Cohort Revenue
Date: 2026-08-16
Revenue by acquisition cohort over time. It’s the honest answer to whether the business compounds — and it separates “we’re growing because we acquired more” from “we’re growing because customers are worth more”, which no aggregate can.
What it is
Cohort revenue tracks the cumulative revenue or contribution generated by each acquisition cohort as it ages.
cumulative contribution per customer
month 0 3 6 12 24
Jan cohort £15.00 £19.20 £22.10 £24.00 £27.00
Feb cohort £15.00 £19.80 £23.00 £25.10
Mar cohort £15.00 £20.40 £23.90
Apr cohort £15.00 £21.00
Read down a column to compare cohorts at the same age — this is the version that answers “are we getting better?”
Read across a row to see one cohort mature and to check when it clears CAC.
Two questions it answers that nothing else does
Is acquisition getting better or just bigger? Aggregate revenue rises with acquisition volume regardless of quality. Cohort revenue at fixed age strips volume out entirely — the January cohort at month 6 versus the March cohort at month 6 is a like-for-like comparison.
In the table above, month-3 contribution rises from £19.20 to £21.00 across four cohorts. Something improved — and no aggregate metric would show it.
When does a cohort clear its acquisition cost? With CAC at £22.50, the January cohort crosses at around month 6. That’s the Payback Period observed rather than modelled.
Revenue retention versus customer retention
The distinction cohort revenue exists to expose:
customers remaining contribution per original customer
month 0 100% £15.00
month 12 28% £24.00
Only 28% of customers are still active, and cumulative value per customer is up 60%. Fewer customers, each worth more. That’s a viable business; the reverse — high customer retention, flat revenue — is not.
Neither a customer-count curve nor a revenue total tells you which you have. See Retention Curves.
Use contribution, not revenue
The same rule as everywhere in this domain. Revenue cohorts look excellent and don’t answer the question, because CAC is real money and revenue isn’t money kept — Contribution Margin, Customer Lifetime Value.
Deduct returns too, or a high-return category’s cohorts are systematically overstated — Return Rate and Reverse Logistics.
The traps
- Incomplete cells. April’s month-12 value doesn’t exist yet. Plotting it as zero or averaging it in drags recent cohorts down and makes them look worse than they are — the single most common error in these charts
- Identity improves with age. Older cohorts have better stitching, so they show fewer customers each worth more. That reads as improving value and is partly measurement — fix the identity resolution date to guard against it
- Only identified customers have cohorts. Guest checkout and anonymous traffic are absent — Anonymous and Identified Users
- Cohort size matters. A brilliant curve on 60 customers is noise. Show counts alongside — Sampling Error
- Data retention caps cohort age. You cannot analyse a 24-month cohort on a 14-month window — Data Retention
What to do with it
- Segment by acquisition channel. The spread between channels is usually larger than any campaign effect, and it’s the strongest argument against allocating budget on first-order return — Channel Mix
- Judge lifecycle changes on early cells. Month 1 and month 3 move first and give the fastest read
- Set acquisition ceilings from observed cohort value, not modelled LTV. A cohort that reached £24 by month 12 is a fact; an infinite-horizon LTV of £41 is a forecast — LTV to CAC Ratio