Cohort analysis follows groups of users who share a starting event, usually the week or month they first visited or signed up, and compares their behavior over the following periods. It separates the effect of when someone joined from the effect of how long they have been around. For SaaS teams it is the standard way to read retention, activation and the long-term value of different acquisition sources.
Reading a cohort table
A cohort table has one row per cohort and one column per period since the start. Each cell shows the share of that cohort still active, or another metric, in that period. Reading down a column compares cohorts at the same age; reading across a row shows one cohort's decay.
Cohort Week 0 Week 1 Week 2 Week 3
Jan 05 100% 42% 35% 31%
Jan 12 100% 45% 37% -
Jan 19 100% 48% - -
(illustrative)Cohorts in marketing analytics
GA4's cohort exploration builds these tables from first-visit or other events. Product analytics tools offer richer versions tied to in-app actions. For SEO, a useful cut is cohorts by first acquisition channel: do users first acquired through organic search retain or upgrade at different rates from paid users?
Pitfalls
Recent cohorts have fewer periods of data, so do not compare their totals with older cohorts. Small cohorts swing a lot from noise. And identity matters: web analytics cohorts based on cookies lose users when cookies expire, so product databases with logged-in IDs give better retention cohorts.
Example
Did onboarding changes help?
Suppose Inkwhistle, a note-taking app for teams, ships a new onboarding checklist in March. Cohorts from after the change show higher week-four activity than earlier cohorts at the same age. Because the comparison is at equal age, the team can attribute the change to timing rather than to users simply having had longer to engage.
Common mistakes
Comparing cohorts of different ages
Always compare at the same period since start.
Using cookie-based users for long retention
Cookie loss looks like churn. Use logged-in identity where possible.
Questions
A segment is any subset of users. A cohort is a segment defined by a shared start time or event, followed over time.
