Superwall
RevenueCatMobile app conversion rate optimization is the practice of raising the share of people who move through each step of an app's money path: store page view to install, install to finished onboarding, onboarding to trial or purchase, and trial to renewal.
For subscription apps, the biggest measured wins usually sit at the paywall and in onboarding, not on the store page.
The fastest way to find them is to test one step at a time and judge each test on revenue per install, not on clicks.
This guide covers the funnel and its metrics, what teams actually test, how app CRO differs from ASO and user acquisition, how small apps can test without huge traffic, and when to bring in tools or outside help.
Every example links to a real test in our app test database.
What app CRO is, in one paragraph
App CRO is the app version of conversion rate optimization.
A website has one main conversion, usually a checkout.
A subscription app has a chain of them: the store page has to earn the install, onboarding has to earn attention, the paywall has to earn a trial or purchase, and the product has to earn the renewal.
Improving any link raises revenue, but a gain at one step can be wiped out at the next.
That last point is what separates app CRO from button-color testing.
Zumba tested a web checkout button on its paywall and saw immediate conversions fall 25% while LTV rose 17% (vendor-reported by RevenueCat).
Judged on the first step alone, it looked like a loss. Judged on the whole chain, it won.
The reverse also happens. BoldVoice tested a paywall with no dismiss button and got more trial starts, but refunds rose by more than the gain.
The step metric went up and the business went down.
The app funnel and the metrics that matter
Most subscription apps can describe their funnel in five steps.
Each one has its own metric, and each metric needs a clear window (for example, "within 35 days of install") so numbers are comparable over time.
| Step | Main metric | Where it is measured |
|---|---|---|
| Store page | Page view to install (store conversion rate) | App Store Connect, Google Play Console |
| Onboarding | Install to onboarding completion | Product analytics |
| Paywall | Paywall view to trial or purchase | Subscription SDK or paywall tool |
| Trial | Trial to paid | Subscription backend |
| Renewal | Renewal rate, churn, refunds | Subscription backend |
Two summary metrics tie the chain together. Install-to-paid conversion tells you what share of new users ever pay.
Revenue or proceeds per install (often called ARPU at a fixed day, such as D30 or D90) tells you what each install is worth.
We explain the formulas in our guide to subscription metrics.
What "normal" looks like
RevenueCat's State of Subscription Apps 2026, built on data from over 115,000 apps, reports a median D35 download-to-paid rate of 2.6% in North America and 1.4% in India and Southeast Asia.
The same report finds hard paywall apps convert a median 10.7% of downloads to paid within 35 days, against 2.1% for freemium apps.
Trial length matters too. In RevenueCat's 2026 benchmark summary, trials of 17 to 32 days convert to paid at a median of 42.5%, while trials under 4 days convert at 25.5%.
The report also says 55.4% of 3-day trial cancellations happen on Day 0.
Benchmarks are a sanity check, not a target.
Category, price and region move them a lot.
For more detail, see our pages on App Store conversion rates and trial conversion rates.
Pick one north-star metric per test
Choose the metric closest to money that your traffic can still measure.
For a paywall test, that is usually proceeds per paywall viewer or per install.
Trial starts alone can mislead: the wellness app in Daphne Tideman's test saw a 14-day trial bring slightly more trial starts but fewer paid conversions than a 7-day trial.
What subscription apps test
The tiles above this article link to every surface hub.
Here we cover why each surface tends to produce the size of results it does, and what kind of change usually moves it.
Paywalls: the most tested surface
Most of the tests in our app database are paywall tests, because the paywall sits right at the point of payment and changes ship without an app update when a remote paywall tool is in place.
Common levers are plan mix, trial offers, price display, layout and timing. Our paywall hub groups them by element.
Two patterns stand out. First, removing options often helps.
Mojo saw +14% new subscription revenue after hiding credit packs from the paywall (vendor-reported by Superwall).
Second, adding a trial where there was none can be large: Stardust saw +131% paywall conversion from adding a 7-day trial in a multivariate test (vendor-reported by Superwall).
Not every simplification wins. Stardust also tested a yearly-only paywall, and it lost on both conversion and proceeds per user.
The same app, the same team, opposite results for two "fewer choices" ideas.
