A/B testing App Store and Google Play screenshots

Screenshots are the biggest lever on store conversion that you fully control. Both major stores let you test alternatives against your live listing with real traffic — but the tools differ, and a test is only as good as the variants you feed it.

Apple: Product Page Optimization (PPO)

• Test up to three treatments against your original product page. You can vary screenshots, app previews and app icons.
• Only one test runs at a time, for up to 90 days. You choose what share of traffic sees treatments; it is split evenly between them, and each visitor keeps seeing the same version.
• New screenshots in a treatment go through App Review before the test starts (icon tests also require the icons in your app binary).
• Treatments can be localized for all or some of your languages — more localizations can mean longer tests to reach significance.
• Apple reports conversion rate and improvement with a confidence level; wait for high confidence (Apple suggests at least 90%) before applying a winner.

Google Play: store listing experiments

• Test up to two variants against your current listing, with an adjustable share of visitors split equally across variants.
• Default graphics experiments test the icon, feature graphic and screenshots in your default language — one at a time.
• Localized experiments can also test descriptions, and you can run up to five at once in different languages.
• Pick a target metric such as install clicks; Play Console estimates how long the test needs, and experiments stop automatically after six months.

What to test (in rough order of impact)

1. The first screenshot — it does most of the work in search results. Test a benefit headline vs. a feature headline, or a hero shot vs. a UI capture.
2. Headline angle across the set: outcome ("Sleep better tonight") vs. mechanism ("Smart sleep tracking").
3. Order — lead with the feature your best reviews mention.
4. Visual style — light vs. dark backgrounds, frames vs. frameless, 2D vs. 3D devices.
5. Social proof — award badges or ratings as overlay elements.

Preparing variants in Storedeck

1. Mark your live set as control: open the variant menu and set its status.
2. Click + next to the variant tabs. The new variant is a full copy of the current one — every platform, screen, language and image.
3. Write the hypothesis ("Outcome-led headlines lift installs") and change only what the test is about.
4. Use the Compare view to see variants side by side for the same platform and language, exactly as a shopper would scan them.
5. Export: pick the variants, platforms and languages in the export dialog. Files are grouped as variant/platform/language/NN-name.png, ready for PPO treatments or Play experiment variants.
6. When the store declares a result, set that variant to winner — the previous control is demoted automatically, so the project always shows what is live.

How long should a screenshot test run?

Long enough to cover at least one full week (weekday and weekend traffic behave differently) and to reach the confidence the store reports as reliable. Low-traffic apps should test bold differences — subtle tweaks need huge samples. Avoid running tests during feature placements or big marketing pushes, which skew who visits your page.

Frequently asked questions