Hard paywall vs soft paywall: choose by revenue per install
Hard paywall vs soft paywall for a subscription app: what the 2025 benchmarks show, when each fits, and how to test the choice by revenue per install.
Use a hard paywall when your app proves its value in the first session, and a soft paywall when value takes days. Then settle it the way a growth lead reads Monday's dashboard: revenue per install for each version, not trial taps.
One note on terms: this post covers subscription apps and AI tools, not the metered article paywalls of news publishers.
What hard and soft paywalls mean in an app
RevenueCat defines the hard version plainly: "users must subscribe to access your app's features and content." Many hard paywalls include a free trial. The user still has to start that trial before using the app.
The soft version, in the same article, "fits into the freemium app model." Users get part of the app for free. They meet the paywall later, at a premium feature or a usage limit.
The real difference is where the paywall sits against the first moment of value. A hard paywall stands before it. A soft paywall stands after it. A closable onboarding offer sits in between: users see the price early but can still reach the product.
Install your app fresh this week. Write down every screen before the paywall and what a user can do without paying.
What the 2025 benchmarks say
The numbers below come from vendor reports. Each one describes that vendor's own customers, mostly consumer mobile apps, so read them as reference points, not targets.
RevenueCat's State of Subscription Apps 2026 covers 2025 data from over 115,000 apps. Its headline finding: "Apps that ask for money upfront convert 5x better than freemium (10.7% vs. 2.1%)." Both figures are median download-to-paid rates at day 35.
The medians hide a wide spread. RevenueCat's distribution runs from 4.2% to 38.7% for hard paywall apps, where 38.7% is the top 10%, and from 0.3% to 8.2% for freemium. RevenueCat reads the spread as a sign that "execution matters more than model alone." The ranges overlap, so a weak hard paywall app can convert less than a strong freemium one.
Revenue per install, total revenue divided by installs, points the same way. At day 60, hard paywall apps earned 8x more per install than freemium, $3.09 against $0.38. At day 14, the report compares hard paywalls with low-priced freemium apps only: a $2.32 median against $0.27. Different windows and comparison groups, so keep the two figures apart.
The early lead fades over a year. In the same dataset, freemium apps kept 28% of yearly subscribers after one year, and hard paywall apps 27%. A hard paywall collects more money sooner. It does not keep payers longer.
Adapty's report seems to say the opposite. Across 2025 data from 16,000+ apps, Adapty found that "Soft paywalls outconvert hard paywalls by nearly 50%", measured from paywall view to payment. The same report shows 21% higher one-year LTV for hard paywalls, measured per paying user. Adapty's LTV measures what each paying user spends over a year, not how long they stay, so it can be higher while retention stays level.
The two vendors measure different steps, so the findings do not conflict. A soft paywall is seen only by users who already engaged, so its view-to-pay rate is naturally higher. RevenueCat divides by every download, including people who never reached a paywall. Adapty also gives no definition of hard and soft, so its groups may not match RevenueCat's.
Pull your own download-to-paid rate at day 35 and revenue per install at day 60 this week. Those two numbers are the baseline for any paywall change.
When a hard paywall fits, and when a soft one does
Four questions set the starting model. Answer them with your own data before you look at anyone's benchmark.
How fast users see value
A hard paywall works when onboarding can show the result before the paywall, or when users reach it inside a short trial. A photo editor that returns an edited picture in minutes fits. A language or habit app that pays off after weeks fits a soft paywall better.
Shortening time to value is mostly an onboarding job. The steps in our guide to SaaS onboarding best practices apply to mobile apps too.
What a free user costs you
AI apps pay for compute on every free generation. A generous free tier then becomes a cost line, not a marketing channel. Gate the expensive actions behind a hard paywall, a trial or a small credit allowance. When free users cost almost nothing and bring in others, a soft paywall pays for its reach.
How much intent users bring at install
Someone who searched for "background remover" arrives with one job to do and accepts an early paywall. Someone who tapped a broad social ad is still curious. A wall before value loses many of them. Split installs by channel and keyword, because the right answer can differ per channel.
How fast paid acquisition must pay back
Timing matters when you buy installs. In RevenueCat's data, 1/3 of all conversions happen on day zero for both methods. For freemium apps, 23% of conversions do not occur until 6+ weeks after download. If you need payback within a month, that late revenue arrives after your budget decision. A hard paywall brings revenue forward, so you can judge campaigns sooner.
The diagram turns the first three questions into a starting point. Payback then tells you how long you can afford to wait for a soft paywall to earn back an install.
| Aspect | Hard paywall | Soft paywall |
|---|---|---|
| Where the paywall sits | Before first value, at install or onboarding | After first value, at a premium feature or usage limit |
| Fits apps where | Value shows in the first session and users install with clear intent | Value builds over days or weeks and free users cost little |
| Early revenue | Collects more money sooner | Many conversions arrive weeks after download |
| Main risk | Losing curious users before they see value | A generous free tier becomes a cost line when free use costs compute |
Answer the four questions in writing this week, one line each, with the data point behind each answer.
