What is churn, and why does cohort analysis matter?
Churn is the rate at which customers leave. Cohort analysis groups customers by when they joined and tracks each group separately — which is the only way to see whether churn is actually improving.
Why the headline churn rate misleads. Aggregate monthly churn mixes customers who joined last month with customers who joined three years ago, and those groups behave completely differently. Growing fast makes churn look artificially low, because recent joiners — who have not had time to leave — dominate the base. The rate then rises sharply when growth slows, which looks like a new problem and is actually an old one becoming visible.
What a cohort analysis shows. Take everyone who joined in January, and track what proportion remain after one month, two months, six months. Repeat for February, March, and so on. Now you can see:
Whether retention is improving over time — later cohorts retaining better than earlier ones is the clearest evidence that product changes are working.
Where in the lifecycle people leave. Heavy loss in the first month is an onboarding problem; steady loss later is a value problem. These need entirely different responses, and the aggregate number cannot distinguish them.
Whether the curve flattens, indicating a genuinely retained core rather than gradual decay to zero.
The distinctions that matter:
Customer churn vs revenue churn. Losing many small customers is very different from losing one large one, and revenue churn is usually the more important figure for a business with varied contract sizes.
Net revenue retention includes upgrades and expansion from remaining customers. It can exceed 100%, meaning existing customers generate more over time than is lost through churn — the hallmark of a strong subscription business and a far more informative metric than churn alone.
Voluntary vs involuntary churn. A meaningful share of subscription losses are failed payments — expired cards, insufficient funds — not decisions. Dunning processes recover a substantial proportion, and this is among the cheapest improvements available.
Why small differences compound. Monthly churn of 5% versus 3% is the difference between an average lifetime of twenty months and thirty-three.
General information, not business advice.