limit the disclosure. The Parties shall ensure that all of their personnel are bound to respect the duty of confidentiality to the extent stipulated in this Terms of Service. A Party may only be released from the duty of confidentiality by a written declaration from the Party whose information the duty protects. The duty of confidentiality shall survive the termination of the Terms of Service. ## 10\. USE OF NAME & TRADEMARKS Neither Party shall use the trademarks or service marks of the other Party in any advertising, promotional or marketing materials without such other Party’s prior written consent, provided, however, that Recombee may identify Subscriber as a Subscriber of Recombee without prior approval. Notwithstanding the above, Subscriber agrees that it shall participate in either or both a press release and case study with Recombee announcing Subscriber’s use of the Services („Press Release“ and „Case Study“). Either or both the Press Release and the Case Study shall be prepared by Recombee and shall be subject to approval by Subscriber, such approval not to be unreasonably withheld. Recombee may use Subscriber’s name and trademark or service marks on Recombee’s web site and on other tangible and electronic marketing materials, provided that Recombee shall comply with such reasonable trademark or service mark usage guidelines as provided by Subscriber from time to time. ## 11\. REPRESENTATIONS, WARRANTIES & DISCLAIMERS ### 11.1\. FREE AND PAID SERVICES Each Party represents and warrants that it has the requisite corporate power and authority to enter into this Terms of Service and to carry out the transactions contemplated hereunder. Subscriber represents and warrants that it will comply in all respects with the export restrictions applicable to any hardware, software and technology delivered to Subscriber and will otherwise comply with the applicable laws and regulations in effect during the term. ### 11.2 FREE SERVICES RECOMBEE DOES NOT WARRANT THAT THE SERVICES WILL BE UNINTERRUPTED, ERROR FREE OR SECURE AND DOES NOT WARRANT THE SERVICES AGAINST MALFUNCTION OR CESSATION DUE TO CESSATION OR MALFUNCTION OF ANY INTERNET SERVICE PROVIDER OR ANY OF THE THIRD PARTY NETWORKS THAT FORM THE INTERNET. EXCEPT AS SET FORTH HEREIN, ALL SERVICES ARE PROVIDED „AS IS“ AND „AS AVAILABLE“ AND RECOMBEE MAKES NO WARRANTIES TO Subscriber OR TO ANY THIRD PARTY INCLUDING, WITHOUT LIMITATION, END USERS, WHETHER EXPRESS, IMPLIED OR STATUTORY, INCLUDING, BY WAY OF EXAMPLE, WARRANTIES OF MERCHANTABILITY, FITNESS FOR ANY PARTICULAR PURPOSE, TITLE, NON-INFRINGEMENT OR RESULTS TO BE OBTAINED FROM USE OF THE SERVICES, ALL OF WHICH ARE HEREBY EXPRESSLY EXCLUDED AND DISCLAIMED. ## 12\. LIMITATION OF LIABILITY ### 12.1 FREE SERVICES IN NO EVENT SHALL EITHER PARTY BE LIABLE FOR ANY LOST PROFITS, LOST DATA, OR LOST EQUIPMENT, ANY WEBSITE OR NETWORK DOWNTIME, COST OF PROCURING SUBSTITUTE SERVICES OR FOR ANY INDIRECT, INCIDENTAL, SPECIAL, OR CONSEQUENTIAL DAMAGES OF ANY KIND, HOWEVER ARISING, WHICH ARE RELATED TO THIS TERMS OF SERVICE AND THE PROVISION OF SERVICES AND PRODUCTS HEREUNDER, EVEN IF RECOMBEE HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES. ## 12.2 PAID SERVICES Recombee shall not be liable for any failure in provision of the Services resulting from: I. Subscriber’s misuse or abuse of the Services; II. Non-performance or negligent or unlawful acts or omissions by Subscriber or Subscriber’s agents or suppliers; III. Unavailability of Subscriber’s network or the network of the party conducting the query, including that resulting from telecommunications failures; IV. Problems or delays associated with third party networks or networks outside Recombee’s control; V. Force majeure events or denial-of-service attacks or similar malicious attacks on Recombee’s infrastructure or Subscriber, its authorized agents or suppliers; VI. Scheduled maintenance notified to Subscriber in advance. ## 13\. GOVERNING LAW The Terms of Service shall be governed by, and construed in accordance with, the laws of the Czech Republic, without regard to its conflict of law principles that would apply the law of another jurisdiction. ## 14\. ARBITRATION ### 14.1 FREE SERVICES Any Dispute arising out of or relating to this Terms of Service, or the breach thereof, will be settled by final and binding arbitration administered by the Arbitration Court, affiliated to the Czech Chamber of Commerce and the Agricultural Chamber of the Czech Republic by one arbitrator appointed by the Chairman of the Arbitration Court in accordance with the on-line Rules of the Arbitration Court. Each Party will bear its own costs relating to such arbitration, and the Parties will equally share the arbitrators’ fees. The arbitration and all related proceedings and discovery will take place pursuant to a protective order entered by the arbitrators that adequately protects the confidential nature of the Parties’ proprietary