In RevenueCat's 2026 data, the median subscription app sees 5% to 7% of installs start a free trial within 30 days, about 32% of those trials convert to paid, and 2.0% of all installs become paying subscribers within 35 days. After that, it depends on the plan. Of monthly subscriptions, 53% to 61% make the first renewal, 14% to 26% are still active at six months and 6% to 14% at a year. Annual plans keep 20% to 40% of subscribers through the first renewal, and weekly plans keep 1% to 2% for a full year. Each of those numbers uses a different denominator and a different window, so they describe six separate measurements, not one cohort travelling down a funnel.
This page is the hub for our subscription benchmark series. It puts every stage side by side with its definition, explains when each one can actually be read for a new cohort, and turns a median funnel into a value per install. Where a stage has its own deep dive, we link to it rather than repeat it. For the strategy that sits around these numbers, start with our subscription app marketing strategy guide.
Key takeaways
- The six funnel benchmarks are medians across different sets of apps with different windows. Use them to find your weakest stage, not to build a single chained forecast.
- Every stage has a date before which it cannot be read. Judging a cohort's renewal or year-one value before then is guessing.
- In our worked example, a monthly-plan cohort read at day 35 has delivered only about 27% of its first-year net revenue. Most of the value arrives after the point where most teams make their spend decisions.
- A 10% lift at the trial or paid stage is worth more than a 10% lift at first renewal, because it multiplies every later payment and arrives sooner.
The 2026 subscription app funnel at a glance
All figures below come from RevenueCat's State of Subscription Apps 2026, built on more than 115,000 apps and over $16 billion in revenue. They are medians across apps, and where the report gives a range it is the spread of category medians. We use one source for the whole table on purpose, for reasons explained further down.
| Stage | Denominator | Window | 2026 median |
|---|---|---|---|
| Install to trial | Installs (trial-based apps only) | Trial started within 30 days of download | Most non-gaming categories 5% to 6%; Health & Fitness 6.9%; Business 9.1% |
| Trial to paid | Free trial starts | Trial converts to a paid subscription | 32.0% App Store, 32.5% Google Play; Health & Fitness 37.7% |
| Download to paid | Installs (all apps) | At least one paid subscription within 35 days of install | 2.0% overall; Health & Fitness 2.9%; hard paywall 10.7% |
| First renewal | Paid subscriptions | Renews at the end of the first billing period | Monthly 53% to 61%; weekly 35% to 58%; annual 23% to 40% |
| Six-month retention | Paid subscriptions | Enough paid renewals to cover six months | Monthly 14% to 26% (top quartile 30% to 50%) |
| Year-one retention | Paid subscriptions | Enough paid renewals to cover twelve months | Annual 20% to 40%; monthly 6% to 14%; weekly 1% to 2% |
RevenueCat defines retention as the share of paid subscriptions that remain active after a given period, where a subscription counts as retained only if it has made enough paid renewals to cover the time. Twelve monthly renewals count as a year. That is a subscription-level measure, so it starts at the first payment, not at install.
Why you cannot simply multiply the stages together
The tempting move is to chain the medians: installs times trial rate times trial-to-paid gives paying subscribers. It is worth doing once, as a check, because it shows how loosely the stages fit.
Published Health & Fitness download-to-paid median (D35): 2.9%
Close, but not the same, and there are three reasons it never will be. First, the trial benchmark is calculated only for apps that use a trial-based acquisition strategy, while download-to-paid includes apps that sell straight from a paywall with no trial. Second, the windows differ: 30 days for trial starts, 35 days for paid conversion. Third, each figure is a median across a different set of apps, and the median of a product is not the product of the medians.
The practical rule: compare each of your own stages to its own benchmark, then use your own chained numbers for forecasting. The benchmarks tell you which stage is weak. Only your data tells you what your funnel is worth.
Benchmarks for each stage of the funnel
Install to trial
RevenueCat calls this download-to-trial: the share of installs that start a free trial within 30 days of download. Business apps lead at a 9.1% median, Health and Fitness sits at 6.9%, most non-gaming categories land at 5% to 6%, and North America leads the regions at 7.1%. Because nearly all trials start on install day, this is a first-session number, decided by the store page, onboarding and whether the paywall appears before the user has seen enough to want it. Our install-to-trial benchmarks break it down by category, market and price point.
