A large DIY and home improvement retailer wanted to see how many of its inactive customers could be re-engaged with the brand and needed a solution to predict future customer churn to advise their retention strategy.


We designed a machine learning model to predict customer behaviour leveraging over 15 engineered variables. The model identified 475,000 out of nearly 1 million customers at high risk of churnWe created an experimentally designed campaign with 5 different sets of incentives to win-back churned customers and supported our client in the deployment of these campaigns. 

The Outcomes

Our Churn model prediction achieved accuracy of over 91%. Over £4 million additional revenue generated from previously inactive customers after the first month. £13.16 million revenue generated after the fifth month. 76, 264 customers shifted from churn back to active. 53% of re-activated customers were still active after six months.

Sign up to our newsletter

Want to make life easier by staying on top of market trends?