Cohort retention

A cohort is every customer who started in the same period — January’s signups, or Q3’s new logos. Cohort retention tracks each group forward through its own month 1, month 2, month 3, rather than through calendar months.

This matters because aggregate retention blends cohorts of wildly different ages. A company acquiring rapidly is constantly adding young customers who have not had time to churn, which flatters the blended number. Growth slows, the mix ages, and reported retention appears to worsen even though no cohort actually got worse.

What the curve tells you

Plotted as retention against months since start, three shapes recur:

• Decaying to zero. Every cohort eventually leaves. The product solves a temporary need, or there is no lock-in.
• Flattening. The curve drops then stabilises — there is a durable core of customers who stay indefinitely. This is what you want to see, and the height of the flat portion is roughly the ceiling on long-term retention.
• Smiling. The revenue curve turns upward after the initial drop, because expansion in surviving accounts outweighs continued churn. This is NRR above 100% shown graphically.

Run it on revenue, not just logos

Logo cohorts show whether customers stay. Revenue cohorts show whether they grow. In B2B SaaS the revenue view is usually the more important one, because a cohort that loses 20% of its accounts while the survivors double their spend is a healthy cohort — and the logo curve alone would tell you the opposite.