Cohort analysis
Cohort analysis groups records by a shared start event and follows that fixed group over time, which is the only way to separate mix change from behavior change.
Period-over-period comparison mixes cohorts: this quarter’s churn rate includes customers acquired under different conditions, pricing and product. A cohort view holds the group constant, so a change in the curve is a change in behavior rather than a change in composition.
Cohorts are the correct basis for retention, conversion and payback measurement alike. The discipline they impose is definitional: a cohort requires an unambiguous start event and a stable identity over time, which surfaces data model weaknesses immediately.
Where it breaks
Retention is reported as a period ratio, so a surge of new customers appears as improved retention.
Related terms
- Net revenue retention (NRR)
- NRR is ending recurring contracted value for a starting cohort divided by that cohort’s starting recurring contracted value, after churn, contraction, and expansion and excluding new customers.
- Funnel conversion rate
- Funnel conversion rate is the proportion of records that transition from one governed stage to the next, measured on a consistent cohort basis.
- Metric definition
- A metric definition is the governed specification of a measure, its formula, source fields, filters, grain, owner and known limitations.
- Unit economics
- Unit economics is the per-unit relationship between what it costs to acquire and serve a customer and what that customer returns over their lifetime.
Field notes on this
- You just inherited RevOps. Start with a charter.
Being told to "own revenue operations" is not a mandate. Until authority is written down, every decision you make is reversible by whoever objects loudest.
- Why four teams report four different revenue numbers
A reconciliation problem may be definitional, technical, or both. Test the metric contract before buying another data platform.