There is no measured public benchmark for subscription app ROAS at D7, D30 or D60. What exists is revenue per install. RevenueCat's State of Subscription Apps 2026, built on more than 115,000 apps, puts the median at $0.23 by day 14 and $0.34 by day 60, and calls a 1.5x rise between the two typical. The one published set of subscription ROAS ranges we found, 15% to 30% at D7 and 40% to 70% at D30, comes with a note from its own publisher saying it is not a verified median. Because ROAS at any window is revenue per install divided by cost per install, its level depends on your CPI and nobody else's number fits you. How much ROAS grows between windows does not depend on CPI, and that is the part of a benchmark you can borrow.
This page covers what each window measures, what has actually been published for 2026, how much ROAS should grow from one window to the next by plan type and paywall model, a worked example of two apps that look opposite at day 7 and identical at day 365, and how to set your own checkpoints. If you want the single deepest window explained in full, our guide to Day 7 ROAS for subscription apps covers why D7 is the operating metric, how creative quality shows up in it before CPI moves, and how to read it under iOS attribution delays. This page is about how the windows relate to each other.
Key takeaways
- Subscription app ROAS at day N is the revenue a cohort of installs has paid by day N divided by what you spent acquiring that cohort. Without the window, a ROAS figure is meaningless.
- No public dataset reports measured D7, D30 or D60 ROAS medians for subscription apps. RevenueCat publishes revenue per install at D14 and D60 instead, which is the honest starting point.
- The ratio between two windows cancels out CPI, so it is comparable across apps: about 1.3x from D14 to D60 for yearly and hard paywall apps, about 1.6x to 1.7x for monthly and weekly ones.
- Plan mix and trial length decide what each window contains. An annual plan with a short trial earns nearly all of year one by day 7. A monthly freemium app may have earned a fifth.
- Set D7, D30 and D60 checkpoints by working backwards from your payback target using your own mature cohorts, not by copying another app's day 7 number.
What D7, D30 and D60 ROAS mean for a subscription app
Cohort ROAS at day N is the cumulative revenue paid by everyone who installed in a given period, counted up to N days after each person's install, divided by the media spend that acquired them. The denominator is spend on that cohort only. The window is days since install, not days since the ad ran. A D30 ROAS of 40% means the cohort has paid back 40p of every pound spent on it within its first 30 days.
Two details change the number more than most teams expect. The first is whether the window includes day N itself. An app with a 7 day trial that starts on install day takes its first payment on day 7, so a D7 window that stops at the end of day 6 can miss most of the first week's revenue while one that includes day 7 catches it. RevenueCat's 2026 timing data shows the effect: hard paywall apps see a spike of 25.7% of all their paid conversions between days 4 and 7, which the report attributes to 7 day trials expiring. Check how your analytics tool defines the window before comparing it with anyone else's.
The second is the revenue basis. RevenueCat's own documentation says its revenue and realised LTV figures are net of refunds but before store commission and taxes. On a UK App Store subscription in its first year, the developer keeps about 58% of what the customer pays, after 20% VAT and Apple's 30% commission. A 100% ROAS on a gross figure is therefore roughly 58% on proceeds. Neither is wrong, but a team quoting gross to its board and planning on net will overestimate how much it can afford to pay for installs. We covered the netting in detail in our guide to break-even CPI for a subscription app. Every ROAS figure in this article states which basis it uses.
Subscription app ROAS benchmarks: what is actually published in 2026
We read the main reports and the pages ranking for this query on 13 October 2026. The table lists what each source publishes, with the definition next to the number.
