A/B testing (App Store)

You can test up to three treatments with alternate app icons, screenshots and app previews against your original. A test runs for 90 days or until you stop it, and a stopped test cannot be restarted. It applies to iOS and iPadOS apps on version 15 or later and is not available for custom product pages.

Text fields such as the app name, subtitle and keywords are not part of a treatment. To judge those, use a before-and-after method with rank tracking and a metadata update. That is weaker evidence, because time and season also change. Changes may take up to 24 hours to appear.

Results appear in App Analytics after five first-time downloads are associated with the test. That count is only when numbers show. A difference between 12 and 15 downloads is noise.

1. Change one element per test.
2. Define the metric first, usually conversion rate.
3. Let the test run long enough to cover weekdays and weekends.
4. Do not change other metadata mid-test.
5. Keep the winner only if the gap is clearly larger than the noise.

A hypothetical habit app tests a screenshot set that leads with widgets against its current set that leads with streaks. After several weeks the widget set converts better, but another custom product page shows the streak set does better for ads that mention streaks. Both facts are useful.

Read App Analytics for where results appear.