Cohort analysis reveals retention, churn, and lifetime value patterns. Here’s how to set up and interpret cohorts.
What a Cohort Analysis Shows
A cohort is a group of users who share a characteristic — usually when they signed up or first converted. Cohort analysis tracks each group’s behavior over time: how many come back in week 2, week 4, week 12. It reveals retention patterns, lifetime value, and whether your product is getting stickier or leakier over quarters. It’s the single most important long-term analytics view for subscription and product businesses.
Building Cohorts in GA4
GA4 Explore → Cohort Exploration. Inclusion: first_visit event. Return criteria: any event (or specific — add_to_cart, purchase). Granularity: daily, weekly, or monthly. Metric: total users, conversions, revenue. The resulting table shows percentage of users returning in each subsequent period. Week 1 cohort, tracked across weeks 1–12 — you see retention decay over time.
Reading Retention Decay
Typical patterns: high initial drop (60–80% of users don’t return after day 1), then stabilization. Healthy consumer apps: 20–30% week-1 retention; 10–15% at week 4; 5–10% by week 12. B2B SaaS: 70%+ week-1, 50%+ at month 1, 30–40% at month 12. Compare your cohorts to industry benchmarks — drastically lower retention signals product-market fit issues, not marketing problems.
Improving Retention
Better onboarding (clear first-action guidance) often lifts week-1 retention 20–30%. Re-engagement emails at drop-off points (day 3, day 14, day 30) recapture users. Product improvements targeting the specific reason users leave (found in user interviews, not just analytics). Cohort analysis reveals WHICH week people leave; user research reveals WHY. Both are needed to move retention curves.
Cohort by Acquisition Source
Break cohorts down by acquisition channel: organic search vs paid social vs referral. Different channels produce different retention patterns. Users from organic search typically retain 30–50% better than paid social. This matters for budget allocation — channels with bad retention are hidden losses that look fine in conversion dashboards. Always cohort-analyze by source monthly.
Calculating LTV From Cohorts
Lifetime value = sum of revenue per cohort over their lifetime. Extend with predictions: if 12-month retention is 20% and monthly revenue is $50, LTV is approximately ($50 × 12) × retention curve area ≈ $300–500. Use LTV to set customer acquisition cost (CAC) limits — a 3:1 LTV-to-CAC ratio is a common target. Cohort analysis is how you get real LTV numbers rather than marketing guesses.
Need help with analytics?
Get a free audit of your analytics setup. We’ll show you exactly where the opportunities are.
Get Free Audit →