The middle options between hard and soft
Few apps run a pure version of either model. Three variants sit in between, and each can be tested on its own.
Onboarding paywall with a trial
In Adapty's data, "Onboarding paywalls with trials convert the highest at 1.78% on average." That is an average install-to-paid rate from a different dataset, so do not set it against RevenueCat's day 35 medians. Placement matters because 90% of trial starts happen on day 0 in the same report. A closable offer lets users who decline still reach the product and meet the paywall again later.
Trial length
In RevenueCat's medians, longer trials show higher trial-to-paid. The comparison is across apps, not a controlled test. Trials of 17 to 32 days converted 42.5% of trial starts to paid, 5 to 9 days 37.4% and 4 days or less 25.5%. Trial-to-paid divides by trial starts, not installs. Short trials also lose people at once: "55% of all 3-day trial cancellations happen on Day 0."
A longer trial also delays revenue, which hurts the payback math above. Test a 7-day trial against a longer one before you commit.
Subscription plus credits
AI apps often pair a subscription with credits or tokens. The subscription covers regular use. Credits cover heavy or occasional use without a monthly commitment, and they cap what each light user costs you.
Pixel+, an AI photo and video editing app, is one example. Its case page names the problem: "the paywall strategy was misaligned with user intent." Kreto redesigned the onboarding flow, including "a paywall architecture split between subscription and token-based models." The page reports 4,500+ active subscribers and subscriber momentum that turned positive for 2+ consecutive months after months of plateau.
List the features that cost you money per use this week. Those are candidates for credits, not for the free tier.
Apple rules to respect
On the App Store, a subscription must keep delivering value. Guideline 3.1.2(a) says "you must provide ongoing value to the customer, and the subscription period must last at least seven days". A one-time result, such as a single export, fits a one-off purchase or credits better.
The paywall must explain the deal. Apple's guidelines say "Before asking a customer to subscribe, you should clearly describe what the user will get for the price." For trials, Apple asks you to "clearly indicate how long the free trial lasts and the price billed once the free trial is over."
Plan trials around one more limit: "Customers can redeem one introductory offer per subscription group." A user who started the onboarding trial cannot get a second trial at a later soft paywall in the same group.
Apple's pages carry no date. We read them on October 15, 2026, so check them again before you ship. If you also ship on Google Play, check its policy separately. Screenshot your paywall this week and confirm the price, the trial length and the price after the trial are all visible.
How to test hard vs soft paywalls
Pick the metric before you build anything. Paywall conversion flatters the soft variant, because only engaged users see it. Read download-to-paid at day 35 and revenue per install at day 14 and day 60. Add refunds and second-month renewals once that data arrives.
Kreto runs paywall tests like this as part of product marketing work, because the paywall sets what you can pay per install. The plan has six parts:
- Variants: run your current paywall against one alternative. Change placement only, not price, so you know what moved the result.
- Assignment: assign each new install to one variant at random, server side, and keep it there. Never switch an existing user.
- Channels: tag every install with its acquisition channel. A hard paywall can win on search traffic and lose on broad social traffic, so read each channel before the blend.
- Sample: set the sample size before launch with a standard power calculator. Size it for your decision metric. Use the mean and spread of day 60 revenue per install from recent cohorts, plus the smallest lift worth acting on.
- Duration: let each cohort reach day 60. Early reads favor the hard paywall, because many freemium conversions arrive late (see the payback section).
- Decision rule: write it down first, for example "pick the variant with higher day 60 revenue per install in our two largest channels."
Keep ad spend and creative steady while the test runs. If the channel mix shifts mid-test, the blended result stops matching the traffic you will buy later.
Write the plan on one page this week: variants, metric, sample, channels and read dates.
Frequently asked questions
Is a hard paywall better than a soft paywall?
Neither wins everywhere. Hard paywalls collect revenue sooner in the benchmarks above, and soft paywalls suit apps whose value builds over weeks. Run both on new installs and pick by revenue per install at day 60.
Can an AI app use a soft paywall?
Yes, if free usage is capped. Offer a small credit allowance or a limited feature set, and gate the expensive generations. Track compute cost per free user on the same dashboard as conversion.
How long should a paywall test run?
Long enough for each cohort to reach day 60, plus the time to fill your sample. Treat day 14 as an early signal only.
Make the paywall a test, not a belief
The paywall is one step in a funnel that starts with the ad and ends with the renewal. Choose a starting model with the four questions, then let revenue per install decide.
If you want help setting up a hard vs soft paywall test and reading the result by channel, talk to Kreto. We can go through your funnel data with you on a discovery call.