Blended ROAS is all the revenue your app received in a period divided by all the ad spend in that period. Cohort ROAS takes the users who installed in a period and divides what those users have paid, up to a stated age, by what it cost to acquire them. For a subscription app the two can disagree for months, because most of this month's revenue comes from subscribers you acquired earlier. Use cohort ROAS to decide where the next pound of ad spend goes. Use blended ROAS to check that the business as a whole can afford its marketing. Do not use either for the other job.
This guide defines both, adds a third version that sits between them, and works through an example in which blended ROAS falls by 15% while the return on new subscribers almost halves. It then covers the opposite case, where blended ROAS looks bad and the cohorts are fine, and ends with a table of which number to use for which decision. It is part of our wider guide to subscription app marketing strategy.
Key takeaways
- Blended ROAS answers "is the business covering its marketing?". Cohort ROAS answers "is this spend paying back?".
- In a subscription app, renewals from old cohorts can be most of the revenue in a blended figure, so it reacts slowly when new cohorts get worse.
- Blended ROAS moves with the size of the budget. It drops when you raise spend and jumps when you cut it, even if every cohort performs the same.
- Cohort ROAS has its own weaknesses: it is slow to mature and it depends on attribution.
- Report both, with the definition written next to each number.
What blended ROAS and cohort ROAS mean
Blended ROAS = all revenue received in a period ÷ all ad spend in the same period
Cohort ROAS at day N = revenue paid by day N by the users who installed in a period ÷ ad spend that acquired those users
The difference is in what gets grouped together. Blended ROAS groups by calendar: everything that landed in November over everything spent in November. Cohort ROAS groups by people: everyone who installed in November, followed for as long as you choose. The blended figure is often called the marketing efficiency ratio, or MER.
In practice there are three versions in use, and teams often mean different ones by the same word.
| Version | Revenue counted | Spend counted | Question it answers | What it cannot see |
|---|---|---|---|---|
| Blended ROAS | All revenue received in the period: new subscribers, renewals and organic users | All ad spend in the same period | Is the business as a whole covering its marketing? | Says nothing about what this period's spend bought |
| New-user cohort ROAS | Revenue to a stated age from everyone who installed in the period, paid and organic | Ad spend in that install period | Are new users, taken together, paying back the spend? | Organic users flatter it, and a change in organic volume moves it |
| Paid cohort ROAS | Revenue to a stated age from users attributed to paid ads in the period | Ad spend that acquired those users | Is this spend, campaign or concept paying back? | Depends on attribution, which is partial and delayed on iOS |
The middle version matters because many apps cannot cleanly separate paid installs from organic ones on iOS. Dividing revenue from all new users by ad spend is a fair fallback, as long as you remember that organic users are included in the revenue and not in the cost. Our guides to day 7 ROAS and ROAS at D7, D30 and D60 cover how to choose the age at which you read a cohort. This guide is about the choice between the cohort and the calendar.
Why the two disagree more for subscription apps
In a shop that sells one-off products, most of the revenue from an ad arrives within days of the spend, so the calendar and the cohort stay close. A subscription app breaks that link in three ways.
Renewals from old cohorts
A subscriber acquired last year pays again this month. That payment sits in this month's blended revenue, but this month's ads did nothing to earn it. The longer an app has been running, the larger this share becomes.
Organic users
People who found the app through search, word of mouth or the store pay too. Their revenue is in the blended figure and no ad spend is set against it.
Trials and late conversions
Spend happens today and revenue arrives later. Adapty's State of In-App Subscriptions 2026 reports that 89.4% of trial starts happen on the day of install, so the trial clock usually starts at once, but the first payment waits for the trial to end. RevenueCat's State of Subscription Apps 2026 reports that 50.6% of paid conversions happen on day 0 and 19.2% happen in week six or later. Both figures are shares of the conversions that eventually happen, not shares of installs. So part of the revenue from this month's spend lands in next month's blended figure, and some lands the month after that.
