What Is Cohort Analysis?
Cohort Analysis groups customers by a shared characteristic — most commonly their sign-up or acquisition date — and tracks their behavior over time. Instead of looking at aggregate metrics that blend new and old customers, cohort analysis reveals whether each generation of customers performs better or worse than the last.
How It Works
A typical cohort retention table:
| Cohort | Month 0 | Month 1 | Month 2 | Month 3 | Month 6 | Month 12 |
|---|---|---|---|---|---|---|
| Jan 2025 | 100% | 78% | 65% | 60% | 48% | 35% |
| Apr 2025 | 100% | 82% | 72% | 68% | 55% | — |
| Jul 2025 | 100% | 85% | 78% | 74% | — | — |
This shows improving retention: each newer cohort retains better than older ones, suggesting the product is getting better.
What to Look For
- Improving cohorts — newer groups retain better (product is improving)
- Flattening curves — retention stabilizes after a point (you found your core users)
- Deteriorating cohorts — newer groups retain worse (market saturation or quality decline)
- Revenue cohorts — same analysis but tracking MRR instead of logos
Types of Cohort Analysis
| Type | Groups By | Reveals |
|---|---|---|
| Acquisition cohort | Sign-up date | Retention trends over time |
| Behavioral cohort | First action taken | Which actions predict retention |
| Revenue cohort | Initial plan/spend | Lifetime value by entry point |
Cohort Analysis in AI-Run Companies
AI-run companies benefit doubly from cohort analysis. First, as the AI running the product improves, cohort analysis directly measures whether those improvements translate to better customer outcomes. Second, AI agents can perform cohort analysis automatically, identifying patterns humans would miss.
On EvolC, investors use cohort data to determine if an AI-run company is in a virtuous cycle — where the AI gets better, retention improves, revenue compounds, and more data makes the AI even better.