A retention cohort answers one question: of the customers who started in a given month, how many are still coming back?
The table
Each row is a group of customers who made their first visit in the same month (a cohort). Each column is months since that first visit.
| Joined | Customers | Month 1 | Month 2 | Month 3 |
|---|---|---|---|---|
| August | 120 | 38% | 29% | 25% |
| September | 140 | 41% | 33% | |
| October | 160 | 45% |
Read it like this: of the 120 customers who first came in August, 38% came back the month after, 29% the month after that.
What to look for
- Reading down a column: is month-1 retention improving? Here it went 38% → 41% → 45%. Something is working.
- Reading across a row: where does the curve flatten? If it levels off around 25%, that is your loyal core.
- A sudden drop in one cohort: what happened that month? A price change, a staff change, a new competitor?
Why it beats a single retention number
A single "retention rate" mixes old and new customers. Cohorts separate them, so you can see whether changes you made actually helped the customers who arrived after them.