| Source | Metric | Published figure | Definition and caveat |
|---|---|---|---|
| RevenueCat, State of Subscription Apps 2026 | Revenue per install at day 14 (median across apps) | All categories $0.23. Health & Fitness $0.48. North America $0.38. Hard paywall $2.32, freemium $0.27. | Total revenue divided by total installs. 115,000+ apps, 2025 data. Not split by paid and organic installs. |
| RevenueCat, State of Subscription Apps 2026 | Revenue per install at day 60 (median across apps) | All categories $0.34. Health & Fitness $0.66. North America $0.55, top quartile above $1.39. Hard paywall $3.09, freemium $0.38. | Same definition. The report calls a 1.5x D14 to D60 multiplier typical. |
| Admiral Media, Mobile App Marketing Benchmarks 2026 | ROAS range shown for subscription apps | D7 15% to 30%. D30 40% to 70%. D90 70% to 110%. | The page states its ranges lack reporting dates, sample counts and a reproducible method and are not verified medians. |
Sources: RevenueCat State of Subscription Apps 2026 and Admiral Media's Mobile App Marketing Benchmarks 2026. RevenueCat's figures are medians across apps for 2025 cohorts. Admiral Media is an app marketing agency, like us, and deserves credit for printing the limits of its own table next to it. Its page says the ranges are not accompanied by reporting dates, sample counts or a reproducible calculation method, and should not be treated as verified 2026 market medians or spending targets.
The other pages ranking for this query mostly explain the windows or divide revenue per install by an assumed CPI. That second method is sound in principle and is the identity this article is built on, but it inherits a problem from the data: RevenueCat's revenue per install divides by all installs, and the report does not split paid installs from organic ones. A paid cohort can monetise better or worse than an app's organic users, so the published medians are a reference for your whole app, not for a Meta campaign.
Why there is no honest D7 ROAS median for subscription apps
Producing one would need four things held constant across thousands of apps: the same window definition, the same revenue basis, the same split between paid and organic installs, and the cost of the installs. The first three vary by tool and team. The fourth is private. The largest subscription dataset is built from revenue events and its report contains no ad spend, while the measured ROAS benchmarks we found were all for games, which monetise on a completely different curve. That is why there is no subscription equivalent to the gaming ROAS tables.
It is also why a single D7 target travels so badly between apps. Our own Day 7 ROAS guide says we work to 100% to 120% at D7 across our health and wellness portfolio, and the same article says that target suits apps where the first payment does most of the work. The agency range above puts subscription apps at 15% to 30% at D7. Both can be sensible, for different apps. The worked example below shows why.
How much should ROAS grow between D14 and D60?
Here is the part that does travel. For a single cohort, ROAS at day 60 divided by ROAS at day 14 is the same as revenue per install at day 60 divided by revenue per install at day 14, because the spend is the same in both. CPI cancels out. So while nobody can tell you what your D60 ROAS should be, the published revenue data can tell you roughly how much it should grow from D14.
ROAS at day N = revenue per install at day N ÷ cost per install
ROAS at D60 ÷ ROAS at D14 = revenue per install at D60 ÷ revenue per install at D14
The table applies that to RevenueCat's published medians.
| Segment | Revenue per install, D14 | Revenue per install, D60 | Implied ROAS growth, D14 to D60 |
|---|---|---|---|
| All categories | $0.23 | $0.34 | 1.48x |
| Health & Fitness | $0.48 | $0.66 | 1.38x |
| Business | $0.31 | $0.50 | 1.61x |
| Gaming | $0.08 | $0.14 | 1.75x |
| North America | $0.38 | $0.55 | 1.45x |
| India and Southeast Asia | $0.08 | $0.11 | 1.38x |
| Yearly plan dominant | $0.36 | $0.46 | 1.28x |
| Monthly plan dominant | $0.18 | $0.29 | 1.61x |
| Weekly plan dominant | $0.19 | $0.32 | 1.68x |
| Hard paywall | $2.32 | $3.09 | 1.33x |
| Freemium | $0.27 | $0.38 | 1.41x |
Medians from RevenueCat State of Subscription Apps 2026. Plan rows classify each app by its most-sold plan duration but count all of its revenue, as the report notes. The ratios are our calculation.
Three caveats keep this honest. A ratio of two medians is not the median of each app's ratio, so treat the column as directional. The plan figures come from the report's chart, which matches the growth percentages in its own commentary, but a contributor summary in the same report quotes different monthly and weekly figures, so we have used the chart. And the growth only runs to day 60: for annual plans almost nothing happens between day 60 and the first renewal, while monthly and weekly cohorts keep compounding.
