What Is a Good Free Trial Conversion Rate? The 2026 Benchmarks

You added a free trial, some people start it, and some of them convert to paying. The question every founder asks next is whether their number is any good. There is no single trial-to-paid benchmark to answer that, because the rate swings hard with your trial length, your paywall, your platform, and where your users live. But there are real 2026 numbers to measure yourself against, as long as you compare like with like.

I have been shipping apps since 2012, so let me lay out what the data actually says, from the two biggest public datasets, and how to read your own number without fooling yourself. One caution up front: people throw around trial-to-paid, download-to-paid, and install-to-trial as if they are the same metric. They are not, and mixing them is how you end up measuring your app against a number that was never counting the same thing.

First, the two rulers, and why not to blend them

Almost every benchmark you will see comes from one of two sources.

RevenueCat's State of Subscription Apps 2026 draws on more than 115,000 apps and over $16 billion in tracked revenue. Adapty's State of In-App Subscriptions report draws on a smaller set, around 16,000 apps and $3 billion. They use different samples and, more importantly, different definitions and denominators. RevenueCat often reports download-to-paid; Adapty reports install-to-trial and trial-to-paid as separate steps. So a 10% from one and a 28% from the other are not in conflict, they are counting different things. Pick the metric that matches the number you actually have, and never average across the two.

Check the vintage as well, because these figures move year to year. RevenueCat's overall trial conversion median was 37.3% in the 2024 report, which drew on a 30,000-app sample rather than today's 115,000. Hard paywall download-to-paid was 12.11% in 2025 and 10.7% in 2026. If you are working from a benchmark slide someone put together a year or two ago, the direction of the finding usually survives but the exact number often does not.

What a good trial-to-paid rate is, by trial length

If you run a free trial and want the cleanest trial-to-paid figure, RevenueCat's 2026 data breaks it down by trial length, and length matters more than almost anything else you control.

Long trials, in the 17 to 32 day range, convert at a median of about 42.5% trial-to-paid. Short trials of four days or less convert at about 25.5%, and the middle band of 5 to 9 days sits at 37.4%. That is a spread of 17 percentage points, about two thirds better in relative terms, from a single decision. The likely reason is habit. A longer trial gives the app time to become part of someone's routine before the charge lands, so the choice to keep paying is easier. Short trials force a snap judgment, and they bleed early. On a three-day trial, about 55% of all cancellations happen on the very first day, and around 84% happen in the first two. If your trial is short and your conversion is low, the length is worth testing before you rewrite the paywall copy again. I come back later to how much of that gap is really cause and how much is just correlation.

What a trial buys you, versus no trial

The prior question is whether to gate access at all. Here the framing shifts to download-to-paid, so keep the metric straight.

In RevenueCat's 2026 data, apps with a hard paywall, where you must subscribe or start a trial before you can use the app, convert about 10.7% of downloads to paid within 35 days. Freemium apps, which let you in for free and offer the upgrade later, convert about 2.1% on the same window. That is roughly a five times gap, and hard paywalls ran even higher a year ago, closer to 12%. But this is download-to-paid, a different measurement from trial-to-paid, and it comes as a trade. A hard paywall converts a much larger share of the people who get in, and turns away everyone who would have poked around for free first. The hard paywall number is also drifting down while freemium holds steady, so the gap narrows slightly year over year.

Where hard, soft, and freemium sit alongside trials

The obvious follow-up is whether trial-to-paid moves with paywall type, and whether a long trial pays off differently behind a hard gate than a soft one. RevenueCat does not publish that cross-tab. Its paywall comparison is hard paywall against freemium at the download-to-paid stage, its trial-length numbers are a separate cut, and it does not report soft paywalls as their own category at all. If you see a blog presenting a tidy paywall-type by trial-length matrix, check where it came from, because it is usually the 10.7% figure relabelled as trial-to-paid.

Adapty does publish one matrix worth having, on paywall placement rather than paywall hardness.

Onboarding plus a trial is the strongest of the four combinations, and every placement does better with a trial than without. Adapty does not say what those percentages are divided by, so the ordering is the useful part rather than the numbers themselves.

