How do I work out why customers are actually churning instead of guessing?
Three passes. First, cohort the churn: group customers by when they signed up and by what they did, then see whether people leave at month three or month eighteen, because those are completely different problems. Second, capture a structured reason at cancellation and classify it consistently, so 'too expensive' and 'missing feature' become countable rather than anecdotal. Third, actually interview a dozen churned customers, because the form answer is almost always the polite version. Then split the list into avoidable and not, and only spend effort on the avoidable half. Most teams do the first pass and stop.
Go deeper
5 resources, 1 India-specific, 5 link-checked.
📄 Article
✓ Link checkedFreeIntermediate
Walks through building the cohort table itself, and separates acquisition cohorts from behavioural cohorts, which is the distinction that turns cohort analysis from a pretty chart into a diagnosis.
An actual interview script, twelve sections deep, designed to get past 'it was too expensive' to the real reason. This is the input your churn dashboard cannot give you.
The ten strategies are sequenced the way you would actually run a churn programme: instrument, classify reasons, segment by whether you can influence it, then offer something reason-specific.
Zoho on the two churn predictors that actually work early (usage depth and NPS) plus a worked cohort example. Written for teams without a data function.
The third pass in the short answer is the one people skip, and this is a practical guide to running it: who should make the call, what to ask, and how to get past the polite first reason. Short enough to watch before you book the interviews.