The useful reading is the spread. Yearly and hard paywall apps earn most of their 60 day revenue early, so their ROAS grows about 1.3x after day 14. Monthly and weekly apps keep collecting, so theirs grows about 1.6x to 1.7x. If your own D14 to D60 growth is well below the figure for your model, revenue is leaking after the first payment. If it is well above, your early windows are understating the cohort and you may be cutting campaigns too soon.
What lands in each window, by plan and trial length
The multiplier tells you how much ROAS grows. This table tells you why, which is what you need to read a single cohort. It assumes trials start on install day, which is where most of them start: RevenueCat finds 82.1% of Health & Fitness trial starts happen on day 0, and as many as 89.9% in Business apps.
| Model | D7 contains | D30 contains | D60 contains |
|---|---|---|---|
| Annual plan, 3 or 7 day trial | The first payment, which is nearly all of year one revenue. With a 7 day trial it lands on day 7, so check whether your tool's D7 includes day 7. | Roughly flat. Early refunds come out. | Flat. The next money is the renewal, around a year later. |
| Monthly plan, 7 day trial | The first month's payment from trial converters, if day 7 is inside the window. | First payments only, plus a few later converters. No renewals yet. | The first renewal, around day 37, is now in. |
| Monthly plan, 14 day trial | Almost nothing from trial users. Only people who skipped the trial. | First payments. | First renewal, around day 44. |
| Weekly plan, no trial | The first payment and possibly the first renewal. | Around four weekly payments from those still subscribed. | Around eight. Weekly ROAS keeps climbing steeply while churn allows. |
| Freemium, any plan | Early buyers only. RevenueCat finds 23% of freemium conversions arrive 6 or more weeks after download. | Still missing a large share of eventual payers. | Closer, but the tail continues. |
Two consequences follow. D30 is the weakest window for a monthly app with a 7 day trial, because it contains the first payment and nothing else, which is why D60 is often the first window worth budgeting on for that model. And freemium apps need patience at every window: RevenueCat reports that more than 60% of paid conversions happen within a week for both freemium and hard paywall apps, but freemium has 1.5 times the late conversions, with 23% arriving after six weeks. Day 60 still has not seen all of them. How your trial users convert is covered in our trial-to-paid conversion benchmarks, and how many installs eventually pay in our download-to-paid benchmarks.
Worked example: two apps, opposite at D7, identical at D365
The two apps below are hypothetical. Neither is a client and the inputs are chosen to make the shape visible, not to represent a benchmark. Both are sold on the UK App Store and ROAS is calculated on proceeds, after 20% VAT, the 30% first year commission and a 3% refund rate on first payments.
- App A, hard paywall, annual plan. 1,000 installs at £3.00, so £3,000 of spend. 12% start a 7 day trial on install day and 40% of those convert, giving 48 subscribers who each pay £59.99 on day 7, or £34.99 in proceeds. 30% renew a year later at the 85% rate.
- App B, freemium, monthly plan. 1,000 installs at £1.70, so £1,700 of spend. 8% start a 7 day trial and 35% convert, giving 28 subscribers on day 7. Another 12 people buy without a trial over the next three months, 4 of them in the first week. Each pays £9.99 a month, or £5.83 in proceeds, and the share still paying falls month by month to 12% by the twelfth payment.
| Window | App A, net ROAS | App B, net ROAS |
|---|---|---|
| D7 | 56% | 11% |
| D30 | 54% | 12% |
| D60 | 54% | 19% |
| D365 | 54% | 54% |
| After the year one renewal (day 372) | 74% | Still renewing monthly |
Both cohorts have returned 54% by day 365. App A got there on day 7. App B had returned 11% on day 7, a fifth of its year, and earned the rest over the following eleven months. Judge App B against App A's D7 and you would switch off a campaign that returns exactly as much within the year. Judge App A against App B's curve and you would wait months for revenue that is never coming, because App A's next payment is the renewal on day 372.
The basis matters too. On the gross figure most revenue dashboards show by default, before VAT and store commission, App A's D7 reads 96%. On proceeds it is 56%. That is the difference between a cohort that appears to have paid for itself in a week and one that is nearly half short, and it is the most common reason a D7 target borrowed from another team turns out to be wrong.
