Reading your book as cohorts
Total customer count hides what is happening underneath. Grouping customers by start month makes retention legible.
Priya Raman//4 min read/Analytics
Group every customer by the month they started, then count how many remain at month 3, 6, 12, and 18. One tab in a spreadsheet, refreshed monthly, and the shape of the business becomes visible.
A cohort that holds 88 percent at month 12 and 74 percent at month 18 is telling you where to spend the next quarter of effort. Averages across the whole book hide that gap entirely.
Compare cohorts by acquisition source once you have six months of data. Channels that look equally priced often separate by 15 points at month 12.
Build the table in one afternoon
Group every customer by the month they started, then count how many remain at month 3, 6, 12, and 18. One tab in a spreadsheet, refreshed monthly, and the shape of the business becomes visible. Total customer count hides all of this, since a book that grows five percent while losing twelve percent underneath looks identical to a healthy one from the outside.
| Start month | Customers | Month 3 | Month 6 | Month 12 | Month 18 |
|---|---|---|---|---|---|
| March | 142 | 96% | 93% | 88% | 74% |
| April | 198 | 95% | 91% | 85% | 71% |
| May | 233 | 92% | 86% | 79% | Pending |
Read down the columns rather than across the rows. The May cohort losing eight percent by month three while March lost four is a signal about who was sold in May and how they were serviced, and it arrives nine months earlier than the annual retention number would deliver it.
Split by acquisition source once you have six months
Channels that look equally priced often separate by fifteen points at month twelve. Paid search customers with an urgent pest problem behave differently from referral customers who bought on a neighbor's recommendation, and the difference is worth more than the difference in cost per lead.
- Tag every customer with the source at the time of sale, since it cannot be reconstructed later.
- Compare month 12 retention by source before comparing cost per customer.
- Multiply retention by gross profit to get a first year value per source.
- Move budget toward the source with the best value, which is rarely the cheapest one.
The cheapest customer to acquire and the most valuable customer to own are usually different people.
Use cohorts to test one change at a time
Cohorts turn operating changes into experiments you can read. Move the price increase off the anniversary in April, and the April cohort tells you at month 14 whether it worked. Add an annual summary letter in June and the June cohort answers the same way. One change per month keeps the reading clean.
Put two lines on the monthly page
Most owners need only two numbers from the table each month, retention at month 12 for the cohort that just reached it, and the month 3 number for the cohort that just started. The first tells you about the year behind you and the second gives you an early read on the year ahead.
A book that holds 88 percent at month 12 and 74 percent at month 18 is telling you where to spend the next quarter of effort, and it is telling you now rather than at the end of the fiscal year.
