What is cohort analysis?
Short answer: Cohort analysis tracks groups with a common starting date or characteristic over comparable periods since that start. It separates customer lifecycle behavior from calendar growth, which can otherwise hide weakening retention or contribution behind new additions.
Common cohorts use acquisition month, first purchase, contract start, channel, product, geography or customer segment. A calendar report might show total customers rising while every recent cohort retains less well. A cohort table aligns each group by month one, month two and later maturity so that pattern becomes visible. Measures can include active customers, recurring revenue, orders, gross margin, contribution, churn and CAC payback. Cohorts based on different product or pricing regimes may need separate labels rather than being averaged together.
How it works
Define cohort membership and the start event once, then prevent later customer movement from silently rewriting history. Choose a stable customer identifier, handle account mergers and reactivations explicitly and reconcile totals to source systems. Measure each cohort at the same age and use the same revenue and cost definitions. Preserve both the starting population and surviving population so selection effects are visible. Segment only where sample size remains useful. Annotate product, pricing or channel changes and compare forecast curves with realized results. For financial planning, multiply cohort retention and contribution curves by expected new-customer volumes.
Cohort retention at month n = active members from the original cohort at month n / original cohort members
Example
January and February each acquire 100 customers. At month three, January retains 85 and February retains 72, so three-month retention is 85% and 72% respectively. January customers contribute 60 each at that point, producing 85 x 60 = 5,100 of cohort contribution. February customers contribute 65 each, producing 72 x 65 = 4,680. The newer cohort has higher contribution per survivor but lower total contribution because retention weakened. A calendar total that includes March additions could conceal that change.
Why it matters
Cohorts reveal whether recent customers, products or channels are performing better or worse than earlier vintages. Management uses the evidence to change product, service and acquisition decisions and improve forecasts. Buyers and investors use mature cohorts to validate retention, LTV, NRR and payback instead of relying on blended averages. Lenders may use contribution and retention curves when recurring cash is central to the credit case.
Small cohorts and short observation periods create noise. Survivorship bias, missing identifiers and changed definitions can materially distort results. Later cohorts have less time to mature, so comparing their lifetime totals with older cohorts is invalid. Product, price, channel and macroeconomic changes may explain differences but do not prove causation. Cohort analysis is a management KPI rather than an accounting-standard measure and should reconcile to financial or operating source data.