Mosaic: Look-alike Modeling for Audience Expansion and Enrichment

Client Company

Initially, the Mosaic was developed for a UAE DMP and MENA region

Illustration with white abstract figure on orange background

Solution Description

A system that provides clients’ data enrichment: identification of people like the target audience and customer’s user segments augmentation for more personalized ad targeting.

Technology background

Primary functionality includes:

  • Reconstruction of users’ gender and age based on their routes and points of interest - shops, sports clubs, beauty salons, etc
  • Application of data processing and AI algorithms to historical data to create 2D and 3D modeling. Laying the results over a map for better visualization
  • Complementation of the solution with tools analyzing and providing additional data: devices people use, browsers, languages they speak, their interests, and others
  • Identification of behaviour and patterns that are most common among people taking specific actions (like making a purchase) for more granular ad targeting and higher engagement rates

Application of machine learning model enabled achievement of results with +/- 80% accuracy.

Results and Current Status

The system has shown its efficiency and is currently used by Inventale’s European and UK clients, as well as clients in the MENA region.