Hard Paywall vs Freemium: Which Model Fits Your Indie App?

Updated Oct 10, 2026

On this page (7 sections)

1. Quick answer
2. Definitions: hard, soft, freemium
3. What the data says (cited)
4. When freemium still wins
5. AI apps and serving costs
6. Decision checklist
7. FAQ

A hard paywall asks users to subscribe (often with a free trial) before they can use the app; freemium lets them use a free tier and pay to unlock more. RevenueCat's 2026 data shows hard paywalls convert about five times better at the median, with nearly identical one-year retention. Freemium still wins when free users drive word of mouth, network effects or long-term scale.

Quick answer

If your app's value is clear in the first session and your audience is focused, test a hard paywall, usually paired with a trial. If your app grows through sharing, needs time to show value, or depends on a large free audience, start freemium. Either way, decide with your own revenue per install, not the median.

Definitions: hard, soft, freemium

What is a hard paywall?

A hard paywall blocks the app's core features until the user starts a subscription or trial, usually right after onboarding. The user can't keep using the main product for free. Many hard-paywall apps pair it with a free trial so people can try everything before the first charge, which lowers the risk of committing.

Soft paywalls sit between the two: users can close the paywall and continue with limited access, so in practice a soft paywall is usually a freemium app with an early, prominent upgrade screen.

What the data says (cited)

Do hard paywalls convert better than freemium?

At the median, yes. RevenueCat's State of Subscription Apps 2026 reports a median download-to-paid conversion by day 35 of 10.7% for hard paywall apps versus 2.1% for freemium apps, about five times higher. One-year retention was nearly identical: 27% for hard paywall and 28% for freemium.

Source: RevenueCat, State of Subscription Apps 2026 and the 10-minute summary.

Caveats worth keeping in mind:

• These are medians. RevenueCat also notes hard paywalls show wider variance, so some do far worse than the median.
• Apps choose their model for reasons (category, audience, product), so the gap isn't purely caused by the paywall.
• Similar retention means the hard paywall's advantage is mainly earlier and higher conversion, not stickier subscribers.

When freemium still wins

RevenueCat itself says freemium remains the right call when free users drive word of mouth, network effects or long-term brand scale. (RevenueCat) In practice, freemium tends to fit when:

• The product spreads through use. Shared lists, invites, collaborative features: free users bring in other users.
• Value takes longer to show than a trial lasts. If the "aha" moment comes after weeks, a trial can expire first. RevenueCat makes the same point in its freemium tier design guide.
• You want a large top of funnel, for example to rank for broad terms through ratings volume or to build a brand.
• You can afford to wait for revenue. Freemium conversions arrive later, so you need runway.

When should an app be freemium?

Choose freemium when free users create value for you: they invite others, share content, leave ratings, or need weeks to see the app's benefit. It also suits broad audiences where reach matters more than early revenue. If none of that applies and value is clear quickly, a hard paywall with a trial is worth testing.

AI apps and serving costs

If every action in your app costs you money (AI model calls, heavy processing, cloud storage), the free tier is not free for you. A few rules of thumb:

• Cap the free tier tightly, for example a small number of AI actions per day, so free usage can't outrun your budget.
• Consider a hard paywall with a trial if the core feature is the expensive one. The trial still lets people try it.
• Watch abuse. Free tiers on costly features attract scripted or repeated sign-ups.
• Price lifetime plans carefully, or skip them. One payment against ongoing costs is a risk.

Build this into your numbers: revenue per install minus serving cost per install is the figure that decides whether a model works.

Decision checklist

Lean hard paywall if most of these are true:

• The core value is clear in the first session or two
• Your audience is focused and has a specific problem
• Each free user costs you real money to serve
• You need revenue soon

Lean freemium if most of these are true:

• Free users invite others or create shareable content
• The app's value builds over weeks
• You're targeting a broad audience and reach matters
• Serving a free user is cheap and you have runway

Then test. Run the alternative on a share of new users, keep the paywall design consistent, and compare revenue per install over a fixed window. See choosing an A/B testing method, and use user research to learn why people don't convert. Engagement also feeds store visibility; see retention and engagement signals.

GROWNX pulls your App Store Connect and RevenueCat data into one free-to-start dashboard, so you can see whether your keyword, metadata and paywall changes actually move revenue. Free plan available.

FAQ

Does Apple allow hard paywalls?

Yes. Apple's App Review Guidelines allow apps that require a subscription, as long as the subscription provides ongoing value and you clearly explain what users get, how much it costs and how it renews before they buy. Check the current App Review Guidelines section 3.1.2 before you submit.

Can I switch models later?

Yes, but plan it. Moving from freemium to a hard paywall can upset existing free users, so many apps keep current users on their old access and apply the change to new installs only. Moving to freemium is easier for users but changes your revenue curve. Test on new users first.

Is a soft paywall a good middle ground?

Often. A dismissible paywall during onboarding shows the offer to everyone at their most interested moment, while still letting people explore. It behaves like freemium for users who close it, so judge it on revenue per install against both alternatives, not against conversion alone.