Cohort Analysis

Longitudinal analysis comparing user behavior and retention across cohorts defined by acquisition timing or segment, measuring long-term engagement and the impact of product changes.

Process

Key Fields

Question it answers	Are newer users more or less engaged? Does a feature or change improve long-term retention? Which user segment has better lifetime value?
Participants & timing	Continuous from analytics · cohort maturation typically 90-180 days · 2-4 week analysis cycles
AI compatibility	AI calculates retention curves, detects significant differences between cohorts, and predicts lifetime value.
Output	Cohort retention curves, comparative analysis, driver identification, lifetime value estimates

Use when

Retention and engagement are critical to the business (subscription, freemium, apps)
Long-term impact of features or changes needs to be measured
User acquisition timing affects behavior (seasonal products)

Skip when

All users have a similar short lifecycle (purchased once, done)
Fewer than 6 months of historical data exist

Common Mistakes

Uncontrolled confounding variables

Cohort 1 vs. Cohort 2 shows lower retention, but marketing, pricing, and product all changed simultaneously. Multiple changes make it impossible to know the cause. Change one thing per cohort when possible.

Ignoring selection bias

A new cohort acquired through paid ads will have different baseline retention than one acquired through referrals. Match cohorts on user characteristics for valid comparison.

Drawing conclusions too early

A cohort less than 30 days old may look worse but stabilize later. Wait for cohorts to mature (90+ days) before declaring success or failure.