When is a revenue cohort mature enough to compare?

A cohort is not mature merely because the reporting month has closed. It is mature for a particular measure only when enough of the customer lifecycle relevant to that measure has been observed.

Start with the decision horizon

For a 90-day payback limit, each compared cohort needs roughly 90 days of opportunity to produce revenue. A cohort acquired three weeks ago cannot be compared with a six-month-old cohort on cumulative revenue without either truncating both at three weeks or clearly modelling the younger cohort’s future.

Annual plans create a different problem. Their first payment makes early cumulative revenue look strong while the second renewal remains unobserved for a year. A monthly-plan comparison may reveal several renewal decisions in the same period.

Use an age-aligned table

Place cohort start month on rows and months-since-acquisition on columns. Compare down a single age column, not across the latest calendar total. Mark cells with partial periods and keep them out of the decision calculation.

Then inspect sample size. A technically mature cohort with 18 payers may still be too noisy for a channel-specific ceiling.

Separate facts from the tail

Observed cumulative revenue should stop at the cohort’s current age. Any continuation beyond that point is a forecast and needs a named method, such as a survival curve or a conservative retention floor. Show the observed and projected portions separately.

Maturity is therefore not one universal date. It is an explicit relationship between cohort age, plan term, sample size and the question being asked.