A hard paywall makes paid acquisition easier on almost every mechanical measure. RevenueCat's State of Subscription Apps 2026 puts the median download-to-paid rate within 35 days of install at 10.7% for hard paywall apps against 2.1% for freemium, and revenue per install at day 60 at $3.09 against $0.38. That means more paying users per pound, more purchase signal for Meta to learn on, and an answer on each campaign within days. What a hard paywall does not do is produce better subscribers: year-one retention is almost identical across the two models. So the choice is about speed and signal, not customer quality, and freemium is still right for apps whose free users drive their own growth.
This page looks at the paywall decision from the ad account rather than from the product. It covers what each model does to your cost per paying user, to the budget Meta needs before it can learn, to attribution and reporting windows, and to the creative you have to make. If you want the benchmark itself in full, with its distribution and caveats, our download-to-paid conversion benchmarks cover it. This page is about what the gap means once you are paying for installs.
Key takeaways
- Hard paywall apps convert about five times as many installs to paid within 35 days (10.7% against 2.1% median) and earn about eight times the revenue per install by day 60.
- Retention and refunds barely differ between the models, so the advantage is volume and speed of payment, not subscriber quality.
- More paying users per install means Meta can optimise on a deep event at a fraction of the budget a freemium app needs.
- Almost a quarter of freemium conversions arrive six or more weeks after download, against about 15% for hard paywalls, after Meta's longest standard click window and Apple's last attribution window have closed.
- The medians describe apps that chose each model. Switching model will not hand your app the other side's number, so test before you commit.
What counts as a hard paywall and what counts as freemium
A hard paywall means a new user cannot use the core product without starting a subscription or a free trial. Many hard paywall apps include a trial, so "hard" describes the gate, not the absence of a free period. Freemium means a useful version of the product is available without paying, with premium features or limits behind an upgrade.
Most apps sit somewhere between. A freemium app that shows a closable paywall during onboarding, often called a soft paywall, behaves partly like each. RevenueCat segments apps by access method into two groups, freemium and hard paywall, so its figures describe the two ends and say little about the hybrids in the middle. Keep that in mind if your app shows a paywall on day one but lets people dismiss it.
Hard paywall vs freemium: the 2026 numbers
Every figure below comes from RevenueCat's State of Subscription Apps 2026, which covers more than 115,000 apps. They are medians across apps and include organic as well as paid installs.
| Metric | Hard paywall | Freemium | Notes |
|---|---|---|---|
| Download-to-paid within 35 days of install (median) | 10.7% | 2.1% | About 5x. Hard paywall top quartile above 20.0%, top 10% at 38.7%. Freemium top quartile above 4.5%. |
| Revenue per install at day 14 (median) | $2.32 | $0.27 | About 8.6x. Total revenue divided by total installs. |
| Revenue per install at day 60 (median) | $3.09 | $0.38 | About 8.1x. Freemium grows slightly more between day 14 and day 60 (1.41x against 1.33x). |
| Share of conversions arriving in week 6 or later | 15.3% | 22.9% | Both models see more than 60% of conversions within a week. Hard paywalls spike in days 4 to 7 (25.7%), driven by 7 day trials ending. |
| Year-one retention, yearly plans (median) | 27% | 28% | Monthly plans 9% against 8%. RevenueCat calls the gap noise. |
| Refund rate (median) | 2.5% | 2.9% | RevenueCat calls this difference probably noise as well. |
Two other figures from Adapty's State of In-App Subscriptions 2026 sit alongside these. Adapty finds soft paywalls convert nearly 50% better than hard paywalls from paywall view to payment, and hard paywalls produce 21% higher average one-year LTV. The first sounds like it contradicts RevenueCat and does not. Its denominator is paywall views, and a hard paywall is shown to every install, including the ones that came for a look. Adapty's figures are averages of per-app rates, mostly from the App Store, so they should not be mixed into the same calculation as RevenueCat's medians.
