Cohort analysis for product is the method of grouping users by a shared characteristic or experience at a specific time — most commonly signup date or feature adoption date — and tracking their behavior over time to reveal whether product changes are improving or degrading retention and engagement.
Cohort analysis is the most honest lens in your analytics stack. You can improve aggregate metrics with acquisition spend while your retention is deteriorating. You can ship features that look positive in aggregate but are actually making things worse for new users. Cohort analysis surfaces the truth that aggregate metrics hide.
This guide shows product managers how to build, read, and act on cohort analysis.
What Cohort Analysis Reveals
H3: The Three Questions Cohort Analysis Answers
- Is retention improving over time? (New cohorts should retain better than old cohorts if the product is improving)
- Does a product change help or hurt? (Compare cohorts before and after the change)
- Are acquired users from different channels retaining differently? (Acquisition quality analysis)
Without cohort analysis, you cannot answer whether you have product-market fit. The retention curve is the single most reliable PMF signal.
Reading the Retention Curve
H3: The Three Retention Curve Patterns
Pattern 1 — Smiling Curve (Product-Market Fit)
100%
|\
| \_____
| \____
| \___________ (flattens above 0%)
+----------------------------------→ Time
The curve flattens above zero. Some percentage of users keep coming back indefinitely. You have a retained user base. This is product-market fit.
Pattern 2 — Cliff (No PMF)
100%
|\
| \
| \
| \______________________ (approaches 0%)
+----------------------------------→ Time
All users eventually churn. Nobody finds long-term value. This is the pre-PMF state.
Pattern 3 — Rising Retention (PMF in Progress)
Cohort 1: drops to 5%
Cohort 2: drops to 8%
Cohort 3: drops to 12%
Cohort 4: drops to 18%
Each successive cohort retains better. The product is improving. This is the signal to keep investing.
According to Lenny Rachitsky's writing on retention analysis, the rising retention pattern across cohorts is the most important signal in early-stage product development — it means the product is learning faster than customers are churning, and if continued, will produce a retained base.
Types of Cohort Analysis
H3: Acquisition Cohort Analysis
Group users by when they signed up (week of signup, month of signup). Track their retention over time. Compare across cohorts.
What to look for:
- Are newer cohorts retaining better than older ones? (Product improving)
- Are there cohorts with unusually high or low retention? (What changed that week?)
- Does retention differ by acquisition channel? (Traffic quality signal)
H3: Behavioral Cohort Analysis
Group users by whether they completed a specific action (feature adoption, onboarding completion, first purchase). Compare retention between groups.
Classic use case: Compare Day-30 retention of users who completed onboarding vs. users who did not. If completed-onboarding users retain at 45% and non-completed users retain at 8%, you've quantified the value of fixing onboarding.
According to Shreyas Doshi on Lenny's Podcast, behavioral cohort analysis is the most actionable form of retention analysis for product teams — it directly answers the question of which product actions are correlated with long-term retention and should therefore be treated as activation milestones.
H3: Feature Cohort Analysis
Group users by whether they adopted a specific feature within their first 30 days. Compare long-term retention.
This analysis answers: which features, when adopted early, predict retention? These are your "aha moment" features — the behaviors you should drive in onboarding.
Building a Cohort Analysis in Practice
H3: Data Requirements
- User identifier: Consistent user ID across sessions
- Signup timestamp: When the user first appeared
- Event timestamps: When each user performs each action
- Cohort definition: The grouping criteria (week, channel, feature adoption)
- Return event definition: What counts as a "return" for retention calculation
H3: Time Intervals to Track
For most products:
- Day 1 retention (did they come back the next day?)
- Day 7 retention (did they come back within a week?)
- Day 30 retention (are they monthly active?)
- Day 90 retention (do they have a habit?)
- Day 365 retention (are they a long-term user?)
For SaaS with monthly billing cycles:
- Month 1, Month 3, Month 6, Month 12 retention
According to Annie Pearl on Lenny's Podcast discussing retention analysis, the Day-7 retention rate is the single most predictive early signal of long-term retention — teams that fix Day-7 retention see the impact in Day-30 and Day-90 numbers two to three months later.
Acting on Cohort Analysis
H3: When Retention Is Declining Across Cohorts
A new cohort with lower retention than the previous cohort is a regression signal. Investigate immediately:
- What changed in the product between cohorts?
- Did acquisition channel mix change (lower quality traffic)?
- Did onboarding change?
H3: When a Specific Cohort Underperforms
An isolated cohort with lower-than-expected retention suggests a product event:
- Did a major bug ship around that cohort's signup date?
- Did a pricing change affect that cohort's willingness to pay?
- Did a marketing campaign bring in an atypical user segment?
FAQ
Q: What is cohort analysis in product management? A: A method of grouping users by a shared characteristic at a specific time — most commonly signup date or feature adoption date — and tracking their behavior over time to reveal retention trends and the impact of product changes.
Q: What is a retention curve in cohort analysis? A: A chart showing the percentage of users from a cohort who return at each time interval. A curve that flattens above zero indicates product-market fit. A curve that trends to zero indicates all users eventually churn.
Q: How do you know if you have product-market fit from cohort analysis? A: When your retention curve flattens above zero — meaning some percentage of users keep returning indefinitely — you have a retained user base and the signal of product-market fit.
Q: What is behavioral cohort analysis? A: Grouping users by whether they completed a specific action and comparing their long-term retention. Used to identify which product actions are activation milestones — the behaviors that predict long-term retention.
Q: What retention intervals should product managers track? A: Day 1, Day 7, Day 30, and Day 90 for consumer products. Month 1, Month 3, Month 6, and Month 12 for SaaS products with monthly billing cycles.
HowTo: Run Cohort Analysis for a Product
- Define your cohort grouping criteria — most commonly weekly signup cohorts or channel acquisition cohorts for retention analysis
- Define your return event — the specific action that counts as a user returning and delivering or consuming value
- Track retention at standard intervals: Day 1, Day 7, Day 30, and Day 90 for consumer or monthly intervals for SaaS
- Plot retention curves for each cohort and look for whether newer cohorts retain better than older ones as the primary PMF signal
- Run behavioral cohort analysis comparing retention of users who completed specific actions versus those who did not to identify activation milestones
- Investigate any cohort that underperforms its neighbors to identify product regressions, acquisition quality changes, or pricing change impacts