Paywall timing and placement
Where the paywall appears can matter as much as what it says.
Greg reported about +400% trial sign-ups, with sign-up-to-trial rising from 3% to 15%, after adding a paywall to onboarding and gating features.
That was a bundle of changes, so the placement alone doesn't get all the credit.
A three-way test shared by Rosie Hoggmascall compared paywall position and an added feature carousel, with trial opt-in of 2% for version A, 8% for B and 15% for C.
Results like these are why many teams test placement before they test design.
Placement changes also change who sees the paywall.
Showing it earlier means more people see it, but some of them have not yet seen why the app is worth paying for.
Watch trial-to-paid and refunds after a placement change, not only trial starts.
Pricing and plans
Price tests often lower conversion while raising revenue.
BuyBye raised its yearly price 50% and saw 19% fewer conversions but 25% more proceeds per user (vendor-reported by Superwall).
This is why price tests should never be judged on conversion rate alone.
Our subscription pricing hub covers anchoring, plan length and regional pricing.
Onboarding
Onboarding tests tend to change who reaches the paywall and in what frame of mind.
Lose It! found a longer onboarding raised trial start rate by double digits, which runs against the common advice to cut steps.
ASL Bloom, working with the agency Applica, reported +15% trial conversion, +12% ARPU and +18% 30-day retention from goal-based onboarding.
That result came from a bundle of changes, so no single screen gets the credit. See the onboarding hub for more.
Retention and win-back
Retention tests are fewer, but some show the largest gaps.
Unscripted saw over +240% trial starts from a reactivation trial offer (vendor-reported by Superwall).
Lose It! reported about 10x more money from in-app message offers compared with its previous approach.
The retention hub covers churn, win-back and renewal tests.
Where the payment happens
Web checkouts outside the app stores are a newer area, and the results are mixed.
Superwall's own early app-to-web tests showed initial conversion down 11.9% but net proceeds up about 19.85%.
RevenueCat's test with Dipsea went the other way: a web-only paywall dropped initial conversion from 27.0% to 18.1%.
Our web-to-app guide covers when the trade can pay off.
App CRO vs ASO vs user acquisition
These three jobs overlap but answer different questions.
User acquisition (UA) asks how to bring more people to the store page at a sustainable cost.
App store optimization (ASO) asks how to rank for searches and turn store page views into installs.
App CRO asks how to turn those installs into paying, renewing subscribers.
| User acquisition | ASO | App CRO | |
|---|---|---|---|
| Main goal | More qualified visitors | Rankings and installs | Revenue per install |
| Typical levers | Ad creative, channels, bids | Keywords, icon, screenshots | Onboarding, paywall, pricing, offers |
| Testing tools | Ad platforms | Apple product page optimization, Google Play store listing experiments | Paywall and subscription tools |
| Feedback speed | Days | Weeks | Days to months (renewals) |
They feed each other. Better in-app conversion raises the value of each install, which lets UA spend more per user.
Better store conversion lowers the cost per install, which gives CRO tests more traffic.
The store side has its own testing rules.
Apple's product page optimization lets you test up to three treatments of the app icon, screenshots and app preview videos.
A test runs for up to 90 days, and new metadata must be approved by App Review first.
Google Play's store listing experiments allow up to two variants against the current listing, and you can run one default graphics experiment or up to five localized experiments at a time. Our ASO hub goes deeper on both.
One caution: a store page that promises something the app doesn't deliver can raise installs and lower trial conversion.
Check downstream metrics after a store test, not just install rate.
How to start with a small app
Most apps don't have the traffic of the brands in case studies.
That doesn't rule out testing, but it changes what you test and how you read results.
Check whether a test can reach a result
A common rule of thumb for an A/B test at 95% confidence and 80% power gives the sample per variant as:
n ≈ 16 × p × (1 − p) ÷ d²
Here p is your baseline conversion rate and d is the absolute change you want to detect.
Say 5% of paywall viewers start a trial, and you want to detect a lift to 6% (a 20% relative gain, so d = 0.01).
That gives 16 × 0.05 × 0.95 ÷ 0.0001 = 7,600 paywall viewers per variant, or about 15,200 for a two-way test.
