Hong Life CalculatorsUnderstanding the cohort retention calculator — concept and principle
Cohort retention tracks one group of users who started in the same period, for example everyone who signed up in January, and shows what share of them are still active one, three and six months later. Product managers, app developers, growth marketers and investors use it because it separates real staying power from the noise of new sign-ups that can hide a leaky product in overall totals.
How it is calculated
Retention in month n = active users in month n ÷ users in month 0 × 100
Drop from month 1 to month 6 = (1 − active in month 6 ÷ active in month 1) × 100
- Users in month 0: the size of the cohort when it started.
- Active in month 1, 3, 6: how many of those same users were active in each later month. Do not add users who joined later.
The principle
Retention is measured against a fixed starting group, so every later figure is a fraction of the same denominator. Plotting these fractions gives a retention curve. A typical curve drops steeply in the first month, when users who were only curious leave, and then flattens as a core of regular users remains. The month 1 → 6 drop isolates that second phase: it uses month 1 as the base, showing how well the product keeps people who got past the first experience.
Worked example
- Month-1 retention: 420 ÷ 1,000 = 42%.
- Month-3 retention: 260 ÷ 1,000 = 26%.
- Month-6 retention: 180 ÷ 1,000 = 18%.
- Drop from month 1 to 6: (1 − 180 ÷ 420) × 100 = 57.14%.
How to read the result
The headline number is month-6 retention, 18% here. More important than any single value is the shape: a curve that keeps falling toward zero suggests weak product–market fit, while one that levels off signals a loyal base. Compare cohorts: if newer cohorts retain better at the same age, recent product changes are working. Benchmarks vary greatly between games, social apps, e-commerce and business software, so compare mainly with your own history and close competitors.
Common mistakes
- Dividing by the current total user base instead of the cohort's starting size.
- Changing the definition of "active" (login, purchase, key action) between cohorts.
- Reading month-6 retention for a cohort that has not yet reached month 6.
Frequently asked questions
What is the difference between classic and rolling retention?
Classic retention counts users active in exactly that month. Rolling (unbounded) retention counts users active in that month or any later one, so it is always equal or higher.
Why did my month-3 retention rise above month 1?
Users can return after a quiet month. With classic retention, later months can occasionally exceed earlier ones.
How big should a cohort be?
Larger cohorts give more stable figures. With only a few dozen users, a few people change the percentage by several points.
Sources: Farris, Bendle, Pfeifer & Reibstein, Marketing Metrics (Pearson) — customer retention · Fader, P. S. & Hardie, B. G. S. (2007), How to Project Customer Retention, Journal of Interactive Marketing