See more projects

The customer chose to protect their confidential information

A Video Recommendation Engine

Media & Entertainment
Customer rating
This ranking reflects our expertise and success
A Video Recommendation Engine

About the client

Andersen was approached by a large digital cinema company that needed to enhance their online video content recommendation system. One of Europe’s leading digital cinema providers, the customer has a significant market share coupled with a loyal client base.

[object Object] on the map

Project overview

The recommendation system case study examined here is focused on using ML algorithms and Data Analytics to generate and propose personalized recommendations for each end-user. Before our company was contracted, the customer had been offering various recommendation options like trending movies, new releases, and personalized picks based on the user’s viewing history and preferences. However, that was not enough, as the company was facing stiff competition from other European digital cinema providers, including Netflix, Amazon Prime Video, Disney+, and HBO Max. The management decided to differentiate their content by offering superior client service and innovative features like a highly accurate and advanced recommendation engine.

Duration12 months
SAS campaign


    • 1 Project Manager
    • 2 Data Engineers
    • 2 Data Scientists
    • 1 Business Analyst
    • 1 Solution Architect
    • 3 Front-end Developers
    • 3 Back-end Developers
    • 2 QA Engineers
    • 2 UX Designers
Visual 1

Project details

Business challenge

The customer runs a large digital cinema platform that contains a vast collection of various movies, TV series, and short clips. After careful analysis, it was identified that most of their users were insufficiently engaged in comparison with competitor platforms and didn't spend much time consuming content. The customer decided to address this situation with multiple actions, including the improvement of the content library and enhancement of its recommendation engine. Due to a lack of expertise in the AI/ML area, the company contacted our team to optimize its video recommendation system and hence provide more relevant "watch next" proposals to end-users, encouraging them to spend more time in the system.

The digital cinema company was struggling with keeping end-users engaged so that they spent more time consuming content via its platform. This was a key business challenge that needed to be addressed to improve client satisfaction and revenue generation rates.

The customer found that their content library didn't meet the needs and preferences of their target audience, which was contributing to low user engagement and retention. This was a specific area of focus that could be tackled with better content acquisition and production strategies.


Based on initial analysis, our specialists identified that their existing recommendation engine was too simple and couldn't effectively handle complex segmentation and customer behavior patterns. Andersen's target was to develop a truly cutting-edge content-based recommendation system example. The following things were planned:

  • Complete 360-degree user profiles with all activities and info;
  • Around 15,000 dynamic client micro-segments instead of the 5-10 that the platform initially had;
  • A/B testing to determine which configuration of NBOs (next best offers) would convert best;
  • Better personalization;
  • Advanced behavioral analysis features to help the platform better understand user preferences and behaviors.

Market trends

Nowadays, there are multiple providers of video content in the market, which offer viewing content on either a monthly subscription basis or a pay-per-view one. Revenue across the entire video streaming app industry reached $72.2 billion in 2021 and is projected to reach $115 billion by 2026. The market is highly competitive, and each of the providers strives to win a larger audience by generating unique and popular content. That is why an effective video recommendation system is crucial for any video platform, as it is the primary method to empower users to discover new content and stay engaged.

Over-the-top (OTT) platforms are becoming increasingly popular, with more consumers cutting the cord on traditional cable TV and opting for streaming services. This trend is expected to continue, and OTT revenues are projected to reach $159 billion by 2025.

The number of streaming platforms continues to grow, and consumers are becoming more selective about the content they choose to watch. This has led to a growing demand for personalized content recommendations tailored to individual viewing habits and preferences.

Visual 2

While the major streaming platforms like Netflix, Amazon Prime Video, and Disney+ continue to dominate the market, there is also a growing trend towards niche content platforms that cater to specific interests and demographics. Examples include Crunchyroll for anime fans, Shudder for horror enthusiasts, and BroadwayHD for theater lovers.