and Confidential Information. In no event will any arbitration award provide a remedy beyond those permitted under these Terms, and any award providing a remedy beyond such will not be confirmed, no presumption of validity will attach, and such award will be vacated. In the event that any of the terms and provisions of this Terms of Service are held to be invalid or unenforceable, such determination shall not affect the operation of the remaining provisions of this Terms of Service, which shall remain in full force and effect. The Parties shall replace the invalid or unenforceable provisions with valid and enforceable provisions that best respect the intended objectives of the invalid or unenforceable provisions. The Terms of Service is executed in two counterparts. Each party shall receive one. The Parties declare that they have read the Terms of Service, understand its content, agree with it in full and desire to be bound by it. --- # Cookies > Source: https://www.recombee.com/cookie-policy > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # Recombee Cookie Policy At Recombee we respect individuals’ rights to privacy and value the relationship that we have with you. Hence, below we would like to inform you about our cookie policy. The reason we use cookies is to ensure that your experience on our website is superb and the most productive. The cookies we use help us to understand your behavior, and when we analyse collected data we aim for the best user experience. ### 1 What Is a Cookie and What Is It Good For? Cookies are small files that contain some information. They are downloaded by your web browser when you visit our website. Cookies do a number of very useful jobs, such as remembering your preferences, telling us how you interact with our website or how you found our website. Our use of cookies is limited to internal use, mainly to better organise our web presence, to find out if some sections of our website work the way we want them to work. ### 2 What Kind of Cookies Do We Use? The Google Analytics cookie is considered to be a third-party cookie, which means it is managed by a third party; however, the data it collects are strictly limited to the cookie tracking mentioned above. Regarding their current privacy policy please visit . We generally use Google Analytics to find out more about you as a customer or potential customer. The data collected varies, depending on whether you are logged in with your Google account or not. This cookie tracks, for example, location data, browser type, origination website, time of your session as well as some demographic data, such as your age bracket and gender. The Hotjar cookie analyses your behavior within Recombee’s website and Recombee’s Admin UI. ### 3 How Long Do You Keep Cookie Data? Each cookie is kept for a different period. We do not keep cookie data for more than 13 months. ### 4 How Do I Manage My Cookie? Most internet browsers automatically allow cookies to be stored on your device. Depending on your browser, you should be able to manage acceptance and management of current cookies yourself. You can decide to disable cookies for our website; you can still use our services, but enabled cookies help us greatly, to provide you with the best experience while using our services. If you want to find out more about cookies, please visit [**www.allaboutcookies.org**](http://www.allaboutcookies.org). --- # GDPR Compliant Recommendation Engine > Source: https://www.recombee.com/gdpr > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # Personalize User Content With GDPR Compliant Solution ## Recombee made sure to be prepared​! ## How Recombee secures customer data and our GDPR compliance. ### 1 What Is the GDPR? The European Union has strengthened data protection safeguards regarding personal information by adopting the General Data Privacy Regulation (“GDPR”). This regulation applies to all individuals and companies that deal with such information in any way. In Recombee’s case, when providing our services, we act as a so-called processor of personal data, which means that we receive data from “data controllers” (our clients) and analyse the data on our servers using a unique algorithm. ### 2 Does Recombee Process Personal Data? Yes, however, the data we process, have been pseudonymised. Pseudonymisation is process that effectively blocks us from identifying data subjects. Any identifiable elements of the individual bits of personal information are unreadable for us. The process of pseudonymisation is done on our clients’ side, so we can never learn who, in fact, is behind the pseudonymised data. This means that the database we work with is virtually free of personal information. ### 3 How Does Recombee Comply With the GDPR? Even though we do not process personal information, we have to comply with the GDPR in general. Although our database is pseudonymised, we have implemented safeguards and security