Cohort analysis groups people by a shared starting point — the month they first bought, signed up or were admitted — and then tracks each group separately over the following weeks or months. Instead of one blended retention figure, you get a row per cohort: how many of January’s new customers were still active after one month, two months, six months, and the same for February’s, March’s and so on.
It matters because averages hide change. An online retailer might see overall repeat-purchase rate holding steady at 30% while, underneath, customers from the new advertising channel repeat at 12% and older customers at 45%. The blended number looks fine until the older customers age out. Cohorts show the shift months earlier. The same lens works for patient follow-up rates in a clinic, or for how quickly new sales hires reach quota.
The usual output is a triangle-shaped table: cohorts down the side, months since start across the top, and a colour scale showing retention fading or holding. Read it diagonally to compare cohorts at the same age — that is the honest comparison. Reading down a column mixes cohorts of different ages and tells you less than it appears to.
In Klayara, a cohort view is a table or heatmap built from a start date and an activity date, with the retention percentage as a calculated field. The SaaS metrics template includes one, and formulas handle the date arithmetic.