The best mobile app A/B testing tools fall into three groups: in-app experiment platforms (Firebase A/B Testing, Statsig, Optimizely Feature Experimentation), paywall testing tools (Superwall, RevenueCat Experiments, Adapty) and store listing tools (Apple's Product Page Optimization, Google Play store listing experiments, SplitMetrics).
Most subscription apps need one tool from the paywall group plus the free native store tests, and add an in-app platform only when they test features beyond the paywall.
We have not used these tools ourselves. This page compares them from their public docs and pricing pages, and shows what teams that ran tests with them actually found, using results from our app test database.
In-app experiments: Firebase A/B Testing, Statsig, Optimizely
These tools test anything your code can switch: onboarding steps, feature defaults, copy, layouts and push messages.
You ship the variants in your app build and use a remote flag or config to decide who sees what.
For the method itself, see our guide to A/B testing a mobile app.
Firebase A/B Testing
Firebase A/B Testing runs on top of Remote Config for UI and feature changes and on Cloud Messaging for notification tests, per Google's docs.
It is a no-cost product on both the Spark and Blaze plans, according to the Firebase pricing page.
Limits matter once you test often. Each experiment allows up to 8 variants including the baseline, and a project can have 24 running experiments at once, shared with Remote Config rollouts, per the Firebase docs.
Results use frequentist inference at a 0.05 significance level.
Downsides: it measures events you log in Google Analytics, so subscription revenue is only as accurate as your purchase tracking.
It has no paywall builder, so every paywall variant needs to be built into the app first.
Not for: teams that mainly want to test paywalls and prices without engineering work on each variant.
Statsig
Statsig combines feature flags, experiments, product analytics and session replay.
As of November 2026, its pricing page lists a free Developer tier with 2M metered events a month, a Pro tier from $150 a month with 5M events and $0.05 per 1,000 extra events, and custom Enterprise pricing. Experiments are included on every tier.
OpenAI agreed to buy Statsig in September 2025, and Statsig said it would keep serving existing customers, as SiliconANGLE reported.
Weigh that ownership change when you plan for the long term.
Downsides: exposures count as metered events, so a busy app can outgrow the free tier quickly.
It does not handle purchases or receipts, so revenue metrics depend on piping in data from your subscription backend.
Not for: solo developers who only need to test one paywall at a time.
Optimizely Feature Experimentation
Optimizely Feature Experimentation offers SDKs for Swift, Android, React Native and Flutter, per its developer docs.
It suits larger product teams that already use Optimizely on the web and want one experiment program across platforms.
Downsides: pricing is not public. The pricing page says every plan is individually packaged and asks you to talk to sales.
Not for: small and mid-size subscription apps. If you are comparing web tools too, see our A/B testing tools roundup and Optimizely alternatives.
Real example: the AEZAKMI Group tested six onboarding flows in an LED controller app and moved conversion from 0.8% to 1.29% (+61%), with ARPU from $0.12 to $0.14. That is the kind of multi-variant onboarding test these platforms (or a subscription backend with experiments) handle.
Paywall testing: Superwall, RevenueCat Experiments, Adapty
Paywall tools let you change the paywall, its products and prices from a dashboard, without an app release.
They also know who paid and renewed, so they report revenue per user rather than just taps.
For the wider tool list, see our paywall tools roundup.
Superwall
Superwall builds tests from campaigns, audiences and placements.
Inside an audience you set a split such as 70% paywall A and 30% paywall B, and you can add holdouts that see no paywall, per Superwall's docs.
Paywalls render remotely, so new designs ship without an App Store update.
Teams that ran tests on Superwall have published clear results (all vendor-reported case studies):
- Mojo saw +14% new subscription revenue after hiding credit packs and selling subscriptions only.
- Stardust added a 7-day free trial on the yearly plan and saw +131% paywall conversion in a multivariate test. A yearly-only paywall lost on both conversion and proceeds per user.
- BuyBye raised its yearly price by 50%: conversions fell 19%, but proceeds per user rose 25%.
- PropGPT moved its weekly price from $6.99 to $9.99 and saw +47.5% proceeds per user.
Downsides: it is a paywall tool, not a general experiment platform, so onboarding or feature tests outside the paywall need another tool. Fees grow with paywall revenue.
Not for: apps that want to test product features rather than monetization. More in our Superwall review.
RevenueCat Experiments
RevenueCat Experiments tests product pricing, trial length and presence, introductory offers, the number and mix of products, and paywall imagery, copy and layout, per RevenueCat's docs.
Experiments is available on the Pro and Enterprise plans.
Its main strength is that results tie straight to RevenueCat's subscription data, including renewals and refunds.
Published RevenueCat results (vendor-reported):
- Pixelcut offered a weekly plan instead of monthly and got +16% paying customers.
- Photoroom tested localized upsell paywalls and saw +50% upsell screen conversion.
- Zumba sent the paywall button to a web checkout: immediate conversions fell 25%, but LTV rose 17%.