that protect the integrity of the data subjects’ information we analyse for our clients. Our servers have both physical and software security measures that minimise the risk of unauthorised persons intercepting, deleting, reading or modifying the data we store. We also perform routine penetration tests. Our policy stands on the principle that only essential, well-selected and trained personnel are allowed to interact with the database; furthermore, those interactions are monitored and logged. When interacting with our clients, all communication is done using secure measures. We recognise that all data are precious and should be kept as confidential as possible. Our servers that contain the data we process are located in the European Union, which is considered as a secure destination. ### 4 Can an Individual Invoke His or Her Rights, Such as the Right to Opt-Out From Being Processed? Since we don’t know whose personal information we analyse, individuals can execute their right to opt-out from processing only with the data controller (our clients). ### 5 Does Recombee Send Any Information From the Database to Third Parties? No, the database of our clients’ information is the most important commodity we have. With our clients we typically sign a data protection agreement that specifies all rules and safeguards in order to comply with the GDPR. ### 6 How Should I, as a Controller, Address My Customers When I Want to Use Recombee’s Services? Based on your business model, you might have to obtain consent to use your clients’ personal information. Each business is very specific, so in some cases your existing consent is sufficient, in other cases you may need to them to update their consent. Some business models might fall under a different processing category (i.e. legitimate interests pursued by the controller). You may also be obliged to inform your data subjects of Recombee’s role in the data processing. If you are not sure how to resolve this matter, we suggest seeking professional legal advice. --- # Privacy Policy > Source: https://www.recombee.com/privacy-policy > For the complete site index, see [llms.txt](/llms.txt). To integrate Recombee, see the developer documentation at https://docs.recombee.com/llms.txt. # Recombee Privacy Policy This Privacy Policy explains how Recombee, s.r.o., with its principal place of business at Rybna 716/24, Stare Mesto, 110 00 Praha 1, Czech Republic (**“Recombee”**, **“we”**, **“us”** or **“our”**), collects, uses, stores, discloses and protects personal data in connection with Recombee’s websites, services, applications, communications and other related activities. This Privacy Policy applies to visitors of Recombee’s websites, persons requesting a demo or contacting Recombee, users of Recombee’s Admin User Interface, representatives of Subscribers, prospective customers, business partners and other persons whose personal data Recombee processes for its own purposes. This Privacy Policy does not replace any separate data processing agreement, order form, subscription agreement or other written agreement entered into between Recombee and a Subscriber. ## 1\. GENERAL TERMS Recombee provides recommendation services, related software tools, documentation, technical support, account management and other related services. In this Privacy Policy, such services are referred to as the **“Services”**. For the purposes of this Privacy Policy: 1. **“Subscriber”** means a person or legal entity using, subscribing to or evaluating the Services; 2. **“Admin User Interface”** means Recombee’s administrative interface through which Subscriber may manage the Services; 3. **“Personal Data”** means any information relating to an identified or identifiable natural person; 4. **“Subscriber Data”** means data uploaded, submitted, transmitted or otherwise provided by Subscriber to Recombee’s systems in connection with the Services; 5. **“Applicable Privacy Laws”** means the GDPR, UK GDPR, ePrivacy laws, the California Consumer Privacy Act as amended by the California Privacy Rights Act, and other privacy or data protection laws applicable to Recombee and Services. ## 2\. ROLE OF RECOMBEE Recombee may process Personal Data in different roles depending on the context. ### 2.1 Recombee as Controller Recombee acts as controller when it determines the purposes and means of processing Personal Data. This usually applies to Personal Data processed in connection with: 1. operation of Recombee’s websites; 2. registration and management of accounts; 3. demo requests, sales communications and business inquiries; 4. customer support and technical communications; 5. billing, accounting and payment administration; 6. marketing communications; 7. security, fraud prevention and abuse prevention; 8. protection of Recombee’s rights and legitimate interests; 9. improvement and analysis of Recombee’s websites and Services. ### 2.2 Recombee as Processor Where Recombee