Why customers leave — and why a discount rarely brings them back

Subscription businesses tend to respond to rising churn the same way: a retention team, a script and a discount for anyone who tries to cancel. It feels like action. Often, it is mostly cost.

One offer for everyone

A single retention discount assumes that customers leave for one reason — price. In a recent project for a consumer subscription service, we tested that assumption. Price was the main reason for leaving in only two of six customer segments. For the others, the reasons were content, service quality or simply low usage. A discount cannot fix any of those. It only lowers revenue from customers who would have stayed anyway, or briefly delays those who will leave regardless.

Combine what customers say with what they do

Neither surveys nor behavioural data are enough on their own:

• Surveys explain motivations — why people value the service and why they leave. In this project we surveyed 4,000 current and former customers.
• Behavioural data shows what actually happens: usage patterns, billing issues, service contacts in the weeks before cancellation.

Together they produce segments that are both measurable in the data and meaningful to the people designing retention offers.

Segment by reason, not just by value

Classic value-based segmentation tells you who matters most. Reason-based segmentation tells you what to do about it. The six segments we built combined customer needs, behaviour and value, and each came with a recommended treatment — a better content package for one group, proactive service for another, a price offer only where price was the real issue.

Predict early enough to act

Most churn signals appear weeks before a customer cancels. A model that predicts risk 60 days ahead gives the retention team time to act before the decision is made — not when the customer is already on the phone asking to leave. The model is only useful if its scores reach the people who act on them, so we integrate it directly into the CRM.

Test offers before rolling them out

Retention offers should be tested like any other marketing decision. Small, controlled tests on selected segments show which offers retain customers and which simply give away margin. Measure retained revenue minus the cost of the offer, not just the number of customers who stayed.

The result

With segment-specific offers and early risk scores, churn in the targeted segments fell by 18% over six months, and discount spend dropped because offers went only to customers for whom price was the real issue.

Where to start

• Ask former customers why they left — and analyse their last 60 days of behaviour.
• Group customers by reason for leaving, not only by value.
• Give the retention team a different offer for each group, and test before scaling.