
How it works
A cohort is a group of customers who share a starting point, most often the month they first paid. Cohort analysis follows each group separately, month after month, and records how many are still paying.
A single blended churn number can hide a lot. If a business is growing fast, new customers, who tend to leave sooner, are mixed in with older ones who rarely leave. Splitting by start month shows whether recent customers behave differently from earlier ones.
Example
A subscription business tracks three monthly cohorts and counts customers still paying after three months.
| Cohort | Customers at start | Still paying after 3 months | Retention |
|---|---|---|---|
| January | 100 | 84 | 84% |
| February | 120 | 96 | 80% |
| March | 150 | 105 | 70% |
Retention after three months is falling: 84% to 80% to 70%. The blended figure across all three groups is (84 + 96 + 105) ÷ (100 + 120 + 150) = 285 ÷ 370 = 77%. That single number would look acceptable and hide the decline. The numbers are illustrative, not market data.
The same method works for revenue: group customers by start month and track the recurring revenue each group still pays after three, six and twelve months. Comparing groups of the same age is what makes the comparison fair.
How buyers read a retention table
Here is a made-up table of customers still paying in each month after they joined.
| Cohort | Start | Month 1 | Month 2 | Month 3 | Month 6 |
|---|---|---|---|---|---|
| January | 100 | 92 | 88 | 84 | 75 |
| February | 120 | 108 | 101 | 96 | 84 |
| March | 150 | 120 | 110 | 105 | not yet |
Read each row from left to right to see how a group fades, and each column from top to bottom to compare groups of the same age. Here the January group keeps 75% by month 6 (75 ÷ 100). At month 3 the groups keep 84%, 80% (96 ÷ 120) and 70% (105 ÷ 150), so newer cohorts are leaving faster. A buyer reads three things: whether the curve flattens (a core of loyal customers), whether newer rows sit below older ones, and how many months of history each row has.
Why it matters in a valuation
A buyer is paying for customers who stay. Cohorts show whether the customer base is getting stickier or leakier, and whether last year's customers are still around. A steady or improving pattern supports the case that revenue will continue. A worsening pattern, as in the example, is a reason for a buyer to lower the price or ask more questions.
VALUERAQ's SaaS valuation tool takes one monthly churn figure and does not read cohort data, so cohort tables are best prepared separately and shared with a buyer during due diligence.
Frequently asked questions
How many customers do I need for cohort analysis?
Enough that one customer leaving does not move the percentage much. With a handful of customers per cohort, group by quarter instead of by month.
Is cohort analysis only for SaaS?
No. Any business with repeat customers can use it, including stores, memberships and newsletters. Group customers by first purchase or first sign-up.
What is the difference between cohort retention and churn?
Churn is a single rate for one period. A cohort table shows how each start group behaves over time, which can reveal changes a blended churn figure hides.
Related terms
Retention rate is the share of customers kept. Churn is the share lost. LTV estimates what a customer is worth over their time as a customer. Net revenue retention tracks revenue from the same customers.