Saturday 14:45–15:20 in Auditorium

Winning (Some) Churners Back

Ferran Elias

Audience level:
Intermediate

Description

In our presentation we will discuss our experience at Booking.com creating a marketing channel using incentive offers to bring churners back. The main elements of our strategy are: 1) A model predicting the probability of churning for each customer in a given timeframe. 2) Experimentation framework to incentivize customers based on the churn model.

Abstract

What is the right strategy to contact customers who have churned and are not shopping with your platform any more? In our presentation we will discuss our experience at Booking.com creating a marketing channel using incentive offers to bring churners back. The main elements of our strategy are: 1) A model predicting the probability of churning for each customer in a given timeframe. 2) Experimentation framework to incentivize customers based on the churn model. This has lead to a ROI of up to 4, which is 8 times larger than the ROI of experiments with no targeting strategy. We will also discuss the steps we are developing at the moment: 1) A framework for putting incentive models full-on to continuously monitor and steer ROI. 2) A Thompson Sampling strategy to efficiently experiment with different rewards offered to churners. 3) Moving from a fixed-timeframe churn model to a Survival model, which facilitates optimization of personalized contact time.

Subscribe to Receive PyData Updates

Tickets

Get Now