Why a hard paywall makes paid acquisition easier
More paying users per pound of media
At the same cost per install, a fivefold difference in download-to-paid is a fivefold difference in media cost per paying subscriber. That is the number that has to fit inside what a subscriber is worth. If you have worked out your break-even CPI, the paywall model is the biggest single input to it. A freemium app needs either much cheaper installs or much more valuable payers to reach the same break-even, and in our experience it rarely gets the first without giving up some of the second.
Enough signal for Meta to learn on
Meta's learning phase guidance says an ad set usually exits learning after about 50 results in the week after its last significant edit. The paywall model decides how many of those results a budget can buy. The example below is hypothetical, with round numbers chosen to show the shape. Neither app is a client.
Both apps run one ad set at £150 a day, about £1,050 a week, at a cost per install of £2.50, which buys 420 installs a week. App A has a hard paywall with a 7 day trial: 15% of installs start a trial and 8% of installs pay within 35 days, 65% of them inside the first week. App B is freemium: 6% start a trial and 2% pay within 35 days, 60% of them inside the first week. Both paid conversion rates sit below RevenueCat's medians, because paid installs usually convert below a benchmark that includes organic users.
| Per ad set | App A, hard paywall | App B, freemium |
|---|---|---|
| Installs per week | 420 | 420 |
| Trial starts per week | 63 | 25 |
| Paid conversions per week, inside the first 7 days | About 22 | About 5 |
| Monthly spend for about 50 trial starts a week | About £3,600 | About £9,000 |
| Monthly spend for about 50 paid conversions a week | About £10,400 | About £45,100 |
| Media cost per paying subscriber by day 35 | £31.25 | £125 |
App A can optimise on trial starts from its first week and could move to optimising on purchases at around £10,000 a month on a single ad set. App B needs about £9,000 a month just to learn on trial starts, and about £45,000 to learn on purchases at the same install cost. Even if App B's installs cost a third less, at £1.70, purchase optimisation still needs about £30,700 a month. So in practice a freemium app on a modest budget optimises for installs, trial starts or a product event that predicts paying, and accepts that Meta is aiming at a proxy. Our guide to the Meta Ads learning phase covers what happens to an ad set that never gets there.
A faster, cleaner answer on every campaign
RevenueCat finds a pronounced spike in hard paywall conversions in days 4 to 7, which it attributes to 7 day trials ending. A hard paywall cohort has shown most of what it will ever show within a week or so. That shortens every loop that matters: how quickly you can call a creative test, how soon you know whether a new market works, and how fast spend comes back as revenue to be spent again. Our guide to the CAC payback period shows why the speed of the money matters even when the total does not change.
What a hard paywall does not fix
The retention figures are the part of this comparison that gets left out of most arguments for gating. Median year-one retention on yearly plans is 27% for hard paywall apps and 28% for freemium, and refund rates are 2.5% and 2.9%. RevenueCat's own reading is that access method is "a conversion choice, not a retention strategy." A hard paywall gets more of the people who would pay to pay, and gets them to pay earlier. It does not appear to make them stay longer.
The conversion gap itself also needs care. Apps that choose a hard paywall are not a random sample. They tend to have a single, specific promise that can be sold before anyone has used the product, and that clarity is part of why people pay. RevenueCat notes the hard paywall range runs from 4.2% to 38.7% and reads the spread as a sign that execution matters more than the model. Put a gate in front of a product with a vague promise and the likely result is the bottom of that range, plus fewer installs.
Where freemium makes paid acquisition harder, and what to do about it
Part of the revenue never reaches the ad platform
RevenueCat finds 22.9% of freemium conversions arrive in week six or later, against 15.3% for hard paywall apps. The longest click window in Meta's standard attribution settings is 7 days, and the longest view window is 1 day. Apple's SKAdNetwork documentation sets three conversion windows that end on day 35 after first launch. A payment in week six falls outside all of them. Those subscribers are real revenue that no ad will ever be credited with, so Meta optimises without them and its ROAS column understates the campaign. Both models lose some revenue this way. A freemium app loses about one and a half times as large a share, from a base that is already five times smaller. On iOS the postback delays add a lag on top of the missing tail.
