Case Study · Predictive Analytics

Customer Churn Prediction Model for the Automotive Industry

A model that predicts which customers are likely to stop using the dealership for service, and when, so retention teams can act before those customers move to independent garages.

← All case studies

Client

Dealer network with high post-warranty service attrition

Industry

Automotive

Delivers

Churn risk and timing for every customer

The problem

Aftersales is one of a dealership's most profitable lines, but many customers stop returning once their warranty ends. By the time a missed service shows up in a report, the customer has usually already moved to another garage.

Approach

Churn definition agreed with the business: no service visit within the expected interval plus a grace period, based on estimated mileage.
Features include service history and intervals, vehicle age, warranty end date, satisfaction scores, complaints, campaign responses and distance to the dealership.
A gradient-boosted classifier for 6-month churn risk, plus a survival model that estimates when churn is likely.
Explanations for each customer show why they are at risk, for example a price complaint, a long gap since the last visit or an expiring warranty.

From prediction to action

Customers are grouped into risk tiers, each with a retention playbook: service-plan offers, personalised reminders, loyalty rewards or a call from their service advisor.
Offers are targeted at customers the intervention is likely to win back, not those who would return anyway.
Risk scores and tasks appear in the CRM, and a retention dashboard tracks results.