There is also a finding that reads like a contradiction until you check the pairs being compared. Adapty reports that soft paywalls convert better than hard ones, while hard paywalls produce higher lifetime value per subscriber. RevenueCat reports hard paywalls converting five times better than freemium. Both hold, because they are not comparing the same things. RevenueCat is measuring a hard gate against no gate at all. Adapty is measuring a hard gate against a dismissible one. A dismissible paywall lets more people through and converts a higher share of them; a hard gate stops more people but the ones who do subscribe are worth more.

Which is the same discipline as the rest of this post. Before you take any paywall benchmark as a target, check what it was measured against, because "converts better" means something different in each of those two sentences. And since roughly 90% of trial starts happen on day zero, whichever gate you pick is doing its work in the first session either way.

The lens that changes everything: platform and geography

Two factors move trial numbers so much that ignoring them makes any benchmark useless.

Platform. In Adapty's data, iOS converts about three times better than Android on annual plans, and around 3.6 times on annual subscriptions specifically. RevenueCat's 2024 report pointed the same way on trial-to-paid, with the App Store at 39.0% against Google Play at 28.5%. If your app is Android-heavy, a lower conversion number can be completely normal, and comparing yourself to an iOS-weighted median will make a healthy app look broken.

Geography. Adapty puts install-to-trial at about 14.5% in North America against roughly 7.6 to 10.2% elsewhere. Your trial funnel is simply narrower outside high-income markets, which matters for the last section.

The kicker: the first payment is almost the only one you get

Here is the finding that should change how much you care about trial conversion. In RevenueCat's 2026 data, only about 5% of subscribers who cancel an annual plan ever come back. Roughly 95% never return. Monthly subscribers return at about four times that rate, which is still low. The reactivation rate does move with geography and price tier, so your own figure will differ, but the shape holds wherever you look: win-back is a rounding error next to first conversion.

The lesson: winning the first conversion is worth far more than most founders treat it, because the industry's win-back rate is close to zero. You do not get many second chances at the trial-to-paid moment, so it is worth getting the trial length, the paywall timing, and, as we will see, the price right the first time.

What people actually aim for

Published medians tell you where the middle sits. They do not tell you what to shoot for, and those are different questions.

Steve P. Young runs App Masters, an app marketing agency that has worked on app growth since 2010, and hosts the App Marketing by App Masters podcast, which is closing in on 700 episodes. On one of those episodes he lays out a set of funnel targets that are useful here, because they cover the steps before the trial as well as the trial itself.

The first step is the easiest to fix and the most often ignored. Someone who never reaches your paywall was never going to convert, so anything in onboarding that buries or delays the offer quietly costs you everything below it.

The second step is where most apps lose the most people. It is close to Adapty's install-to-trial figure of 11.2%, though not the same measurement: Adapty divides by installs, Steve divides by people who actually saw the paywall. The closer your install-to-paywall-view gets to 100%, the more those two numbers converge.

The third step is the trial-to-paid rate this whole post is about, and his 30 to 40% target lands inside RevenueCat's 25.5 to 42.5% band by trial length. A practitioner rule of thumb and a 115,000-app dataset landing in the same place is a decent sign the range is real.

My read on why the two datasets disagree

Everything above is published data. This next part is my own read of it, so treat it as a hypothesis to test against your own numbers rather than a finding.

I think the paywall disagreement comes down to who ends up standing in the denominator. A gate does not only change how many people convert, it changes which people are in the pool being measured. A hard gate makes everybody decide at install, so the trial cohort fills up with the merely curious. A soft gate only sends people to the paywall once they have used the app, so that cohort is smaller and far more motivated.

Run it on a thousand installs and both findings appear at the same time. Under a hard gate, all 1,000 see the paywall, 200 start a trial, and 35% of those convert. That is 70 subscribers, or 7% of downloads. Under a soft gate, only 400 ever reach the paywall, 40 start a trial, but 50% convert because they already liked the app. That is 20 subscribers, or 2% of downloads. The hard gate wins download-to-paid by three and a half times. The soft gate wins trial-to-paid by fifteen points. Both numbers are right, and they are describing different groups of people.