The fix is to stop asking the platform for the full answer. Judge ads against each other inside Meta, where the missing tail hits every ad roughly equally, and judge budgets on cohort revenue from your subscription data. Our subscription app ROAS benchmarks work through a hard paywall app and a freemium app that look opposite at day 7 and identical at day 365.
Cheap installs are easier to buy and harder to read
A free app is an easier yes, so a freemium app can often buy installs for less. The risk is that creative which wins on cost per install is winning on curiosity, and curious installs are the ones least likely to pay. With a hard paywall that problem shows up on day one as a weak trial rate. With freemium it hides for weeks. The earliest honest read for a freemium app is usually an in-product event that predicts payment, and if you only watch CPI, the cheapest concept will look like the best one. Our install-to-trial benchmarks explain how to tell an ad problem from an onboarding problem at that stage.
The creative has a different job
This is the part we spend most time on with clients, and it is where the paywall decision reaches the ad account most directly. Behind a hard paywall the ad has to sell the outcome, because the next thing the user meets after onboarding is a price. The promise in the ad and the promise on the paywall need to be the same promise, or the install rate holds while the trial rate falls. Behind freemium the ad only has to sell the first session, which allows broader, more curious angles, but the product then has to do the selling the ad did not. Neither is easier creatively. They fail in different places, and the hook patterns that work for one are not automatically right for the other.
When freemium is still the right choice
RevenueCat's own summary of the debate says freemium remains the right call when free users drive word of mouth, network effects or long-term brand scale. That is the important point for a paid acquisition team, because none of that value appears in a cohort ROAS report. A free user who invites two friends, shares a result or leaves a five star rating is worth something, and the paid-media view of the business will always undercount it.
The table below is how we would frame the decision for a subscription app that intends to grow mainly through paid social. It is a judgement framework, not a rule.
| If this describes your app | Paid acquisition is usually easier with | Why |
|---|---|---|
| The outcome can be shown in a 20 second ad and believed before anyone opens the app | Hard paywall, usually with a trial | The ad and onboarding can do the selling, so there is little to gain from letting people in free. |
| Free users invite others, create content or make the product more valuable for everyone | Freemium | The value of a free user sits in organic installs that paid reporting never credits. |
| Value only becomes obvious after several sessions or days of use | Freemium, or a hard paywall with a long trial | Gating before value is felt asks people to pay on faith, which only the strongest promises survive. |
| Meta is your main acquisition channel and monthly spend is modest, say under £10,000 | Hard paywall makes this much easier | At low budgets, only a hard paywall is likely to generate enough deep events for Meta to learn on. |
| Serving each free user has a real cost, such as AI features | Hard paywall or a tight free tier | A contributor to RevenueCat's report notes AI apps are making free products less generous for this reason. |
| You need cash back from campaigns within weeks to keep spending | Hard paywall | Revenue lands earlier, so the same budget recycles faster. |
The pattern underneath is simple. The more of your product's value an ad can carry, the more a hard paywall helps paid acquisition. The more of it depends on using the product, or on other users, the more a hard paywall taxes the growth that paid acquisition cannot see.
How to test a switch without fooling yourself
If you are considering moving from freemium to a hard paywall, or adding a stronger onboarding paywall, these are the rules we would apply.
- Measure per install, not per paywall view. Paywall conversion will almost certainly fall when every user sees the paywall. What matters is paying users and revenue per install.
- Wait for the freemium tail. Compare cohorts on download-to-paid at day 35 and revenue per install at day 60, or you will credit the hard paywall with revenue the freemium cohort simply had not collected yet.
