“What’s a good retention rate?” is one of the most-asked and least-answerable questions in analytics. A daily-use messaging app and a tax tool people touch once a year both have “retention,” and a number that’s excellent for one is a disaster for the other. So the honest answer comes in two parts: a few rough heuristics by model, and — far more useful — how to read your own curve.
Why there’s no universal number
Retention is the share of a starting group still active some period later, and its inverse is churn. What counts as good depends on natural usage frequency (is this a daily habit or an occasional tool?), how you define “active,” the period you measure, and your audience. Comparing your weekly app retention to someone’s annual subscription renewal rate is meaningless. Treat published benchmarks as loose context, never targets.
Rough ranges by model
With that caveat firmly in place, some widely-observed heuristics:
- Consumer subscription (SaaS, media). Healthy products often retain a majority of users month over month; best-in-class can show negative revenue churn as existing customers expand.
- Consumer mobile apps. Notoriously leaky — it’s common to lose the large majority of new users within the first weeks, so even modest day-30 retention can be respectable.
- B2B / enterprise. Logo retention is usually high (long contracts, switching costs); the sharper signal is seat and usage retention inside accounts.
- E-commerce. Measured as repeat-purchase rate rather than continuous use; a meaningful minority of customers returning to buy again is a strong sign.
Notice how wide these are. That’s the point — the ranges overlap and vary so much that your own trend is the only benchmark you can act on.
The curve beats the number
A single blended retention figure hides the story. Plot retention by cohort instead — group users by when they joined and track each group over time — and a shape appears. Retention almost always falls fastest early, then either:
- Flattens to a plateau — a stable core keeps using the product. That plateau, even if low, is the clearest signal of product–market fit, because those users found durable value.
- Decays toward zero — no plateau means few users stick, and no amount of acquisition will fix a bucket with no bottom.
A rising plateau across newer cohorts means your product is getting stickier — usually a better thing to celebrate than any absolute number.
How to measure your own
Start with the simple math: the free retention & churn calculator turns start, new, and end counts into a retention rate, churn rate, and implied customer lifetime (lifetime ≈ 1 ÷ churn). Then move to the shape with the cohort retention visualizer, which renders a cohort grid you can read at a glance. In Pug, retention cohorts are built in over your real events, filterable by any profile trait, so you can compare retention by plan, source, or cohort directly.
The bottom line
Stop hunting for the magic percentage. Define “active” honestly, measure retention by cohort, look for a flattening curve, and work to lift the plateau over time. A product that retains a committed core beats one with a flashier headline number and nothing underneath it. For related targets, see what a good conversion rate is.