Aggregate churn hides as much as it reveals. Cohort retention analysis groups customers by when they joined and tracks each group over time, so you can see whether retention is improving for newer cohorts or quietly getting worse. Here's how to build and read one.
Rows are join cohorts (e.g. "joined Jan 2026"). Columns are months since joining (Month 0, 1, 2…). Each cell is the percentage of that cohort's original revenue (or logos) still retained. The result is a triangle — older cohorts have more columns filled in.
Build two grids. Logo retention shows the percentage of original customers still active — it only ever declines. Revenue retention shows the percentage of original MRR retained — and because of expansion, a cohort's revenue retention can climb back above 100%. The gap between the two grids is your expansion story.
For cohort C at month N: retained revenue = MRR from customers in cohort C who are still active at month N. Cell value = that ÷ cohort C's Month 0 MRR. You need each customer tagged with a join cohort and their MRR tracked every month — the same customer-keyed model that powers MRR movement.
A blended churn number can look stable while newer cohorts churn harder, masked by loyal old customers. By the time aggregate churn moves, the problem is months old. Cohorts surface it early.
The grid is tedious to build by hand — it's a lot of conditional sums keyed by cohort and month-age. A template that already structures the cohort grid lets you paste your export and read the triangle instead of constructing it.
The template we recommend builds both logo and revenue cohort grids from your data, so you can see retention curves and expansion at a glance.
It groups customers by when they joined and tracks each group's retained revenue or logos over time, producing a triangle chart that reveals whether newer cohorts retain better or worse than older ones.
Logo retention tracks the share of original customers still active and only declines. Revenue retention tracks original MRR retained and can rise above 100% when expansion outpaces churn.
Aggregate churn can look stable while newer cohorts churn harder, hidden by loyal old customers. Cohorts surface a worsening trend months before it shows up in the blended number.
Page built 2026-06-14 from public, dated buying-intent signals. Updated as new signals land.