Cohort Retention Curve

One churn figure cannot tell these shapes apart. The curve shows when customers decide the product is not worth keeping, and whether a loyal core exists at all. A flattening curve is a clearer sign of Product-market fit than any single monthly percentage.

Comparing curves also tests changes. If cohorts that joined after a new onboarding sequence flatten higher than earlier ones, the sequence worked. Nothing else about their situation differs except the date.

• Group by signup period and plot the share still paying each month, never one blended line.
• Read the shape, not just the end point.
• Watch where the curve bends most sharply. That marks the moment to improve onboarding or follow-up.
• Keep enough months of history for the flattening to show. Judging too early mistakes a slow leak for a plateau.
• Split curves by channel or plan when the total looks calm but the business feels uncertain.
• Say what is counted, customers or revenue, and keep it fixed.

Take everyone who signed up in one month, count how many are still paying one month later, two months later and so on, and plot the percentages. Month zero is always 100 per cent. Each later point shows the share who remain. Repeat for each signup month and the curves can be laid over each other.

Three shapes are worth recognising. A curve that drops and then flattens means some customers leave early and the rest stay. A curve that keeps sliding means churn never stops. A curve that dips and then rises, which happens when revenue is measured, means remaining customers are spending more, as in negative churn.

• Averaging all cohorts into one line, which hides a recent decline.
• Judging a recent cohort that has only a few months of data against older ones.