Trial to paid
The share of free trial starts that become a paid subscription. The store medians are almost identical at 32.0% on the App Store and 32.5% on Google Play, which is striking given how far iOS leads at the earlier stages. Trial length, category and market all move it: Health and Fitness has a 37.7% median and India and South East Asia 15.2%. Our trial-to-paid benchmarks cover every slice the report publishes.
Download to paid (D35)
The share of installs that produce at least one paid subscription within 35 days. The median is 2.0%, with Health and Fitness at 2.9%, Gaming at 1.0%, hard paywall apps at 10.7% and freemium apps at 2.1%. On the App Store the median is 2.6% against 0.9% on Google Play. This is the most useful single number for paid acquisition, because its denominator is the thing you buy. Our download-to-paid benchmarks explain why D35 is a floor rather than a final count, and the hard paywall versus freemium comparison explains why that 10.7% is not an automatic case for a hard paywall.
First renewal and beyond
Renewal is where plan type takes over. The table shows RevenueCat's ranges of category medians for each plan.
| Plan | First renewal | Third renewal | Retained at one year |
|---|---|---|---|
| Weekly | 35% to 58% | 77% to 81% | 1% to 2% |
| Monthly | 53% to 61% | 73% to 82% | 6% to 14% |
| Annual | 23% to 40% | 56% to 70% | 20% to 40% |
Two patterns matter for acquisition. Renewal rates rise after the first cycle, because the people most likely to leave do so first: monthly plans go from 53% to 61% at the first renewal to 73% to 82% by the third. And the first renewal is the steepest single drop after the trial, which is why subscription consultant Alice Muir Kocourková describes it in the report as the point where subscriptions are won or lost. Annual plans look weak at the first renewal only because that renewal falls a year after the payment.
Year-one retention has slipped a little. Between the 2023 and 2024 cohorts the medians fell from 31% to 28% for annual plans, 10% to 8% for monthly and 1.7% to 1.2% for weekly. Price matters too: low-priced annual plans retain 36% at a year against 23% for high-priced ones. How these differences play out in money terms is the subject of our weekly versus monthly versus annual comparison.
Six-month retention
RevenueCat puts monthly plans at a 14% to 26% median six months in, with the top quartile at 30% to 50%. We do not quote a six-month weekly figure, because the report gives two that do not agree, 5% to 15% in one place and 3% to 6% in another. When a source contradicts itself, the honest move is to leave the number out.
When each benchmark becomes readable for a new cohort
The most common misuse of funnel benchmarks is comparing a cohort to them before the cohort is old enough. This is the order in which the stages mature.
| Stage | Readable from | Why |
|---|---|---|
| Install to trial | Within a few days of install | Most trials start on install day: 89.4% in Adapty's data, 78% to 90% by category in RevenueCat's. |
| Trial to paid | Trial length plus a few days | The conversion happens when the trial ends, and failed first payments can take a few days to resolve. |
| Download to paid (D35) | Day 35, and even then not final | RevenueCat finds 19.2% of paid conversions arrive in week six or later. |
| First renewal | Trial plus one billing period | About day 37 for a monthly plan with a seven-day trial; over a year for an annual plan. |
| Six-month and year-one retention | Six or twelve months after the first payment | No shortcut. Early renewals predict them, but only for the plan mix you actually sold. |
On iOS there is a second clock. Apple's attribution postbacks for a paid install arrive in delayed windows, so the campaign-level view of a cohort fills in over several weeks, which we cover in our guide to SKAdNetwork postback delays. The result is that the stages you can read quickly, trial starts and early trial conversion, are the ones you end up optimising hardest. That is fine as long as someone checks later that the renewals followed.