Many streaming platforms are now integrating with social media platforms like Facebook, Twitter, and Instagram to make it easier for users to share and discover new content. This can include personalized content recommendations based on the user's social media activity or social media sharing features that enable users to share their favorite shows or movies with their friends and followers.

In order to stay competitive, many streaming platforms are investing heavily in creating original content production – movies, TV shows, and documentaries – as well as acquiring exclusive distribution rights covering popular titles. The goal is to build a loyal fan base and differentiate themselves from competitors by offering unique and engaging content.


The NBO recommendation system of the video platform did the following:

  • Split the client base into 5-10 segments that were manually configured by the marketing team;
  • Provided NBOs based on a black box logic relying on simple rules inputted manually;
  • Due to the absence of A/B capabilities, testing changes was risky and required lots of time to collect and analyze data manually. Thus, the previous recommendation approach had quite limited effectiveness and often displayed irrelevant suggestions;
  • The video platform had a vast collection of movies, TV shows, and short clips across various genres and languages. However, due to the sheer volume of content, it used to be difficult for users to discover new titles matching their interests;
  • The platform had a user-friendly interface that allowed users to browse content by genre, language, and other filters. However, the previous video recommendation system often suggested titles that were not relevant to the user's interests, which was frustrating for users;
  • The video platform offered a subscription-based model, allowing users to access unlimited content for a monthly fee. Alternatively, users could rent or purchase individual titles on a pay-per-view basis;
  • The platform was accessible via a variety of devices, including desktop PCs, smartphones, tablets, and smart TVs. However, some users reported technical issues when trying to access content via certain devices and platforms.

It was decided that the target solution of this movie recommendation system project would include:

  • Comprehensive client profiles: Development of a 360-degree view of client profiles that would include all user activity and information, such as viewing history, ratings, preferences, and demographics;
  • Advanced segmentation: Creation of around 15,000 dynamic client micro-segments using ML algorithms to group users based on similar interests, behaviors, and other relevant factors;
  • Personalized NBOs: Usage of advanced AI/ML algorithms to generate personalized NBO recommendations for each client based on their unique preferences and behavior patterns;
  • A/B testing: Incorporation of A/B testing capabilities to evaluate the effectiveness of different NBO configurations and optimize the recommendation engine for maximum engagement and revenue;
  • Continuous improvement: Continuous refinement and improvement of the recommendation engine by collecting and analyzing client feedback, as well as leveraging new technologies and data sources as they become available.
Visual 3

Main components of the platform:

  • Data sources. All possible data concerning the user: previous views, preferences, demographics, behavior, metadata, social, engagement level, etc.;
  • API gateway. This provides interfaces for data sources to proactively stream data into our platform and for our platform to extract data from data sources;
  • Data integration (ETL) pipeline. Data sets are mapped, identified, extracted, and transformed. This ensures high data quality and good data preparation – this step is very important;
  • Data storage/data lake/client 360 profile. All data is stored securely for easy access and used to construct metrics concerning each client;
  • AI/ML engine. The movie recommendation system using machine learning models of the predictive class typifies clients according to patterns found in their data and generates NBOs for the right channel, at the best moment and in the appropriate context;
  • Integration with the video library to fetch all possible offers;
  • Reporting and visualization. Reports and automated trend-tracking capabilities enable continuous AI/ML model improvement;
  • Delivery channels. The customer integrated our recommendation channels into their existing ones: the website, mobile app, and Android TV app. The user's response/interaction with the offer is captured and processed by the back-end part to enrich ML models.
Scheme 1


Previously, offers were assigned using, effectively, a black box system, and we do not know the exact logic for the test/control assignment or the offer assignment. This situation is quite typical, as the offers can be managed by different departments and teams and change over time.

Because of that, there's no guarantee that this assignment was randomized – therefore, the historical data can have arbitrary biases.

