What is retention curve?

A retention curve plots what share of a starting group, called a cohort, is still active after one day, one week, one month and so on. It shows when and how quickly people leave. It matters because it reveals whether a product keeps people or loses them, and the shape says more than any single average.

How it works

For each period, retention = users still active ÷ users who started. The curve usually falls fastest at the start, when people try the product and leave, and then it either keeps falling or flattens. A curve that flattens at a level above zero means a core group stays; a curve that keeps falling toward zero means everyone eventually leaves.

The area under the curve is the average number of active periods per user, which links retention to revenue and lifetime value. Raising early retention usually has the largest effect, because every later point is a share of those who stayed.

Example

Say 1,000 players install a game on day one. On day 1 of play, 40% return, so 400 remain. By day 7, 20% remain, so 200. By day 30, 10%, so 100, and by day 90 the curve has flattened at 8%, so 80. The product keeps a core of 80 per 1,000 installs. If a better tutorial lifts day-1 return to 50%, and the later shape scales along with it, day-90 might reach about 10%. These are illustrative numbers.

How to model it in Machinations

Use a Pool of active users fed by a Source at the start, and a Drain whose rate falls over time, set by a Register that follows a decay formula with a floor. A state connection links the Pool to the Drain. Chart the Pool as a share of the starting cohort. Play and change the early drop-off Lever to see how the curve shifts.