How to set your own D7, D30 and D60 targets
Work backwards from the outcome you need, using the shape of your own mature cohorts. Three steps.
- Choose the horizon and the return. For most subscription apps that is the ROAS on proceeds you need by the end of your payback window. How long that window can be is a cash question, which we cover in our guide to CAC payback for subscription apps.
- Measure your own curve. From cohorts old enough to have reached that horizon, calculate the share of horizon revenue each had earned by day 7, 30 and 60. Use several cohorts and the median share, and rebuild the curve whenever you change price, trial length or plan mix.
- Multiply. Each checkpoint is the required return times the share for that window.
Checkpoint at day N = required ROAS at your payback horizon × share of horizon revenue your mature cohorts had earned by day N
Applied to App B with a target of 100% by day 365: its curve earns 20% of year one revenue by day 7, 22% by day 30 and 36% by day 60, so its checkpoints are about 20%, 22% and 36%. The actual cohort read 11% at D7, so on day 7 you already know it is heading for roughly half of what it needs, and the fix is cheaper installs or better conversion, not more patience. App A, with almost all of year one in week one, needs slightly more than 100% on proceeds at D7 to pay back inside the year without counting renewals, which is the kind of target our Day 7 guide describes.
A new app has no mature cohorts, so borrow the shape rather than the level. Use the D14 to D60 growth for your model from the table above as a first estimate, set provisional checkpoints from your target, and replace them with your own curve as soon as two or three cohorts reach day 60. The same back-solve sets your maximum acquisition cost, which our guide to setting a target CAC walks through.
Why Meta's ROAS column will not match your cohort ROAS
Ads Manager answers a different question. It credits purchase value to ads inside an attribution window after a click or a view, with 7 day click and 1 day view as the default setting. A monthly renewal on day 37 can never fall inside that window, so for a subscription app the platform's ROAS is structurally closer to an early window than to D60, whatever date range you select. Cohort ROAS, built from your subscription revenue and your spend, has no such limit.
On iOS the platform figure also arrives late and incomplete, because attribution postbacks are delayed by design. Our guide to SKAdNetwork postback delays covers how long each one takes. The working rule we use: compare ads and concepts inside Meta on Meta's own figure, because the bias is roughly the same for every ad, and make budget decisions on cohort ROAS, because that is the money.
What to do when a window comes in below target
Reading the level and the growth together tells you where to look. This is the order we work through on client accounts.
| What you see | What it usually means | Check first |
|---|---|---|
| D7 below checkpoint, growth from D7 to D60 normal for your model | The cohort is the usual shape at a lower level. Fewer people are reaching trial or paid, or installs cost more. | Cost per install and install to trial for the new creative, before anything else. |
| D7 on checkpoint, D30 or D60 growth below your usual multiple | Early revenue arrived but did not stick. Renewals, refunds or late conversions are weaker than before. | First renewal rate, refund rate and whether a new angle is overpromising what the product delivers. |
| D7 low, later growth unusually strong | Revenue is arriving later, not disappearing. Common after moving to a longer trial or towards freemium. | Whether the plan mix or trial length changed. Re-derive the checkpoints rather than cutting spend. |
| Every window below checkpoint across all creative | Not a creative problem. Pricing, the paywall, a market change or a measurement break. | Tracking first, then the funnel from install to paid. Creative changes will not fix it. |
The first row is the most common, and it is often a creative problem that looks like a cost problem. A concept that attracts people outside the real use case can hold its CPI while fewer of its installs start a trial, which is why install-to-trial rate by concept is the earliest read available, resolving within a day of launch. The second row is the one teams miss when they only watch D7: a cohort that hits its early checkpoint and then grows less than the usual multiple is paying for an angle the product does not live up to. For where ROAS sits among the other numbers a subscription app has to manage, see our subscription app marketing strategy guide.
How we report ROAS for subscription app clients
Our Meta Ads service reports cohort-level ROAS rather than the platform figure, with the window and revenue basis agreed at the start. In practice that means a D7 checkpoint for creative decisions, a later window for budget decisions, and both derived from the client's own cohorts rather than from a benchmark. Where a client has no mature cohorts yet, we start from the growth multiples above and replace them as the data arrives.
