Retention curves: what they tell you and what they hide

A retention curve is the most honest graph in your analytics stack. It doesn’t smooth over bad months or hide seasonal effects. It shows you, cohort by cohort, how many of the people who started with you are still here.

But it hides things too — and those gaps are where most teams go wrong.

What the curve tells you

The shape of the curve matters more than any single number. A curve that drops steeply in the first two weeks and then flattens is a different problem than one that declines slowly and steadily. The steep early drop is an onboarding problem. The slow decline is a value problem.

The inflection point — where the curve starts to flatten — tells you your “aha moment” window. Users who make it past that point tend to stay. Users who churn before it never found what they came for.

What it hides

Aggregate retention curves hide segment differences that are enormous. A curve that looks acceptable at the company level might be masking a strong segment (enterprise, annual, referral) carrying a weak one (SMB, monthly, paid ad). When you split the curve by acquisition channel, plan type, or company size, you often find that one segment retains at 2x the rate of another.

That gap is a strategy. The high-retention segment is who your product is actually built for — even if it’s not who you’re marketing to.

The right question to ask

Don’t ask “how do we improve retention?” Ask “which cohort retains well, and what’s different about how they found us, onboarded, or use the product?” That question is answerable. The first one isn’t.

Retention is an outcome. The inputs are acquisition quality, onboarding completion, and depth of feature adoption. Fix those — in that order — and the curve improves without ever looking directly at it.