Acquisition is visible, retention is endured. A month when thirty new customers arrive gets celebrated. A month when thirty old ones leave quietly triggers nothing, although the net result is identical and the second costs far more to repair.
It is also the one number that cannot be dressed up. You can buy growth, you cannot buy retention.
Definition
Retention is the share of customers still active after a given delay from signup. It mirrors Churn rate exactly: 92 percent monthly retention equals 8 percent churn.
The useful measurement is never done in aggregate but by cohort: the group who joined in January, tracked month by month, separately from February's. A global average mixes loyal veterans with fragile newcomers and hides exactly what you are looking for.
The shape of the curve matters more than its height
| Curve shape | What it means |
|---|---|
| Falls then flattens | A core found its value: good sign |
| Falls without ever flattening | Nobody stays: the product is not needed |
| Ticks back up | Customers returning: rare and excellent |
| Collapses in month one then holds | An onboarding problem, not a product one |
The last row is by far the most common and the most fixable. A product whose curve flattens at 40 percent after a brutal initial drop does not have a value problem, it has a starting problem.
A curve that flattens, even low, is the most reliable Product-market fit signal you have. The height of the plateau can be worked on afterwards, its existence cannot be decreed.
The levers, in order of return
First result, as fast as possible
Most departures happen before the customer has obtained anything at all. Identify the action matching your product's first real benefit, then organise the whole welcome around it. Nothing else matters in the first seven days.
Make the value visible
A customer using your product without noticing will cancel the next time they review spending. A monthly recap of what they saved or produced costs little and changes the conversation at renewal.
Handle failed payments
Some departures are not departures at all, they are expired cards. Dunning recovers a meaningful fraction of what wrongly appears as voluntary churn.
Frequently asked questions
What retention should I target?
The absolute number depends too much on sector for a single benchmark to mean anything. Look at two things instead: does your curve flatten, and does the resulting LTV (customer lifetime value) cover your CAC (customer acquisition cost) with enough margin.
Retention or acquisition, where should effort go?
Retention first, nearly always, because it multiplies everything else. Better retention raises LTV, which permits a higher CAC, which unlocks acquisition channels previously out of reach. The reverse is not true.
How do I measure retention without subscriptions?
Through repeat purchase. On a one-off model, track the share of customers buying a second time within twelve months. Same reasoning, different frequency.
How many customers before the measure is reliable?
A cohort of thirty already gives a readable trend. Below that, a single departure moves the curve several points and you are interpreting noise.