Cohort Retention Table
Paste cohort sizes and active counts per period and get the retention table in percent with heat shading, weighted and simple averages, and data checks.
A retention matrix (active ÷ size) with blanks where a cohort is too young, weighted (pooled) and simple averages per period, and flags for rising counts and unreadable lines.
Example: Four monthly cohorts of 1,000 / 1,200 / 900 / 800 users retain 40–44 % in period 1: pooled average 41.8 %, simple mean 41.9 %; only January has reached period 4 (22 %).
Each row a cohort,
each column a period.
How to enter the data, how the percentages and averages are formed, and the checks.
Entering cohorts
One line per cohort: a label, the cohort size, then the number still active after period 1, 2, 3 … (or the percentages, if you switch the mode). Comma, tab or space separated; a cohort that has not reached a period simply has fewer numbers and shows a blank, not a zero.
Percentages and averages
Retention = active ÷ cohort size. The weighted average per period pools users — Σ active ÷ Σ size over the cohorts that have reached that period — so big cohorts count more; the simple mean averages the cohort percentages equally. Both are shown because reports use either.
Checks
A count that rises from the previous period is flagged (re-activation is possible; a data slip is more common), as are lines that cannot be read. The heat shading is proportional to the percentage and carries no threshold. Nothing leaves the browser; the same four anonymous usage counts as the rest of the site apply.
SOURCES
- Retention = active in period ÷ cohort size; weighted average = Σ active ÷ Σ size over cohorts that reached the period; simple mean of cohort percentages — arithmetic definitions
Last reviewed 20 September 2026. How results are checked: How we verify.