Personalization on all levels of the marketing funnel


Launching personalization in four stages

Data collection
Forming an algorithm
Design and UI

Data Collection

History Data

Uploading data from the resource

Resource Pixel

A small program that collects information on users and their behavior

Public Data

Social Networks and DMP

Algorithm Formation

When creating a model, we take 3 important parameters into consideration

Content Similarity

We combine content according to general attributes and divide them into groups

User Similarity

We identify users with similar characteristics and behavior patterns

Unique Rules

Each project has its own requirements that we will tailor the recommendations for

Design and UI

What we recommend isn’t the only thing that affects the efficiency of the recommendations, but also the way it looks as well. Before the launch, we participate in the design of the banners to be used in the advertising campaigns and we provide consultations on UX issues in order to plan the recommendation sections on the site or in a message if it’s opened in an app.

The best offers for each user at the sign-in

Recommendations on the home page

Banner template with personalized content

Recommendations in social network tickers


The uniqueness of the model requires a high recommendation quality grade and for that reason we conduct a series of tests

Testing by a small group of experts

We identify mistakes in the way the algorithm is working

Server Traffic

We verify the stability of the servers’ operation at different traffic levels

Comparing the Results

We take a look at how the recommendation results are different from non-personalized selection

Find out more in our video

Still have questions?

Read documents