Worked example: blended ROAS holds up while new cohorts get worse
Every figure here is hypothetical. It is an illustration, not client data and not a benchmark. Take an app with one monthly plan at £14.99. After UK VAT and a 30% store commission, each payment is worth about £8.75 in proceeds. Of 100 new subscribers, 60 pay for a second month, 45 for a third, 36 for a fourth, 30 for a fifth and 26 for a sixth, and after that 90% of those left renew each month. On those assumptions a new subscriber is worth about £35.58 in their first 12 months and about £46.46 over their whole life.
The app has spent £30,000 a month on ads for a long time at £30 per paid subscriber, and it also gains 250 organic subscribers a month. Then the cost of a paid subscriber starts to rise, as it does when creative wears out and nothing new replaces it. Spend stays at £30,000.
| Month | Ad spend | Cost per paid subscriber | New paid subscribers | Total proceeds | Blended ROAS | 12-month cohort ROAS |
|---|---|---|---|---|---|---|
| Baseline | £30,000 | £30 | 1,000 | £57,000 | 1.90 | 119% |
| Month 1 | £30,000 | £33 | 909 | £56,300 | 1.88 | 108% |
| Month 2 | £30,000 | £37 | 811 | £55,100 | 1.84 | 96% |
| Month 3 | £30,000 | £41 | 732 | £53,600 | 1.79 | 87% |
| Month 4 | £30,000 | £46 | 652 | £51,900 | 1.73 | 77% |
| Month 5 | £30,000 | £51 | 588 | £50,100 | 1.67 | 70% |
| Month 6 | £30,000 | £57 | 526 | £48,300 | 1.61 | 62% |
| Month 7 (spend halved) | £15,000 | £40 | 375 | £45,600 | 3.04 | 89% |
The last column is the paid cohort's proceeds in its first 12 months divided by what it cost, which is £35.58 divided by the cost per paid subscriber. Total proceeds are rounded to the nearest £100.
Read the two right-hand columns against each other. Over six months blended ROAS falls from 1.90 to 1.61, a drop of 15%. A team that looks only at that number sees an account that still returns £1.61 for every pound and concludes that things have softened a little. Over the same six months the 12-month return on new paid subscribers falls from 119% to 62%, which is almost half. From month 2 a new cohort no longer earns back its cost inside a year. By month 6 the app is paying £57 for a subscriber worth about £46.46 over their whole life, so each one is bought at a loss of more than £10.
The blended figure barely notices because renewals and organic users make up 81% of proceeds at the start and 86% by month 6. The new paid cohort is too small a share of the month to move the ratio much.
Now look at month 7. The team finally cuts spend in half and keeps only the better campaigns, so the cost per subscriber falls back to £40. Blended ROAS jumps to 3.04, the best figure in the table. Nothing about the ads has been fixed. The new cohort still returns only 89% in a year. The ratio improved because the bottom of the fraction halved while the old subscribers kept paying.
That is the pattern to remember. For an app with a subscriber base, the fastest way to raise blended ROAS is to stop advertising. A metric that rewards you most when you stop growing cannot be the one that guides growth.
The opposite case: blended looks bad and the cohorts are fine
Blended ROAS can mislead in the other direction too. Take a simpler hypothetical. An app spends £5,000 a week, every subscriber comes through a seven-day trial, and each week of spend produces £3,000 of first payments one week later. Ignore renewals and count only first payments. In week 2 the team doubles the budget for one week.
| Week | Ad spend | First payments received | Received ÷ same-week spend | Cohort first-payment ROAS |
|---|---|---|---|---|
| Week 1 | £5,000 | £3,000 | 0.60 | 0.60 |
| Week 2 (spend doubled) | £10,000 | £3,000 | 0.30 | 0.60 |
| Week 3 (spend back down) | £5,000 | £6,000 | 1.20 | 0.60 |
| Week 4 | £5,000 | £3,000 | 0.60 | 0.60 |
Every cohort returns exactly the same 0.60. The calendar figure halves in the week spend goes up, because the payments that arrive that week belong to last week's smaller cohort. It then doubles in the week spend comes back down. A team steering by the calendar figure would cut the budget in week 2 and feel proved right in week 3.