processes Personal Data contained in Subscriber Data on behalf of Subscriber and pursuant to Subscriber’s instructions, Recombee acts as processor or service provider, as applicable. Unless explicitly agreed otherwise in writing, Recombee’s standard Services are not intended for processing Personal Data contained in Subscriber Data. If Subscriber wishes to process Personal Data through the Services, Subscriber and Recombee must enter into a separate written agreement, including a data processing agreement where required by Applicable Privacy Laws. Subscriber is responsible for determining whether Subscriber Data contains Personal Data and whether Subscriber has a lawful basis and all required notices, consents and permissions for such processing. ## 3\. PERSONAL DATA COLLECTED BY RECOMBEE Recombee may collect and process the following categories of Personal Data. ### 3.1 Contact and Identification Data Recombee may process names, email addresses, telephone numbers, job titles, company names, country, business address and similar contact information. ### 3.2 Account Data Recombee may process account registration data, login information, organization details, role and permission settings, account preferences and information about collaborators. ### 3.3 Commercial and Billing Data Recombee may process subscription plan information, payment status, billing address, tax information, invoice data, payment method information and related business records. Payment card details may be processed by third party payment providers and are not intended to be stored directly by Recombee. ### 3.4 Technical and Usage Data Recombee may process IP addresses, browser type, operating system, device information, identifiers, log data, timestamps, pages visited, referring URLs, API activity, Admin User Interface activity and other data generated through use of Recombee’s websites and Services. ### 3.5 Communications Data Recombee may process messages, support requests, demo requests, emails, call notes, feedback and other communications exchanged with Recombee. ### 3.6 Marketing Data Recombee may process preferences relating to newsletters, product updates, events, surveys and marketing communications. ### 3.7 Cookies and Similar Technologies Recombee may use cookies, pixels, local storage, analytics tools and similar technologies to operate its websites, remember preferences, analyze usage, improve performance and support marketing activities. ## 4\. SOURCES OF PERSONAL DATA Recombee may collect Personal Data: 1. directly from the data subject; 2. from Subscriber or Subscriber’s authorized representatives; 3. from Subscriber’s collaborators; 4. through Recombee’s websites, Services and Admin User Interface; 5. from payment processors, analytics providers, marketing tools and other service providers; 6. from public sources, public registers, business directories, events, referrals and similar business sources. ## 5\. PURPOSES OF PROCESSING Recombee may process Personal Data for the following purposes: 1. to provide, operate and maintain the Services; 2. to create, administer and secure accounts and organizations; 3. to provide access to the Admin User Interface; 4. to respond to inquiries, demo requests and support requests; 5. to communicate with Subscriber and its representatives; 6. to issue invoices, process payments and maintain accounting records; 7. to monitor usage limits, service performance and system stability; 8. to prevent abuse, unauthorized access, fraud, security incidents and unlawful activity; 9. to maintain and improve Recombee’s websites, documentation and Services; 10. to send product updates, service notices and administrative messages; 11. to send marketing communications; 12. to comply with legal, tax, accounting and regulatory obligations; 13. to establish, exercise or defend legal claims; 14. to conduct business planning, reporting and internal administration. ## 6\. LEGAL BASES FOR PROCESSING Where EU, EEA or UK data protection laws apply, Recombee relies on the following legal bases: 1. **Performance of contract**, where processing is necessary to provide the Services, manage accounts, administer subscriptions or communicate about the contractual relationship; 2. **Legitimate interests**, where processing is necessary for Recombee’s business, security, service improvement, fraud prevention, customer support, business communications or internal administration, provided that such interests are not overridden by the rights and freedoms of the data subject; 3. **Compliance with legal obligations**, where processing is necessary for tax, accounting, corporate, regulatory or legal compliance; 4. **Consent**, where consent is required for specific cookies, marketing communications or other processing activities; 5. **Establishment, exercise or defense of legal claims**, where processing is necessary in connection with disputes, investigations or legal