Cohort Analysis

What Cohort Analysis actually means

A cohort analysis groups customers by a shared starting point, usually their acquisition month, then follows each group forward. It exists because blended averages hide almost everything interesting. A company with worsening retention can post a flat blended number for a year, simply because older healthy cohorts are large enough to mask what the newer ones are doing.

Worked example

Blended monthly retention holds steady at 92 percent. Split by cohort, customers acquired eighteen months ago retained 95 percent at month six, while customers acquired three months ago are retaining 84 percent at the same point in their life. Retention is deteriorating sharply and the blended number would not have shown it for another two quarters.

Why the board cares

Cohort views are how a board distinguishes a real trend from a mix shift. They are also the only credible way to validate the lifespan assumption inside an LTV model, since a cohort curve can be extrapolated while a blended average cannot. Boards that have been burned by an optimistic LTV tend to ask for cohorts by default afterward.

Common mistakes

• Cohorting by calendar month while comparing groups at different ages, which mixes time and maturity.
• Reading cohorts too early, when the sample is too small to separate signal from noise.
• Cohorting only by date when the more useful split is acquisition channel or first product purchased.

Related terms

• Customer Lifetime Value (LTV)
• Churn Rate
• Net Revenue Retention (NRR)
• Average Revenue Per User (ARPU)

Back to the full marketing glossary