The same thing happens over longer periods. A young app with almost no renewal base, or one selling mostly annual plans whose next payment is a year away, will show a weak blended figure while it scales, even when every cohort is on course to pay back. Cost per subscriber has the same calendar problem, which we cover in our guide to calculating subscription app CAC. How quickly each plan length brings money in is set out in weekly vs monthly vs annual subscriptions.
Five signs blended ROAS is hiding a problem
You do not need a full cohort model to spot the first pattern. These checks use numbers most teams already have.
| Sign | Where to look | What it tells you |
|---|---|---|
| Renewals are a growing share of revenue | Revenue split by new and renewal transactions | Old cohorts are carrying the number. In the example the share rises from 81% to 86%. |
| New paid subscribers are flat or falling on the same spend | New paid subscriptions per month next to spend | Each subscriber costs more, whatever the ratio says. |
| Cost per trial is rising while blended ROAS holds | Ad account, by week | The earliest sign available, weeks before revenue shows it. |
| Recent cohorts sit below older ones at the same age | Cohort revenue per install at day 30 or day 60 | The clearest evidence. Compare matured cohorts only. |
| Blended ROAS jumps when you cut spend | The month after a budget cut | The base is doing the work. The jump says nothing about the ads. |
Rising cost per trial with steady revenue is often the first sign of creative wearing out. Our guide to creative fatigue for mobile apps explains how to confirm it.
Where cohort ROAS falls short
Cohort ROAS is the better guide to spending decisions, but it is not a clean number either.
- It is slow. A 12-month cohort ROAS takes 12 months. In practice you read an early window and project forward, which is only as good as your older cohorts. Our CAC payback guide covers how far ahead it is reasonable to look.
- It depends on attribution. A paid cohort is only as accurate as the system that decides who was paid. On iOS that information is partial and arrives late, as our guide to SKAdNetwork postback delays explains.
- It can miss revenue the ads caused. Someone who sees an ad and installs days later from a store search may be counted as organic. That revenue shows up in the blended figure and not in the paid cohort.
- Small cohorts are noisy. A weekly cohort for one campaign may hold a few dozen subscribers. Look for a direction across several cohorts before acting on one.
- It ignores cash already coming in. A business with a large renewing base can afford a longer payback than one without. Only the blended view shows that.
This is why the honest answer is both. When the paid cohort figure and the blended figure move in the same direction, you can act with confidence. When they disagree, the disagreement is the thing to investigate.
Which number to use for which decision
| Decision | Use | Why |
|---|---|---|
| Raise or cut the budget | Paid cohort ROAS on matured cohorts, read at the margin | Only the cohort shows what the next pound is likely to buy. |
| Keep, scale or stop a creative concept | Cost per trial now, then cohort ROAS for that concept once it matures | Blended ROAS cannot see a single concept at all. |
| Check the company can afford its marketing this quarter | Blended ROAS on proceeds, including fees and creative costs | Cash is a whole-business question. |
| Judge an agency or a new hire | Cohort ROAS on cohorts acquired on their watch | Blended ROAS credits them with renewals they did not earn, or blames them for a base they did not build. |
| Report to investors or the board | Both, each with its definition written next to it | One shows the health of the business, the other the health of growth. |
| Sense-check attribution on iOS | New-user cohort ROAS and blended ROAS together | Neither depends on attribution, so they catch paid revenue that attribution missed. |
"At the margin" in the first row means looking at what the last increase in budget bought, not the average across the account. Our guide to scaling Meta Ads for a subscription app works through that calculation. The figure in Meta Ads Manager is a fourth number again, limited by its attribution window. We explain why it will not match either of these in the ROAS benchmarks guide linked above.
How to build a cohort read
- Group by install date. Use your subscription platform, not the ad platform. In RevenueCat, for example, the Revenue chart credits each payment to the period the transaction occurred in, which makes it the top half of a blended figure. The Realized LTV per Customer chart groups customers by the date they were first seen and lets you set how many days of revenue to count, which makes it the starting point for a cohort figure.