Those inputs are invented, but the ratios they produce land close to the real ones: 3.5x on download-to-paid against RevenueCat's 5x, and 43% better trial-to-paid against Adapty's roughly 50%. Close enough that I suspect cohort composition explains most of the disagreement, rather than either company being wrong.

Now look again at which side of that model ends up with more customers. The hard gate produces 70 paying users against the soft gate's 20, while posting the worse trial-to-paid rate. That points the same direction the practitioners do: show the offer early. Adapty's own placement figures in the table above say it too, with onboarding plus a trial as the strongest of the four combinations. Nothing here is an argument for hiding your paywall.

What it does argue is that trial-to-paid is the wrong number to steer by. It is a ratio you can improve by showing your offer to fewer, warmer people, which is also a way to shrink the business. An app that moves from freemium to a hard paywall should expect its trial-to-paid rate to fall while its revenue rises, which reads as a regression on a dashboard and is not one. Judge that change on subscribers and revenue, and keep trial-to-paid for comparing yourself against apps that gate the same way you do.

The same logic covers the other gap between the two sources. Adapty's global trial-to-paid of 27.8% sits well under RevenueCat's overall median, and Adapty notes that short trials are increasingly common among the top performers in its sample. A 27.8% median lands just above RevenueCat's 25.5% band for trials of four days or less. A sample weighted toward three and seven day trials should produce roughly that number, so the two datasets are not in conflict. They have different trial-length mixes.

Three things I would test

Whether your trial length is causing your rate or just correlating with it. Apps that offer thirty-day trials are often different apps, sold to different people, so the median may not transfer to yours. The mechanism I find most plausible is the day-zero cancel: more than half of three-day trial cancellations happen before the app has really been used, and the user keeps access until the trial lapses anyway. A longer trial is more time for the app to become necessary before that door shuts, and more chances to reverse a decision someone made in the first thirty seconds. Test it by changing the length on your own app rather than adopting the median.

Whether the short-trial playbook is wearing out. Between the 2024 and 2026 reports, short trials fell from 31.2% to 25.5% while long trials only slipped from 44.9% to 42.5%. The gap widened from about 14 points to 17. Over the same period short trials became more popular. Those two facts together suggest a crowded tactic losing potency as users learn to cancel on sight.

Whether geography is thinning the funnel before price is even in play. Install-to-trial runs at 14.5% in North America against 7.6 to 10.2% elsewhere, so fewer people abroad reach your trial at all. What happens at the end of the trial is a separate problem, and I will come back to it.

How to read your own number

Put it together and you can actually diagnose your rate instead of panicking over it. Before you decide your conversion is bad, line it up against the right benchmark: your trial-length band, your paywall model on the correct metric, your platform mix, and your geography. A 20% trial-to-paid on a three-day Android trial serving emerging markets is a very different verdict than 20% on a 30-day iOS trial in the US. Same number, opposite story.

The price the trial converts into

One lever hides inside all of this, and it is the one I build for. A free trial does not convert into a signup, it converts into a price. If that price is wrong for the market, the trial-to-paid step breaks exactly where the funnel is already weakest, outside your home country, where install-to-trial is lower to begin with.

When you set one base price and let the stores convert it, the number a trialing user in India or Brazil is asked to accept at the end is often two or three times what the market can bear, so they cancel at the charge instead of converting. Localizing that price to local buying power gives the trial something acceptable to convert into. You can see your own per-country prices free, no login, and the parity pricing evidence covers why it moves revenue.

There is no magic number

There is no universal trial-to-paid rate to hit, but there are real ranges: roughly a quarter to over 40% depending on trial length, with platform and geography swinging it further. Measure against the right band, give trials enough time to build a habit, and check the price sitting at the end of the trial before you rewrite the paywall copy again. If you are still setting up the offer itself, introductory offers and pricing covers the mechanics, and how to price an app covers the number.

The price is usually the cheapest thing on that list to fix. If your trial converts fine at home and badly everywhere else, start there, because a US price run through a currency conversion is the most common reason a trial in India or Brazil ends at the charge screen instead of a subscription. You can check your own per-country prices in the free localizer in about two minutes, no login, and when you want them live, PricePush pushes the corrected prices to the App Store and Google Play in one step.