• Homepage
  • Blog
  • A new level of recommendation accuracy - AI Recommendations module

A new level of recommendation accuracy - AI Recommendations module

7 min read
Cover photo with  recommendations module

In order to supply good recommendations, it is important to ensure accuracy in building the recommendation model, its updates and the proper calibration of details. The more details we are able to define, the better our recommendations will respond to customer needs.

The Synerise team understands these needs, that's why uses their latest achievements in the field of artificial intelligence, to improve proprietary recommendation engine based on AI algorithm - Cleora. The innovation and reliability of Synerise solutions has been emphasized many times as evidenced by a number awards including won the competition organized by Rakuten Institute of Technology and 2nd place in the AI / ML competition organized by Booking.com. 

Graphic presenting using object document

Based on the improved AI Recommendations module, clients can upload to Synerise any number of stores and items catalogs, which allows you to menage of all recommendations within one profile. The update gives clients a possibility to independently create and configure every model of recommendation campaigns and measure their effects thanks to a simple and intuitive interface. The system itself will take care of data hygiene, parameter tuning and, if needed in the event of a change in the recommendation model, relearning the algorithm.

All in your hands - self-service recommendation generation

The new recommendation update is a collection of big changes that offer huge opportunities. One of the most important functionalities of recommendation module is the possibility of self-service configuration of recommendation powered by our proprietary AI engine, Cleora. As part of this change, customers have the opportunity to adjust recommendations to the needs of their company and to edit them as needed.

Screenshot  from Synerise with cart recommendation settings


Configuring new recommendations is extremely simple and consists of several steps:

  • Sending the product data file (feed)
  • Selecting product attributes for the recommendation model
  • Starting training on the selected recommendation model
  • Set up a recommendation model for the source

The basic requirement to start using recommendation model is to provide an appropriate product data set (product feed), configure the AI engine and meet the minimum requirements for interaction and event data, which can be found in this article. After appropriate data integration, recommendation model gives the ability to fully monitor the effects of recommendations, which in turn allows users to react quickly and adapt the model to new situations, changes in products or consumer behavior.

New opportunities - Multiple product catalogs under one customer profile

New AI Recommendation module allows full control over data within one profile and more accurate measurement of the effects of their activities, but also expands the possibilities of customers who have more than one product, shop or sales market. In practice, this means that within one profile we can implement many of stores and item catalogs and based of them implement different content, video,  policies for prices, discounts, types of promotions and methods of displaying recommendations.

Screenshot presenting the type of recommendation

This solution introduces a new way of managing product catalogs and the recommendation models within a single customer profile. From now on, having multiple stores or several separate products, the customer can independently create a recommendation model for each of them separately.  Within one profile we can be used with various type of recommendations for similar items, cross-selling, cart recommendations or others based on visual similarity, etc.

Where can you use AI Recommendations module?

Currently, Synerise allows users to use recommendations in such channels as:

  • on webstites and in stores
  • in emails
  • in web push notifications
  • in mobile push notifications
  • in applications

What will you gain from AI recommendation module?

Thanks to AI recommendations, you can increase conversion at any stage of the customer journey, from the home page, category page, through the cart, to post-purchase campaigns. Recommendations are made in real time for recognized, unrecognized and new customers based on different types of interactions.

The new version of recommendations allows you to personalize the customer experience at many points of contact, within various communication channels, through mobile applications, websites, mobile notifications and more. The Synerise solution allows you to adjust the recommendation result to your business needs, which is possible thanks to self-service configuration, advanced filtering and sorting of data within the product catalog. On the other hand, the possibility of uploading multiple items databases within one profile makes the use of the new recommendations tailored to the needs of each business. When using recommendations powered by an AI engine - Cleore, you do not have to worry about manual data processing, catching errors or tuning model parameters - the platform does it for you and leaves the most important thing in the customer's hands, the possibility of full control over the entire recommendation creation process.


Social proof shows that other people's choices affect our choices. The very slogan "others have also seen" or "others bought" motivates us to browse products chosen by other consumers, compare ourselves to them and finally buy products that become more attractive to us. Recommendations are therefore not only a way to increase the number of transactions, but also to build a brand and relationships with customers. There is a place for modern recommendations based on the AI Cleora engine, which allow you to independently configure the recommendation model and adapt to the needs of various businesses. It is extremely important to be able to create and manage multiple types of content under one account in order to build more consistent brand communication, better analyze customer needs and respond to them.