A/B Test Calculator
Most PPO tests are called after three days on a difference that was never real. Work out how much traffic yours actually needs before you start — and check the finished numbers before you ship the winner.
Impressions per arm
41,914
each treatment and the original
Total impressions
83,828
2 arms
Days to get there
42
Apple caps a test at 90
Apple allows 3 treatments against the original, runs a test for up to 90 days, and recommends waiting for at least 90% confidence — which is why 90% is the default here rather than the textbook 95%. App Store Connect reports the confidence for your test; this works out what the test needed in the first place.
Decide the size before you look. Checking a running test repeatedly and stopping the moment it crosses the line inflates the false positive rate well past the number on screen, however the maths is corrected.
This models impressions to downloads. It cannot see whether a treatment brings worse users — a screenshot set that wins on install rate and loses on retention is a real outcome, and only your own analytics will show it.
Testing the creative itself? Screenshot teardown and icon teardown show what you are up against before you pick what to test.