Cohort Retention Calculator

How it is calculated

Cohort retention tracks a group of customers acquired in the same period and measures what share remain active at each subsequent period.

Retention at month n = customers from the cohort active in month n ÷ cohort size. The output is a curve rather than a single figure.

Active has to be defined for a non-subscription business, and the definition, purchased within the last N days, determines the shape of the curve.

The same thing with real figures

A cohort of 1,000 customers: 340 buy again within 90 days, 210 within 180, 150 within 270, 120 within a year.

The curve falls steeply and then flattens around 12%, which means roughly 120 of every 1,000 acquired become durable repeat customers.

Those 120 are worth far more than the average: if they buy 3.4 times a year at $58, that is $23,400 of annual contribution from 12% of the cohort.

The other 880 contribute their first order and little else, $28,072 of contribution, against $27,000 spent acquiring the whole cohort at $27 each.

The catch

Aggregate retention across all customers blends cohorts at different ages and moves when the growth rate changes rather than when retention does. A fast-growing business will always look worse on the blend.

The curve's early points are also noisy for small cohorts, and reading a trend from a single month's cohort of 200 customers is reading noise.

Applying it

Compare cohorts at the same age rather than at the same date. Month-three retention for the January cohort against month-three for the June cohort is the comparison that shows whether anything improved.

Then look for the flattening point, since the level at which the curve stabilises is what determines the long-term value of every cohort acquired.

What the shape tells you

A curve that flattens indicates a durable core, some proportion of customers who will keep buying indefinitely. The height of the flat portion is the single most important number in the analysis.

A curve that keeps declining has no core, and the business is renting customers rather than acquiring them. That is a fundamentally weaker position and it requires perpetual acquisition to stand still.

Improving the flattening level is worth far more than improving the early decline, because it applies for years rather than months. That usually means product and category fit rather than marketing, which is why retention curves are as much a product diagnostic as a marketing one.

Presenting the curves as a triangle chart: cohorts down the side, age across the top, makes improvement visible at a glance, since each column can be read down to compare cohorts at the same age.

That format is standard for good reason: it separates the two variables that a single retention figure blends together.

Annotating the chart with product launches, pricing changes and site releases makes it possible to attribute a shift in the curve to something specific rather than guessing at it later.

Tracking cumulative revenue per cohort alongside the retention percentage answers the commercial question directly, since a cohort with lower retention and higher spend can be worth more.

Comparing curves between the best and worst acquisition channels usually produces a wider spread than any product change, which makes channel selection a retention decision.

Reporting the twelve-month cohort value alongside the retention curve gives finance a figure they can plan against, where a percentage curve on its own rarely leaves the marketing team.