Cohort Analysis

Cohort analysis is how you understand whether your product is actually getting better over time. Instead of looking at aggregate metrics (total users, overall revenue), you group users by when they signed up and track each cohort separately. **Why cohort analysis matters:** Aggregate metrics lie. If your total MAU (monthly active users) is growing, that looks good. But if you dig into cohorts, you might find that every cohort churns at 80% after 3 months—you're just masking churn with new signups. That's a leaky bucket. Cohort analysis reveals this. You can see: - Are newer cohorts more engaged than older ones? (Product improvements working) - Do cohorts retain better over time? (You're learning what makes users stick) - Which features drive retention in successful cohorts? (Double down on those) **How to do cohort analysis:** 1. **Define your cohort**: Usually by signup month (Jan 2026 cohort, Feb 2026 cohort, etc.). You can also cohort by acquisition channel, feature adoption, or customer segment. 2. **Choose a metric**: Retention, engagement, revenue per user, feature usage—whatever matters for your business. 3. **Track over time**: Measure the metric for each cohort at 1 week, 1 month, 3 months, 6 months, etc. 4. **Compare cohorts**: Are newer cohorts performing better than older ones? That's progress. **Example:** You launch a SaaS product. Jan cohort: 100 signups, 40% active after 30 days. Feb cohort (after onboarding improvements): 100 signups, 60% active after 30 days. Your product is getting better—new users are more likely to stick around. Without cohort analysis, you'd just see "80 active users in Feb, 120 active in March" and think you're growing. With cohort analysis, you see that Feb cohort retained better than Jan, proving your onboarding changes worked. **Common cohort metrics:** - Retention curves (% of cohort still active over time) - Revenue per cohort (LTV by signup month) - Feature adoption (which cohorts adopted key features faster) - Churn rate by cohort (are you reducing churn over time?) **Tools:** Most analytics platforms support cohort analysis: Mixpanel, Amplitude, PostHog, Google Analytics 4. If you're early stage, a simple spreadsheet works—group users by signup month, calculate retention, compare. The most important question cohort analysis answers: Are we getting better at keeping users? If yes, you're on a path to product-market fit. If no, you're growing but not improving.

Netflix uses cohort analysis to see if content changes improve retention. A Jan cohort might churn at 5%/month. If Feb cohort (with new originals) churns at 3%/month, content is working. Spotify cohorts by acquisition channel to see which sources bring sticky users.