Increasing purchase frequency and average order value through AI personalization

YVES ROCHER

Objective:

Increasing purchase frequency while maintaining and growing average order value

Tools:

Programmatic

Challenge

A highly switchable category: users can build their care routine from different brands and switch brands within the same product type.
User consumption is non-linear

Objectives

Increase purchase frequency
Maintain and increase average order value

Solution

We moved from classic retargeting, where we simply catch up with the user, to managing the next purchase. We started predicting the next purchase and steering the consumption sequence.

How it worked

In the campaign data architecture, we used two data sources:

In-app user behavior (AppMetrica)
Client CRM data

MMP data was then joined with client CRM data in our DMP, where a large number of audience segments were built. The segments were then passed to Yandex via API for campaign launches.

As a result, every user became predictableSegmentation was based on
2 parameters:

Repeat purchase probability

User retention and timely reminders about which products to buy to continue the care routine.

Next Best Product

Basket expansion through recommending a product the user is highly likely to buy next.

About AI agents

For this campaign, we connected our own cascade of AI agents to build the creative matrix, because the number of segments and messages was very large. The output was a creative matrix with segment sets and creative recommendations for every segment and product: a personal offer, copy and visuals.