Why RevenueCat and Adapty report different funnels
Adapty's State of In-App Subscriptions 2026 is the other large public dataset, and its numbers look different. Neither is wrong. They are built differently.
| Point | RevenueCat 2026 | Adapty 2026 |
|---|---|---|
| Sample | 115,000+ apps, $16bn+ revenue, 1bn+ transactions | 16,000+ apps, $3bn subscription revenue |
| Statistic | Medians across apps, often with the top quartile | Averages of per-app rates |
| Windows | Stated: trials within 30 days, paid within 35 days | Not stated on the report page for most funnel stages |
| Store mix | Both stores, with store splits published | Mostly App Store, Google Play where stores are compared |
| Headline funnel | Separate benchmark per stage, no single chained funnel | Interactive view: 11.2% install to trial, 27.8% trial to paid, 59.2% first renewal |
The renewal stage shows the difference most clearly. Adapty's interactive funnel gives 59.2% at the first renewal, 45.1% at the second and 37.1% at the third, falling each time. RevenueCat's monthly medians rise each time. That pattern is consistent with Adapty counting the share of original subscribers still renewing and RevenueCat counting the share of those who reached the previous renewal, but Adapty's page does not say which, so we do not convert one into the other. Pick one source for any comparison and name it.
Paid installs usually convert below these medians
Every benchmark above mixes paid, organic, search and referral installs. People who searched for your app are closer to paying than people whose scroll was interrupted by an ad. Adapty's Apple Ads benchmarks, drawn from more than a million ad groups, put Apple Ads install-to-paid at 1.92% against 0.91% for other paid channels, and trial starts at 3.66% against 1.45%. Adapty frames "other paid" as channels such as Meta and TikTok. If your Meta cohorts sit below the overall medians, that is expected. The comparison that matters is your paid cohorts this quarter against your paid cohorts last quarter.
What a median funnel is worth per install
Here is a hypothetical Health and Fitness app selling a $9.99 monthly plan, the North American median monthly price in RevenueCat's data, with a free trial, on the App Store. Every rate is taken from the medians above, using the midpoint where the report gives a range. These are not client figures.
| Point in the cohort | Subscribers | Assumption |
|---|---|---|
| Install (month 0) | 10,000 installs | Spend depends on your CPI |
| Trial starts | 690 | 6.9% install to trial |
| First payment | 260 | 37.7% trial to paid |
| First renewal (month 1) | 148 | 57% renew |
| Second renewal (month 2) | 105 | 71% renew |
| Third renewal (month 3) | 82 | 78% renew |
| Month 6 | 52 | 20% of first payers still active |
| Month 12 | 26 | 10% of first payers still active |
Months four, five and seven to eleven are interpolated smoothly between those points. Across the first year the cohort makes about 965 monthly payments, or 3.7 per first payer. At Apple's standard first-year rate the developer keeps 70% of each $9.99 payment, about $6.99.
Per install: $6,750 ÷ 10,000 = $0.67
Net revenue at day 35 (first payments only, because a seven-day trial puts the first renewal at about day 37): 260 payments × $6.99 ÷ 10,000 = $0.18 per install, about 27% of the year-one figure
Three things follow. First, a monthly-plan app that judges paid acquisition at day 35 is looking at about a quarter of the first year's value, so a cohort that looks unprofitable then may not be. Our guide to break-even CPI shows how to set a ceiling that accounts for that.
Second, the levers are not equal. In this model, a 10% relative improvement in install-to-trial or trial-to-paid lifts year-one revenue per install from $0.67 to about $0.74, because it adds subscribers who then renew. The same 10% relative improvement at the first renewal, 57% to roughly 63%, lifts it to about $0.72, and that money arrives a month later. Earlier stages compound.
Third, the median is a poor plan. At $0.67 per install over a full year, a Meta CPI of a few dollars cannot pay back on this funnel. Most apps that grow profitably on paid social sit well above the median at one or more stages, usually through a higher price, an annual plan mix or a much stronger first session. That is why the gap between your stage and the median tells you where to work.
Where to look when a stage is below benchmark
- Install to trial is low. Usually a first-session problem: onboarding length, paywall timing, or an ad that promised something the app does not show early. Our piece on ad-to-paywall message match covers the version of this that creative causes.
- Trial to paid is low. Check measurement first, then whether users reach the product's value during the trial, then trial length and price. Split by market before concluding anything: the regional spread is wider than most apps' week-to-week movement.