- Hold the ads constant. Run the same creative and the same optimisation event through the test, or you are testing two changes at once. Expect a change of optimisation event afterwards to restart learning.
- Watch what paid reporting cannot see. Track organic installs, ratings and referral activity through the test. A gate that wins on paid cohorts and quietly halves word of mouth has not won.
- Check the first renewal before declaring a winner. RevenueCat's medians say retention should hold. Your app is not a median, and the first renewal is where a mis-sold subscription shows itself.
How we approach it with subscription app clients
We do not set clients' paywalls, and the access model is a product decision before it is a marketing one. What we do on our Meta Ads service is build the account around whichever model the app has: the optimisation event the budget can support, reporting windows that match when the money actually arrives, and creative whose promise matches the paywall the user meets next. Where the funnel after install is the constraint, we say so. Our Steps & Beasts case study is an example: the 1112% revenue increase came from heavy creative testing and an onboarding overhaul that lifted install-to-subscriber conversion, not from ads alone.
The paywall model is one of five decisions that shape a subscription app's paid growth, and our subscription app marketing strategy guide covers the other four. If your app is already spending on Meta and you want the account built around how your paywall actually converts, see how we work with consumer and subscription apps. We work with accounts spending at least £3,000 a month on Meta, and current fees are on our pricing page.
Frequently asked questions
Do hard paywalls convert better than freemium?
Per install, yes, by a wide margin. RevenueCat's State of Subscription Apps 2026 puts the median download-to-paid rate within 35 days of install at 10.7% for hard paywall apps and 2.1% for freemium apps. Per paywall view the picture flips: Adapty's 2026 report finds soft, closable paywalls convert nearly 50% better from paywall view to payment, because a hard paywall is shown to every install, including the ones that were never going to pay. For paid acquisition the per-install figure is the one that pays for media.
Do hard paywall subscribers retain better than freemium subscribers?
Not according to RevenueCat's 2026 data. Median year-one retention on yearly plans is 27% for hard paywall apps and 28% for freemium apps, and 9% against 8% on monthly plans. Refund rates are similar too, 2.5% against 2.9%. A hard paywall gets more people to pay, sooner. It does not appear to change how long they stay.
Which paywall model is better for Meta ads?
A hard paywall is easier to run on Meta because it produces more paying users per install and most of them pay within the first week, so the algorithm sees enough purchase or trial events to learn on at a lower budget. A freemium app can still run profitably on Meta, but usually has to optimise for a shallower event such as trial start or a product milestone, and needs longer reporting windows because a share of its paying users convert after Meta and Apple have stopped attributing.
Should I switch my freemium app to a hard paywall?
Test it rather than assume it. The 10.7% median comes from apps that chose a hard paywall, which tend to have a promise clear enough to sell before anyone has used the product. A freemium app with a vague value proposition will not inherit that rate. Run the change as a measured test, judge it per install on download-to-paid and revenue per install at day 60, and watch organic installs and ratings, because a gate can cost you word of mouth that never shows up in paid reporting.
How long should I wait before judging a freemium campaign?
Longer than a hard paywall campaign. RevenueCat finds more than 60% of conversions happen within a week for both models, but 22.9% of freemium conversions arrive in week six or later, against 15.3% for hard paywall apps. Use an early window for creative decisions, as long as it ranks your past cohorts the same way their mature results did, and use day 60 or later for budget decisions.
The short version
On paid acquisition alone, a hard paywall is the easier model. About five times as many installs pay within 35 days, revenue per install is about eight times higher by day 60, Meta gets enough deep events to learn on at a realistic budget, and most of the answer arrives within a week. It buys speed and signal, not loyalty, because retention and refunds look the same either way. Freemium is harder to run on Meta and needs longer windows and a proxy event, and it is still the right model when free users create growth that no ad report will ever show you. Decide on the product first, then build the account around the choice.