Shahin

Retention Matrix

Intermediate
Updated Jul 11, 2026 Shahin editorial 2 min read
PronunciationEN/rɪˈtɛnʃən ˈmeɪtrɪks/

A retention matrix is a table whose rows are cohorts by start period and whose columns are periods since start, with each cell holding that cohort's return share.

The table has two reading directions and each answers a different question: moving along a row is one cohort ageing, which is its retention curve, while moving down a column compares different cohorts at the same age and shows whether the product improved over time. The shape is triangular because recent cohorts have not lived long enough to reach the later columns; the cells at the young end are either empty or based on an incomplete period, and they are the single biggest source of wrong conclusions. Cohort triangle and retention heatmap are other names for the same table, not separate concepts.