How retention cohorts work
You pick a start event (say, signup) and a return event (any meaningful action). Users are grouped into cohorts by when they first did the start event — day, week, or month. The analysis then plots what percentage of each cohort came back and did the return event in each later period, usually as a heatmap where each row is a cohort and each column is a period since they started.
Why it matters
Retention is the clearest signal of product-market fit. Acquisition can be bought, but retention has to be earned — if new users don’t come back, growth leaks out the bottom. Cohorts also let you tell whether a recent change helped: compare the retention curve of cohorts from before and after a launch.
Reading the curve
A retention curve that flattens — dropping at first, then holding steady — means you’ve found a core of users who stick. A curve that decays to zero means the product isn’t yet habitual. The goal is a flat (or “smiling”) curve, not 100% retention, which no product achieves.
A worked example
Take the March cohort — 1,000 signups. Week by week, the share returning to do a meaningful action reads 100% → 48% → 39% → 36% → 35% → 35%. It falls hard for three weeks, then flattens near 35%: about a third of every cohort becomes a durable, habitual user. That plateau — not the week-one number — is the product-market-fit signal. If the April cohort flattens at 40% instead, something you shipped is working.
How Pug does retention
Pug’s Retention insight builds a cohort heatmap from a start event and a return event, so you can see who comes back and when, filterable by any property. Because activity is tied to a person, retention reflects real users across devices, not cookies. The free cohort retention visualizer and retention calculator let you explore the math first.