Andersen's team addressed the above challenge with four steps:

  • Implemented a procedure for bias control and correction that would address potential biases in the historical data;
  • Built a core model per product/channel that would estimate the view probability conditioned on different types of content, which further would help to determine the most efficient NBO strategy for each client, video, and channel;
  • Implemented the content selection logic based on the probability;
  • Established the evaluation procedure to assess the quality of the model based on the bias-corrected data.
Scheme 2

AI/ML models

Supervised learning

ML algorithms predict the probability of a customer’s action by learning patterns based on historical data. In the context of video streaming platforms, this could involve analyzing a customer's viewing history, preferences, ratings, and other factors to predict the probability of them taking a particular action, such as watching a specific movie or TV show.

An individual prediction model is trained for each of the products/services to be recommended to clients.

The resulting model scores are then calibrated across content to generate comparable probability scores in order to rank the videos for each client.

For the first version, each model applies the same classification algorithm (e.g., gradient boost) trained on a standardized dataset (client demographics, interaction with products and customer service, product purchases, etc.).

Later, each individual model is improved (the parameters are tuned).

Reinforcement learning

This approach can be particularly useful in situations where the environment is constantly changing, as it allows the system to adapt and optimize its recommendations in real time based on client feedback and other signals. However, reinforcement learning can also be more complex and computationally intensive than supervised learning and may require more data and resources to be effective.


The primary channels for presenting recommended videos are mobile app, website, and TV app, where suggested videos are displayed on the home screen. Based on collected data, it was proposed to use additional ways to communicate with the clients: SMS, mobile app notifications, and emails. Considering the history of interactions and views, we prepared a target audience selection for each channel that was supposed to achieve the highest results, without sending too many annoying ads to all the client base. Regular campaigns were set up to send weekly digests with new or targeted videos across all communication channels.

A/B Testing

In order to ensure our theoretical model corresponds to real-life scenarios, it is important to do actual testing on clients. During A/B testing, the platform could display either legacy recommendations or those from our system. Even preliminary results showed our system had higher conversion rates in comparison with the legacy system. After the model enhancement stage, we were able to significantly increase the performance of the recommended section.

Scheme 3

By using A/B testing in the following way, we were able to make data-driven decisions about which features to add to the platform and how to optimize them for the best user experience:

First, we needed to develop a hypothesis about the changes we wanted to test. For example, we might have hypothesized that changing the algorithm used to generate recommendations would result in higher engagement among users.

Next, we needed to create two variations of the platform to test against each other. One variation would use the new recommendation algorithm, while the other would use the legacy algorithm.

Then, we needed to choose a representative sample of users to test the variations on. This sample should've been large enough to provide statistically significant results but not too large to negatively impact the user experience.

The sample of users should've been randomly divided into two groups, with one group seeing recommendations by the new algorithm and the other group seeing recommendations by the legacy algorithm.

It was necessary to choose a metric to measure the success of each variation. This could be anything from click-through rates to user engagement time.

The variations were then launched and tested by our specialists on the sample of users. Data on the chosen metric was collected and analyzed.

Finally, we analyzed the data to determine if there was a statistically significant difference between the two variations. If the new algorithm performed better, we would implement it on the platform. If not, we would continue testing until we found an improvement.

Project results

While working on this modern recommendation system example, our team delivered a framework for building the NBO models. This framework includes bias correction, propensity modeling, uplift estimation, and model evaluation.

For each of these steps, there are many methods and techniques that provide different trade-offs in terms of complexity, the restrictiveness of assumptions, and the quality of results. Andersen evaluated all these options and came up with the solution.

Scheme 4

This implementation has significantly improved client segmentation practices and made it possible to show more relevant recommended offers. The following results have been achieved:

Scheme 5

Let's talk about your IT needs

What happens next?

An expert contacts you after having analyzed your requirements;

If needed, we sign an NDA to ensure the highest privacy level;

We submit a comprehensive project proposal with estimates, timelines, CVs, etc.

Customers who trust us


Let's talk about your IT needs