Create intermediate Works with: Any AI

Cohort Retention Analysis

Table output Has example

See whether users stay.

The Open buttons send the prompt text as currently shown below to that site only when you click. Fill in the placeholders first.

Analyze retention cohorts. Cohort data (signup period, users, active users in each following period): [COHORT_DATA]. Product changes during the period: [CHANGES].

Calculate: retention = active users / starting users for each cohort and period; average retention by period; and the difference between early and recent cohorts. Show the arithmetic for two cohorts.

Output format:
1. A retention table with periods as columns. Mark any cohort of fewer than 100 users as unreliable.
2. The shape of the curve (declines then flattens, or keeps falling) and what each shape implies.
3. One sentence stating whether newer cohorts retain better or worse than older ones, with the size of the difference in points.
4. The period with the biggest drop, with two hypotheses and the data that would check each.
5. Recommended actions and experiments (onboarding improvements, activation prompts).
6. Cautions about small cohorts, the definition of "active", and survivorship bias.
7. A one-paragraph summary for the team.

Fill in before use

Optional: type your details and the prompt above updates. Nothing is saved or sent.

Replace each bracketed placeholder with your own details, then paste into Any AI or another assistant.

January cohort: 400 signups. Active in month 1: 220 (55 percent). Month 2: 160 (40 percent). Month 3: 140 (35 percent).
March cohort (after onboarding change): 450 signups. Month 1: 288 (64 percent). Month 2: 212 (47 percent).
Retention = active / starting. 288 / 450 = 0.64; 212 / 450 = 0.471.
Difference at month 1: 64 - 55 = 9 points; at month 2: 47 - 40 = 7 points.
Curve shape: drops then levels, which suggests a core of retained users.
Hypothesis: the new setup checklist helped. Check: compare checklist completers with non-completers. Caution: only two cohorts and no control group, so treat as promising, not proven.

• retention