What it calculates

Cohort revenue analysis groups customers by the period they joined and tracks each group separately over time. It separates the behaviour of new customers from the behaviour of old ones, which a blended monthly number cannot do.

How it is calculated

Assign every customer to the cohort of their start period, then measure revenue from that cohort in each subsequent period. Retention is the cohort's revenue in a later period as a percentage of its first period. Net revenue retention includes expansion from the customers who remain, so it can exceed 100% even while some customers churn. Plotting cohorts on the same axis shows whether newer cohorts behave better or worse than older ones.

How to read the result

Watch the shape of the retention curve, not just its level. A curve that flattens means you have found a durable customer segment; one that keeps declining means there is no stable base and growth must come entirely from acquisition. Compare cohorts against each other: if recent cohorts retain worse than older ones, either acquisition quality has fallen or the product has changed for the worse, and blended metrics will hide this for many months.

Worked example

The January cohort starts at 100,000 monthly revenue. By month 6 it is at 82,000, so 82% retention; by month 12 it is at 91,000 because expansion from survivors more than offset churn, giving 91% net revenue retention against gross logo retention of perhaps 70%.

Common questions

What is the difference between gross and net revenue retention?

Gross retention counts only losses - churn and downgrades - so it is capped at 100%. Net retention also counts expansion from remaining customers and can exceed 100%. A business with 85% gross and 115% net retention is losing customers but growing the ones it keeps.

Why not just track overall churn?

Because a blended churn rate mixes cohorts with very different maturities. Rapid growth in new customers, who always churn faster early on, can make overall churn look terrible while every individual cohort is improving - or mask deterioration when growth slows.

How many periods should a cohort be tracked for?

Long enough for the curve to flatten, which for most subscription businesses means at least 12 to 24 months. Truncating early systematically underestimates lifetime value, because the retention curve is steepest at the start and shallowest later.