If your paywall gets 500 views a day, that test takes about a month.
If you want to detect a 10% relative lift instead, the sample roughly quadruples.
Our app A/B testing guide covers sample size, test length and stopping rules in more detail.
Practical rules for low traffic
- Test big changes, not tweaks. A new plan mix, adding or removing a trial, or moving the paywall in onboarding can produce gaps large enough to see. A new button color usually can't.
- Test where the traffic is. More people see the onboarding paywall than any later screen. Start there.
- Use a metric you can measure soon. Trial-to-paid takes a week or more to show up. With little traffic, use paywall conversion plus a check on refunds and early cancellations.
- Run one test per surface at a time. Overlapping tests on the same users make both results hard to read.
- Write down the decision rule before you start. Decide the metric, the minimum sample and what result would make you ship.
A simple order of work
- Instrument the funnel. Make sure you can see each step from install to renewal, split by platform and country. Without this, no test result can be trusted.
- Find the biggest drop. Compare your numbers with the benchmarks above. The step furthest below the norm is usually the best place to start.
- Fix the business model questions first. Hard paywall or freemium, trial or no trial, and which plans to offer tend to move revenue more than any design change. Our guide to paywalls covers these choices.
- Then test the paywall and onboarding details. Copy, layout, plan display and social proof come after the structure is right.
- Return to retention. Once new users convert well, renewal and win-back offers protect what you have earned.
Common mistakes
- Stopping a test early. Small samples swing a lot in the first days. A test that looks like a big winner on day three often settles near zero.
- Judging on trial starts alone. As the BoldVoice and wellness app tests show, more trials can mean more refunds or fewer paid users.
- Mixing platforms. iOS and Android users often behave differently. Read results for each platform before you combine them.
- Copying a winner without testing it. A result from a dating app may not hold for a meditation app at a different price.
Treat before-after results with care
Many small-app wins in our database are before-after comparisons, not A/B tests.
Rootd reported 5x revenue after moving its paywall to the start of onboarding.
That is a strong signal, but seasonality, ad spend and app updates can all shift numbers over the same period.
Each test page shows its evidence type so you can weigh it.
When you copy an idea from a case study, test it rather than ship it.
The same change can win in one app and lose in another, as the two Stardust tests above show.
Tools and when to hire help
You need three things to run app tests: a way to change the paywall or flow without an app release, a way to split users into groups, and revenue data tied to each group. Subscription platforms bundle all three.
Two widely used options:
- RevenueCat handles subscription infrastructure and includes experiments and paywalls. As of November 2026, its pricing page lists the Pro plan as free up to $2,500 in monthly tracked revenue, then 1% of tracked revenue. Its best fit is teams that want billing, analytics and testing in one place. It is less of a fit if you want deep, design-heavy paywall testing and already run your own backend, though it now offers a growth-tools option for that case.
- Superwall focuses on remote paywalls and paywall experiments. As of November 2026, its pricing page lists subscription infrastructure as free and charges 1% of paywall-attributed revenue once that revenue passes $10,000 a month. It suits teams that test paywalls often. It is not built for store page tests or broad product analytics, so you will still need other tools for those.
Both vendors publish many of the case studies in our database, so treat their numbers as vendor-reported.
We have not used either tool ourselves. For a wider comparison, see our roundup of paywall tools.
Whatever paywall tool you pick, you will also want product analytics for onboarding steps and the store consoles for page-to-install data.
Many small teams start with the free tier of a subscription platform plus the store consoles and add more only when a question can't be answered.
When outside help makes sense
Hiring help tends to pay off in three situations: you have enough traffic to test but nobody owns the roadmap, your tests keep coming back flat and you need fresh hypotheses, or you are making a big structural change such as moving from freemium to a hard paywall.
Several tests in our database were run by agencies and consultants, such as Applica's work with Five Minute Journal, which reported +20% ARPU from a jobs-to-be-done onboarding.
Before you hire, ask to see past tests with their evidence type and the metric they were judged on.
A good partner will talk about proceeds per install and refunds, not just conversion rate.
We are building an app growth experts directory for this.
If your app also sells on the web, the same discipline applies there.
Our mobile CRO guide covers mobile web checkout and landing pages.