Cohort
A cohort is a defined group of customers or users sharing a starting event, such as beginning their first paid subscription in January. Cohort analysis follows that group over time and compares groups at the same age since entry, keeping newer and older customer journeys separate.
Prepared with AIContent updated
Define entry and activity
Decide whether entry means sign-up, first payment or first active subscription. Define activity and the treatment of pauses, reactivation and multiple subscriptions per customer. The group must be reproducible using the same definition.
The same cohort after three months
Illustrative cohort of 500 customers with a consistent activity definition.
350 / 500 × 100 = 70%
No new customers enter the original cohort. These are repeated observations, not values to add together.
Retention with a fixed denominator
Customer retention at age X = active customers from the original cohort at X / customers in the starting cohort × 100. A cohort of 500 with 350 active after three months has 70% retention at month three. New customers outside that group must not be added to the numerator.
Revenue retention needs a separate definition. Net retention can include expansion and exceed 100%, whereas customer retention within a fixed original group cannot.
Compare the same elapsed time
Compare January and March at three months after each group’s entry, rather than on the same calendar date. A recent cohort has no observed full-year retention yet. Show unavailable observations as unknown rather than zero.
Patterns do not establish causes
Lower retention in a campaign cohort can relate to audience, price, contract length or product experience. Customers receiving personal onboarding may differ from self-service customers. Investigate differences and test hypotheses before attributing an effect to one initiative.
Use cohorts in planning
Combine observed retention with pricing and delivery costs to model CLV and future subscription value. State uncertainty, especially for small or immature groups, and separate projected months from observed results.
Questions and answers
How does a cohort differ from a segment?
A cohort is typically tied to a shared entry event and followed over time. A segment can be a current group such as a pricing plan. Segments can subdivide cohorts, but changes need clear treatment.
Why compare groups at the same age?
Otherwise the older group has had longer to experience cancellations and renewals. Compare the same number of completed observation periods.
Does better retention prove onboarding worked?
No. Other differences between the groups can explain the change. Use the analysis to develop hypotheses and investigate them with suitable comparisons.