proceedings. ## 7\. COOKIES Recombee may use cookies and similar technologies for the following purposes: 1. strictly necessary operation of the websites and Services; 2. remembering user preferences; 3. measuring website performance and usage; 4. improving website functionality; 5. supporting marketing and analytics activities. Where required by law, Recombee will request consent before using non-essential cookies. A user may manage cookies through browser settings or, where available, Recombee’s cookie preference tools. Disabling certain cookies may affect the availability or functionality of Recombee’s websites or Services. ## 8\. DISCLOSURE OF PERSONAL DATA Recombee may disclose Personal Data to the following categories of recipients when they need to know them: 1. hosting, infrastructure and cloud service providers; 2. payment processors and billing service providers; 3. analytics, monitoring and security providers; 4. customer support and communication tools; 5. marketing and sales tools; 6. accounting, tax, legal and business advisors; 7. public authorities, courts, regulators or law enforcement bodies; 8. successors or potential successors in connection with a merger, acquisition, financing, restructuring or sale of assets. Recombee does not sell Personal Data in the ordinary meaning of the word. ## 9\. INTERNATIONAL TRANSFERS Recombee is established in the Czech Republic. Personal Data may be processed in the European Union, the United Kingdom, the United States and other countries where Recombee or its service providers operate. Where Personal Data is transferred outside the European Economic Area, the United Kingdom or Switzerland, Recombee shall use appropriate safeguards where required by Applicable Privacy Laws. Such safeguards may include the European Commission’s Standard Contractual Clauses, UK transfer mechanisms, adequacy decisions or other lawful transfer mechanisms. ## 10\. SECURITY Recombee shall use commercially reasonable technical and organizational measures designed to protect Personal Data against unauthorized access, accidental or unlawful destruction, loss, alteration, disclosure or misuse. Such measures may include access controls, authentication, encryption where appropriate, logging, monitoring, backup procedures, confidentiality obligations and security review of relevant systems. No method of transmission over the Internet or electronic storage is completely secure. Recombee therefore cannot guarantee absolute security. ## 11\. RETENTION Recombee retains Personal Data for as long as necessary for the purposes described in this Privacy Policy. The retention period depends on the nature of the Personal Data, the purpose of processing, the type of relationship with Recombee, legal obligations, limitation periods, accounting requirements and legitimate business needs. Account data is generally retained for the duration of the account or subscription and for a reasonable period thereafter. Billing and accounting records may be retained for the period required by tax and accounting laws. Marketing data is retained until the recipient unsubscribes or Recombee otherwise removes the data in accordance with applicable law. ## 12\. RIGHTS OF DATA SUBJECTS Subject to Applicable Privacy Laws, a data subject may have the following rights: 1. the right to access Personal Data; 2. the right to correct inaccurate or incomplete Personal Data; 3. the right to delete Personal Data; 4. the right to restrict processing; 5. the right to object to processing; 6. the right to data portability; 7. the right to withdraw consent where processing is based on consent; 8. the right to lodge a complaint with a supervisory authority Requests may be submitted to Recombee using the contact details stated in this Privacy Policy. Recombee may need to verify the identity of the requesting person before responding. Recombee may refuse or limit a request where permitted by law, including where the request is manifestly unfounded, excessive, affects the rights of others or conflicts with legal obligations. ## 13\. CALIFORNIA PRIVACY NOTICE This Section applies to California residents where the California Consumer Privacy Act, as amended by the California Privacy Rights Act, applies. ### 13.1 Categories of Personal Information Collected In the preceding twelve months, Recombee may have collected the following categories of personal information: 1. identifiers, such as name, email address, business address, IP address and account identifiers; 2. commercial information, such as subscription, billing and payment-related records; 3. Internet or other electronic network activity information, such as website usage, log data and interactions with Recombee’s Services; 4. professional or employment-related information, such as company name, job title and business contact details; 5. geolocation information at a general level, such as approximate location derived from IP address; 