- Early stages are fine but download-to-paid lags on paid traffic. Compare paid cohorts against each other, not against the blended median, and check whether a cheaper CPI has brought in lower-intent users.
- First renewal is low. The product did not deliver what the purchase promised, or the price and plan mix are wrong for the audience. Ads can contribute if they attract people for a reason the app does not satisfy.
- Everything is near median but CAC is too high. Then it is an acquisition problem: creative, targeting or market choice. The iOS versus Android economics alone can change which campaigns are worth running.
How we use funnel benchmarks in an account
We use these benchmarks to decide where the first month of work goes, not to set targets. If an app's install-to-trial rate is far below its category, more creative will mostly buy more installs that do not start trials, so the first conversation is about onboarding. Our Steps & Beasts case study is the example on our site: heavy creative testing ran alongside onboarding changes that lifted install-to-subscriber conversion, and active subscriptions rose 118%.
Once the funnel is sound, the work moves to reading cohorts at the right age. Our Meta Ads management reports cohort-level ROAS from your own revenue data rather than the platform's figure, which is the only way to see whether the later stages followed the early ones. Our ROAS benchmarks at D7, D30 and D60 show how that value builds over time. If your app already spends at least £3,000 a month on Meta and you want a second view of where your funnel is leaking, you can apply to work with us.
Frequently asked questions
What is a good conversion funnel for a subscription app in 2026?
Using RevenueCat's 2026 medians, a typical subscription app sees 5% to 7% of installs start a free trial within 30 days, about 32% of trials convert to paid, and 2.0% of all installs pay within 35 days. For monthly plans, 53% to 61% of subscriptions make the first renewal, 14% to 26% are still active after six months and 6% to 14% after a year. Compare against your own category, store and plan type rather than these overall figures, because each of them moves the numbers more than most apps move week to week.
Can I multiply install-to-trial by trial-to-paid to get download-to-paid?
Only as a rough sanity check. Each benchmark is a median across a different set of apps: download-to-trial is calculated only for apps that use trials, while download-to-paid includes apps that sell without a trial. The windows also differ, 30 days for trials and 35 days for paid. For Health and Fitness the product of the two medians is about 2.6% against a published download-to-paid median of 2.9%, which is close enough to sense-check your own data but not close enough to treat the stages as one cohort.
What is a good first renewal rate for a subscription app?
RevenueCat's 2026 category medians are 53% to 61% for monthly plans, 35% to 58% for weekly plans and 23% to 40% for annual plans. Renewal rates then rise at each later renewal for monthly plans, to 65% to 77% at the second and 73% to 82% at the third, because the subscribers most likely to cancel leave first.
What is a good one-year retention rate for a subscription app?
RevenueCat measures one-year retention as the share of paid subscriptions that have made enough renewals to cover the year. The 2026 category medians are 20% to 40% for annual plans, 6% to 14% for monthly plans and 1% to 2% for weekly plans. Median one-year retention fell slightly between the 2023 and 2024 cohorts, from 31% to 28% for annual, 10% to 8% for monthly and 1.7% to 1.2% for weekly.
Why do RevenueCat and Adapty report different funnel benchmarks?
They measure different things in different ways. RevenueCat publishes medians across more than 115,000 apps with explicit windows, such as trials within 30 days and paid conversion within 35 days. Adapty averages per-app rates across more than 16,000 apps, mostly on the App Store, and its report page does not state a measurement window for most funnel stages. Their renewal figures also appear to be built differently: Adapty's fall with each renewal while RevenueCat's rise. Use one source per comparison and never mix them in a single table.
The short version
The 2026 median subscription funnel runs roughly 5% to 7% of installs to trial, about 32% of trials to paid, 2.0% of installs to paid within 35 days, then 53% to 61% of monthly subscribers through the first renewal and 6% to 14% through a full year. Treat each stage as its own benchmark with its own denominator and its own date of maturity, compare your paid cohorts with each other rather than with blended medians, and remember that for a monthly plan most of the first year's value arrives after the point where most spend decisions get made.