Retail and e-commerce recommendation systems

Use the Yandex Cloud data platform, ML, and data visualization technology to create a recommendation system that can cope with any load and volume of data, without capital spending or expert maintenance.

Analyzing your users' behavior to predict the probability of them choosing specific products or services can help improve user experience, increase the average number of items and total amount of your average order, and reduce customer churn.

What makes Yandex Cloud the right choice?

Reliable, managed data platform
Build a data platform for your recommendation system based on reliable and fault-tolerant Yandex Cloud solutions. Integration with popular services and ML tools ensures a full cycle of data operations.
Professional support
Our engineering team ensures the platform’s stable operation, contributes to major open-source data storage projects like PostgreSQL and ClickHouse, and develops managed services already integrated into Yandex Mail, Yandex Taxi, and Yandex Drive.
Transparent pricing
Only pay for the resources you actually consume, and add more capacity as needed. Get detailed information about the amount of resources used and track your spending in the billing system.
Data visualization and monitoring
Make detailed charts and dashboards with key metrics using the convenient Yandex DataLens BI service. Monitor the performance of your recommendation system and your services' workload in real time.
Automatic scaling
Our flexible infrastructure automatically scales resources for your recommendation system and storage of user preference data, ensuring that all services run without lags, even under peak or unpredicted loads.
High availability and security
Host your data platform in three geographically distributed Yandex Cloud data centers, in compliance with the requirements of Russian Federal Law 152-FZ, GDPR, ISO, and PCI DSS.

Solution architecture

Use advanced technology and services to design your recomendation system’s architecture: ClickHouse, PostgreSQL, MySQL®, Spark, S3 Object Storage, and DataSphere with GPUs. We’ll help you find the solution best suited to your project needs.

Sample implementation of a recommendation system¹

1. Data about user activity from the frontend is stored in the operational DBMS. This includes purchases, click-throughs, likes, comments, and other information.

1. An implementation option. An optimal solution for your project should be selected individually. Our experts or partners can help you with this.
2. Apache Spark is a registered trademark of The Apache Software Foundation in the United States and/or other countries.
3. Greenplum is a registered trademark of VMware, Inc. in the United States and/or other countries.