- Match spend to the same install period. Spend in the first week of November goes against the users who installed in the first week of November, whenever they pay.
- Fix the age. Compare every cohort at the same number of days since install, and mark any cohort that has not reached that age as incomplete.
- Use proceeds. RevenueCat's documentation says its realised LTV figures include the revenue the stores later deduct for commission and taxes. Take those off before you compare with spend. Our agency audit guide shows a reported ROAS of 1.3 becoming 0.76 once VAT and commission are removed.
- Split new revenue from renewals in the blended view. The same Revenue chart can be segmented by transaction type, including new and renewal. That one split shows how much of the blended figure this month's marketing could have influenced.
- Put both on one page. Blended ROAS, the renewal share of revenue, and cohort ROAS at a fixed age for the last several matured cohorts.
If your funnel figures also mix paid and organic users, the same problem affects your cost targets. See how to calculate break-even CPI for how to adjust for it.
How we report it
Our Meta Ads management reports cohort-level ROAS, not the platform's figure. We think a client should see the blended number next to it, as a check on the whole business. Headline results in agency case studies are usually business-level numbers, and ours are no different. Our Steps & Beasts case study reports revenue up 1112% and active subscriptions up 118%, which describes the business, not a single cohort. Numbers like that are the right way to show an outcome and the wrong way to decide which ad to scale on a Tuesday. If you are reviewing an agency's reporting, ask to see both views, and ask which one they use to move budget. Our guide to what an agency should do in the first 90 days sets out when the first matured cohorts should appear.
If your app is already spending on Meta and you are not sure which of the two patterns in this guide you are in, that is the kind of account we work on with consumer and subscription apps. Fees and the minimum spend we work with are on our pricing page.
Frequently asked questions
What is the difference between cohort ROAS and blended ROAS?
Blended ROAS is all the revenue an app received in a period divided by all the ad spend in the same period. Cohort ROAS takes the users who installed in a period and divides what those users have paid, up to a stated age such as day 30 or month 12, by what it cost to acquire them. Blended ROAS describes the whole business this month. Cohort ROAS describes what your ad spend is buying.
Why is my blended ROAS healthy when the app is not growing?
Usually because most of the revenue in the blended figure comes from renewals by subscribers you acquired months ago and from organic installs. Neither depends on this month's ads. If the cost of a new paid subscriber is rising, the blended figure moves slowly because the new cohort is a small share of revenue. Check the share of revenue that comes from renewals, the number of new paid subscribers, and the ROAS of recent cohorts at a fixed age.
Is blended ROAS the same as MER?
In most teams, yes. Marketing efficiency ratio (MER) is total revenue divided by total marketing spend, which is how this guide defines blended ROAS. Some teams include agency fees and creative costs in the spend figure for MER and only media spend for blended ROAS, so check the definition before comparing two numbers.
How old does a cohort need to be before I can trust its ROAS?
Old enough for the trial to end, the first payment to be taken and attribution to settle, and older still for longer windows. RevenueCat's 2026 report found that 19.2% of paid conversions happen in week six or later, so a cohort younger than about six weeks is still missing part of its first payments. Compare cohorts only at the same age, and mark any cohort that has not reached that age as incomplete.
Should an agency report cohort ROAS or blended ROAS?
Both, clearly labelled. Cohort ROAS on matured cohorts should drive decisions about budgets and creative, because it shows what the spend bought. Blended ROAS belongs in the report as a check on the whole business and on cash. A report that shows only blended ROAS cannot tell you whether the agency's work is improving or damaging the account.
The short version
Blended ROAS tells you whether the business covers its marketing. Cohort ROAS tells you whether new spend pays back. In a subscription app, renewals and organic users can hold the blended figure up for months while each new subscriber costs more than they will return, and a budget cut will make it look better still. Make spending decisions on matured cohorts, compared at the same age and on proceeds. Keep blended ROAS for cash and as a cross-check, and watch the renewal share of revenue to see how much of it your current ads can take credit for.