How cohort analysis works
Anchor each customer to a start period, usually the month of first closed-won or first invoice. Pick a metric and read it forward: how much of each cohort's original revenue remains at month 6 and again at month 12. Lay the cohorts out as rows with age columns across the top, and the shape of the business appears. If your January cohort holds stronger month-12 retention than last January's, onboarding or product changes are working. If newer cohorts decay faster, you have a leak that a whole-base number will not surface for months.
Why forecasts read the base by cohort
Forecasting existing revenue is a retention question, and retention is a cohort property. Gross revenue retention measures the floor each cohort holds after churn and contraction. Net revenue retention adds expansion revenue and shows whether a cohort grows without new logos. Once you know the retention and expansion curve of past cohorts, you can project what this year's cohorts will be worth in the quarters ahead. That projection carries the existing-base half of a revenue forecast, the part that new pipeline never touches. A monthly ARR waterfall reads the same way, resolving beginning ARR into contraction and expansion cohort by cohort instead of smearing the movement across the whole base.
Reading expansion by cohort
Expansion hides inside blended numbers even more than churn does. A cohort can lose logos while the survivors expand enough to lift net retention above 100%. Reading expansion cohort by cohort reveals which start periods produce accounts that grow, which tells you where acquisition spend actually pays back. Segmenting those cohorts by plan or customer segment sharpens the read further, since a self-serve cohort and an enterprise cohort follow different curves and should never share one retention line.
Frequently Asked Questions
What is cohort analysis?
Cohort analysis groups customers by the period they started and tracks a metric such as retention or revenue across each group's age. It answers whether recent customers retain and expand better or worse than earlier ones, a trend a single blended rate averages out of sight.
How is cohort analysis different from a snapshot metric?
A snapshot measures the whole base at one moment and blends customers of every age into one number. Cohort analysis holds age constant by comparing each start-period group at the same point in its life, so real improvement or decay stands out. See [cohort retention vs snapshot retention](/glossary/cohort-retention-vs-snapshot-retention/) for the full comparison.
What should you measure by cohort?
Start with logo retention and revenue retention, then separate gross from net retention so churn and expansion do not cancel each other in the average. A monthly ARR waterfall by cohort shows exactly where each group grows or shrinks.
How many cohorts do you need?
You can read one cohort's curve right away, but the value is in comparison. Twelve monthly cohorts give a year of side-by-side curves, enough to separate a real trend from month-to-month noise and to project forward with confidence.
Put these metrics to work
ORM builds custom revenue forecast models that turn concepts like cohort analysis into prescriptive action for your team.
Schedule a Demo