- In RevenueCat's own test for Dipsea, a web-checkout-only paywall dropped initial conversion from 27.0% to 18.1%.
The Zumba and Dipsea results show why revenue-aware reporting matters: a test that loses on conversion can still win on LTV, or lose outright.
Downsides: you need RevenueCat as your subscription backend, and the paywall editor is less design-focused than Superwall's.
Fees grow with tracked revenue, not just tested revenue.
Not for: apps that don't want to move purchase handling to a third party.
See our RevenueCat review and Superwall vs RevenueCat.
Adapty
Adapty runs A/B tests on subscription prices, paywall design, copy and layout, trial periods and onboarding designs, with no code changes for each test, per Adapty's docs.
It offers regular tests on one placement and cross-placement tests that show the same variant to a user everywhere.
Downsides: like RevenueCat, it works best when Adapty is your subscription backend, which makes switching later a migration project.
Its docs page does not state variant limits or the statistical method.
Not for: teams already settled on another subscription backend.
More in our Adapty review and Adapty vs RevenueCat.
RevenueCatStore listing tests: Apple PPO, Google Play experiments, SplitMetrics, Storemaven
Store listing tests change what people see before they install: icon, screenshots and preview video.
Start with the free native tools, then add a paid tool if you need pre-launch tests or deeper data.
Our Product Page Optimization guide covers the Apple side in detail.
Apple Product Page Optimization
PPO tests up to 3 treatments against your original product page for up to 90 days, per Apple.
You can test icons, screenshots and app preview videos, localized or not.
New assets need App Review approval, and alternate icons must be included in the app binary.
Apple recommends at least 90% confidence before applying a treatment.
Downsides: it can't test the app name, subtitle or description, and low-traffic apps may wait a long time for a result.
Google Play store listing experiments
Google Play lets you test up to 2 variants against your current listing, per Play Console Help.
Default graphics experiments cover the icon, feature graphic and screenshots.
Localized experiments also cover descriptions in up to five languages.
You can run one default graphics experiment or up to five localized experiments at a time, and experiments stop after 6 months.
SplitMetrics Optimize
SplitMetrics Optimize sends paid traffic to pages that copy App Store and Google Play listings, so you can test icons, screenshots and metadata before they go live, per SplitMetrics.
Pricing is quote-based. The tradeoff: you pay for the traffic, and visitors on a copied page may not act exactly like real store visitors. See our SplitMetrics review.
Storemaven
Storemaven was an app store testing platform known for predictive creative testing.
Zynga acquired it in September 2022 and planned to combine it with its own games and ad platform, as 80.lv reported.
When we checked in November 2026, storemaven.com no longer showed the product.
Confirm it is available before you plan around it.
For other store tools, see our ASO tools roundup.
Pricing compared
Prices below are as of November 2026, taken from each vendor's pricing page.
Revenue-based tools charge on revenue they track, so check how each one defines it.
| Tool | Type | Free tier | Paid pricing |
|---|---|---|---|
| Firebase A/B Testing | In-app | No-cost on Spark and Blaze | Free (pricing) |
| Statsig | In-app | 2M events/month | Pro from $150/month; Enterprise custom (pricing) |
| Optimizely Feature Experimentation | In-app | None listed | Quote only (pricing) |
| Superwall | Paywall | Up to $10,000/month paywall revenue | 1% of paywall-attributed revenue above that (pricing) |
| RevenueCat | Paywall + backend | Up to $2,500 monthly tracked revenue | 1% of tracked revenue above that; Enterprise custom (pricing) |
| Adapty | Paywall + backend | Under $5,000/month | 1% of tracked revenue above that; Enterprise custom (pricing) |
| Apple PPO | Store listing | Free | Free |
| Google Play experiments | Store listing | Free | Free |
| SplitMetrics Optimize | Store listing | None listed | Quote only, plus traffic costs (pricing) |
One detail: Superwall meters only revenue that converts through a Superwall-rendered paywall, while RevenueCat and Adapty meter all revenue they track.
At the same total revenue, the bills can differ a lot depending on how much flows through the paywall.
Which tool to pick
- You mostly want to test paywalls, prices and trials: pick Superwall, RevenueCat or Adapty. Superwall if paywall design is the focus; RevenueCat or Adapty if you also want them to run purchases and subscription analytics.
- You test onboarding and features too: add Firebase A/B Testing if budget is tight, or Statsig if you want stronger analytics and can stay inside its event limits.
- You run experiments across web and app at enterprise scale: Optimizely Feature Experimentation, if the sales-led price fits.
- You want more installs from the same traffic: start with Apple PPO and Google Play experiments. Add SplitMetrics for pre-launch tests or when native tests take too long.
Whatever you pick, judge tests on revenue per user, not conversion alone.
BuyBye's price test lost 19% of conversions and still earned 25% more per user, and Zumba's web checkout lost on immediate conversion but won on LTV.
Browse more results in our paywall test library before you plan your first test.