The short answer
Cohort ROAS follows a defined group acquired during the same period and compares its revenue with the associated ad spend. It is useful when revenue continues after the first transaction. Comparisons need the same elapsed observation horizon.
Define membership once
Choose the acquisition event, customer identifier, source assignment, and acquisition dates. Then follow those customers without changing group membership to favor successful outcomes. Specify whether revenue means first-order, cumulative collected, or another clearly stated basis.
Use an age-matched example
Illustrative inputs: cohort A has $2,000 acquisition spend and $6,000 observed revenue by day 60, giving 3×. Cohort B has $2,000 spend and $3,000 revenue by day 15, giving 1.5×. That comparison does not show B is a worse day-60 cohort because B has not reached that age.
If A had $3,200 by its own day 15, its age-matched ratio is $3,200 ÷ $2,000 = 1.6×. The appropriate early comparison is then 1.6× versus 1.5×, with the later outcome still unknown.
Account for repeat-purchase marketing
If follow-up ads or email expenses help generate later revenue, explain whether those costs are included. A ratio using only initial acquisition media omits ongoing marketing costs. Do not call it a complete profitability measure. Track contribution after service, fulfillment, and follow-up costs alongside cumulative revenue.
Keep forecasts visually separate
Future cohort revenue based on historical retention is an estimate. Store actual and projected values in different fields, record the source cohorts, and revisit the assumptions when pricing or customer mix changes. Do not fill an immature cohort’s missing months with historical averages and label the total observed.
Make comparisons reproducible
Record the cohort definition, elapsed age, extraction date, revenue adjustments, and spend-allocation rule. If customers move across accounts or cannot be matched, state the limitation. Cohort analysis improves the time basis of a comparison; it does not remove attribution uncertainty or establish causality by itself.
Build an age-matched comparison
- Gather: Fix membership by acquisition period and save revenue at the same elapsed horizons.
- Produce: Separate acquisition-only spend from follow-up marketing costs in the labels.
Decision boundary: An immature cohort cannot yet demonstrate its own longer-term return.
Next reading
Examples are illustrative, not client results or benchmarks. The calculations use the inputs shown; real-world decisions depend on your records and assumptions. Report a correction.