6. inferences drawn from the above information for business communication, service improvement and account administration. ### 13.2 Purposes of Collection Recombee collects personal information for the purposes described in Section 5 of this Privacy Policy, including providing the Services, administering accounts, responding to requests, processing payments, securing systems, improving services, marketing where permitted and complying with legal obligations. ### 13.3 Disclosure of Personal Information Recombee may disclose personal information to the categories of recipients described in Section 8 of this Privacy Policy. ### 13.4 Sale or Sharing of Personal Information Recombee does not sell personal information in the ordinary meaning of the word. Recombee does not knowingly sell or share personal information of persons under sixteen years of age. To the extent any use of cookies, analytics or advertising technologies is considered a “sale” or “sharing” under California law, Recombee will provide a method to opt out where required. ### 13.5 California Rights California residents may have the right to: 1. know what personal information Recombee collects, uses, discloses, sells or shares; 2. access personal information; 3. correct inaccurate personal information; 4. delete personal information; 5. opt out of sale or sharing of personal information; 6. limit the use and disclosure of sensitive personal information, where applicable; 7. not be discriminated against for exercising privacy rights. California residents may submit requests using the contact details in this Privacy Policy. Recombee may verify the request before responding. ## 14\. MARKETING COMMUNICATIONS Recombee may send marketing communications to business contacts, prospects, Subscribers and other persons where permitted by applicable law. Recipients may unsubscribe from marketing emails by using the unsubscribe link in the email or by contacting Recombee. Recombee may continue sending non-marketing communications, including service, security, billing and administrative messages. ## 15\. CHILDREN Recombee’s websites and Services are not directed to children. Recombee does not knowingly collect Personal Data from children under the age at which parental consent is required under applicable law. If Recombee becomes aware that it has collected Personal Data from a child without required consent, Recombee will take reasonable steps to delete such Personal Data. ## 16\. SUBSCRIBER RESPONSIBILITIES Subscriber shall ensure that its use of the Services complies with Applicable Privacy Laws. Subscriber is responsible for: 1. determining whether Subscriber Data contains Personal Data; 2. providing all notices required to end users or other data subjects; 3. obtaining all required consents or establishing another valid legal basis; 4. ensuring that Subscriber Data is accurate, lawful and appropriate for use with the Services; 5. entering into a separate data processing agreement with Recombee where required. Subscriber shall not upload, transmit or otherwise provide Personal Data to Recombee unless the Parties have entered into an appropriate written agreement allowing such processing. ## 17\. THIRD PARTY WEBSITES AND SERVICES Recombee’s websites, documentation or Services may contain links to third party websites, applications or services. Recombee is not responsible for the privacy practices, content or security of such third parties. Use of third party websites or services is governed by the terms and privacy policies of those third parties. ## 18\. CHANGES TO THIS PRIVACY POLICY Recombee may update this Privacy Policy from time to time. If Recombee makes material changes, Recombee may provide notice by posting the updated Privacy Policy on its website, sending an email, displaying a notice in the Admin User Interface or by other reasonable means. The updated Privacy Policy becomes effective on the date stated at the beginning of the Privacy Policy, unless a different date is specified. ## 19\. CONTACT Questions, requests or complaints concerning this Privacy Policy or Recombee’s processing of Personal Data may be sent to: * Data Protection Officer (DPO): Our DPO oversees our data protection strategy and compliance with privacy laws. For any concerns related to privacy practices, data protection rights, or personal data inquiries, the DPO can be contacted at: * Email: dpo@recombee.com * Postal Address: Recombee s.r.o., Václavské náměstí 846/1, 110 00 Nové Město, Praha, Czech Republic * Privacy Compliance Team: For general privacy inquiries or to report a privacy incident, our Privacy Compliance Team is available to assist. They can be reached at: * Email: data-privacy@recombee.com * Postal Address: Recombee, s.r.o., Rybná 716/24 110 00 Staré Město, Praha, Czech Republic Where required by Applicable Privacy Laws, Recombee may designate additional privacy contacts, representatives or data protection contacts.