Goal
Turn raw user activity into a cohort retention table and explain what it says about how long users stay.
Inputs
- Product and what counts as active: [e.g. a meal planning app; active means opened a plan]
- Cohort period: [weekly or monthly]
- Number of periods to track: [e.g. 6 months]
- Data: [paste rows with user ID, signup date and activity dates, or pre-counted cohorts]
- Changes shipped during this time: [e.g. new onboarding in March]
Steps
- Assign each user to the cohort of their signup period.
- For each cohort, count users active in period 0, 1, 2 and onward.
- Divide by cohort size to get retention percentages.
- Compare cohorts before and after each listed change.
Output format
- Cohort table: rows are cohorts with their size, columns are periods, cells are retention percentages.
- Average curve: the weighted average retention for each period.
- Drop-off point: the period with the steepest loss and its size.
- Findings: four numbered points, each with a figure and a comparison.
- Next questions: three things to investigate, such as which actions retained users share.
Check before answering
- Period 0 is 100 percent for every cohort by definition.
- Cells for periods that have not happened yet are left blank, not zero.
- Averages are weighted by cohort size, not a simple mean of percentages.
- Cohorts under 50 users are marked as low sample.
- Any link to a product change is stated as possible, not proven.
- The definition of active is repeated above the table.






