Cohort analysis is a method of grouping users by a shared starting characteristic — most commonly the time period in which they first signed up or made their first purchase — and then tracking how that group’s behavior evolves over time. Rather than calculating metrics across the entire user base at a single point in time, cohort analysis calculates metrics for specific groups as they age, which makes it possible to distinguish meaningful behavioral trends from changes in the composition of the user base.
The clearest illustration of why this distinction matters: a product that is growing quickly will have a user base increasingly dominated by recent signups. If recent cohorts retain at lower rates than older cohorts (perhaps because the product has expanded into new markets and is acquiring users with lower fit), the average retention rate across all users could appear stable even while the product is experiencing a retention crisis in its newer cohorts. Cohort analysis surfaces this pattern where aggregate analysis hides it.
How Cohort Analysis Works
A cohort analysis table typically has cohorts in rows (Week 1 signups, Week 2 signups, and so on) and time periods in columns (Week 0, Week 1, Week 2, Month 1, Month 2, and so on). Each cell shows what percentage of that cohort was still active — or had made a repeat purchase, or had reached some other outcome — at the corresponding time period after their initial event.
Reading cohort analysis tables: the diagonal from top-left to bottom-right represents the same calendar period for all cohorts. A column reading vertically compares cohorts at the same age (all at Week 1 after signup, for example). A row reading horizontally shows how a single cohort’s behavior changed as they aged. These three readings serve different analytical purposes: the diagonal identifies current-period changes affecting all cohorts simultaneously, the vertical comparison shows whether the product is improving or degrading for new users over time, and the horizontal reading characterizes a specific cohort’s retention curve.
Retention Cohort Analysis
Retention cohort analysis answers: what percentage of users who signed up in a given period are still active at each subsequent period? A typical retention curve shows a steep initial drop — many users try a product once and do not return — followed by a flattening as the engaged users separate from the disengaged ones. The point at which the retention curve flattens is called the “retained” or “habituated” baseline, and its height is one of the most important metrics in product analytics: a product whose retention curve flattens above 40% has strong product-market fit in its core user segment; one that never flattens (continually declining toward zero) has not yet found users for whom the product is genuinely valuable.
Comparing retention cohorts over time identifies whether product changes are improving retention for new users. If the Month 3 retention rate for users who signed up in Q3 is consistently higher than the Month 3 rate for users who signed up in Q1 and Q2, that improvement could indicate that product changes, onboarding improvements, or changes in acquisition channels are producing a better-fit user population. Cohort analysis is the analytical method that makes these comparisons interpretable.
Revenue and LTV Cohort Analysis
Cohort analysis applies to revenue as directly as it applies to user retention. A revenue cohort analysis tracks how much revenue each signup cohort produces in aggregate over time. The cumulative revenue per customer from each cohort, plotted against time since signup, produces the customer lifetime value curve — which shows how long it takes the average customer from a given cohort to repay their acquisition cost and begin generating margin. If the LTV curve for the most recent cohort is below the curve for older cohorts at the same point in their lifecycle, the product is acquiring less valuable customers and the economics of the business are deteriorating.
Cohort Analysis for Marketing Attribution
The most powerful application of cohort analysis for marketing teams is comparing cohorts defined not just by signup period but by acquisition source. A cohort of customers who signed up from paid search in a given month versus a cohort of customers who signed up from organic content in the same month can be tracked through the same retention and revenue analysis. If the organic cohort retains at 45% after 6 months while the paid search cohort retains at 22%, that information directly affects where to allocate marketing budget — not because paid search is inherently lower quality, but because the specific campaigns or keywords driving paid search acquisition in that period are attracting users with lower product fit.
This analysis requires capturing acquisition source at the individual user level and connecting it to the downstream data store (product analytics or CRM) where retention and revenue data lives. UTM parameters captured at first touch and passed through to signup are the most common mechanism. The analytical work is then grouping users by source-cohort (paid-search-Q3, organic-Q3, referral-Q3) and running parallel retention analyses on each group.