The same discipline applies to reading anyone's results, ours included. Our Tonaly case study reports 3x ROAS sustained at scale for an app that helps musicians and songwriters, alongside a 175% increase in revenue. Like most published ROAS figures, the case study does not state the window, and that is the first question to ask of any ROAS claim you are shown.
If your app is already spending on Meta and the question is which creative moves the later windows rather than only the first week, our work with consumer and subscription apps is built around that problem. We work with accounts spending at least £3,000 a month on Meta, and current fees are on our pricing page.
Frequently asked questions
What is a good D7 ROAS for a subscription app?
There is no measured public median, so any single number is a guess. The answer depends on how much of your first year's revenue lands in the first week, which is set by your plan mix and trial length. An annual plan with a 3 or 7 day trial collects most of its year one revenue in week one, so its D7 ROAS needs to be close to its payback target. A monthly freemium app may collect only a fifth of its year one revenue by day 7, so a D7 ROAS of 20% can be on track. Work out your own checkpoint by dividing your payback target by the share of mature revenue your past cohorts had earned by day 7.
What is a good D30 ROAS for a subscription app?
Again, it depends on your revenue curve rather than on a market figure. The only published subscription range we found, from Admiral Media's 2026 benchmarks page, shows 40% to 70% at D30, but the page itself says its ranges lack reporting dates, sample counts and a reproducible method and should not be treated as verified medians. For a monthly app, D30 contains the first payment but no renewals yet, so it understates the cohort more than for an annual or weekly app. Judge D30 against the D30 your own mature cohorts showed on their way to payback.
How much should ROAS grow between D14 and D60?
RevenueCat's State of Subscription Apps 2026 describes a 1.5x multiplier in revenue per install between day 14 and day 60 as typical. Because ROAS is revenue per install divided by cost per install, the same cohort's ROAS grows by the same multiple. The ratio varies by model: about 1.3x for apps whose most popular plan is yearly and for hard paywall apps, about 1.6x to 1.7x for monthly and weekly dominant apps. These are ratios of published medians, so treat them as directional.
Should subscription app ROAS be calculated on gross or net revenue?
Use net for decisions, and always say which one you mean. RevenueCat's own documentation says its revenue and realised LTV figures are net of refunds but before store commission and taxes. On a UK App Store subscription in its first year, the developer keeps about 58% of the price the customer pays after 20% VAT and Apple's 30% commission, so a 100% gross ROAS is roughly 58% net. Two teams comparing ROAS without agreeing the revenue basis are comparing different numbers.
Why is the ROAS in Meta Ads Manager different from my cohort ROAS?
Because they answer different questions. Ads Manager credits purchase value to ads within an attribution window after a click or view, with 7 day click and 1 day view as the default, so a monthly renewal on day 37 can never appear in it. Cohort ROAS takes every user who installed in a period and divides everything they have paid by day N by what it cost to acquire them. On iOS, attribution delays add a further gap. Use the platform figure to compare ads inside Meta and the cohort figure to decide budgets.
Which ROAS window should a subscription app use to make decisions?
Use the earliest window that predicts your mature outcome reliably, checked against your own cohorts. For many apps with short trials that is D7 for creative decisions and D30 or D60 for budget decisions, with D60 the first window that contains a monthly renewal. The test is simple: plot D7 against D180 or D365 for your past cohorts. If the ranking of cohorts is stable, D7 is a usable operating metric. If it is not, wait for the window where it becomes stable.
The short version
Nobody publishes a measured D7, D30 or D60 ROAS median for subscription apps, and the ranges that circulate say so themselves. What you can benchmark is the shape: RevenueCat's 2026 data shows revenue per install, and therefore ROAS, growing about 1.5x from day 14 to day 60, less for yearly and hard paywall apps and more for monthly and weekly ones. Agree the window definition and the revenue basis, measure how much of your mature revenue each window contains, and set checkpoints by working back from your payback target. A D7 number only means something next to the curve it sits on.