Why Your iOS And Android Numbers Do Not Match

The short version

Comparing iOS and Android performance side by side compares two measurement regimes, not two audiences. On iOS a large share of users are measured through a privacy-preserving path that is delayed, coarse, and incomplete. On Android, measurement is more complete.

So when iOS looks worse, the first question is not "why is iOS performing badly" but "how much of this gap is measurement?"

Where the gap comes from

Users who declined tracking. Every declined permission prompt moves that user into the coarse path. Their conversions may be attributed late, coarsely, or not at all.

Delayed reporting. Privacy-preserving postbacks arrive after a timer rather than in real time. Recent iOS periods look worse simply because the data has not landed yet, and they improve retroactively.

Privacy thresholds. When a breakdown contains too few installs, detail is withheld entirely. Small iOS campaigns lose exactly the granularity you want to analyze.

No user-level join. Long-window analysis at user level is unavailable on the coarse path, so iOS LTV looks thinner because it is measured differently, not because users are worth less.

Crucially, the missing data is not randomly distributed. It correlates with who declines, so what you can measure on iOS is not a representative sample of what you bought.

What to do instead of comparing

Compare within platform over time. iOS this month against iOS last month is a valid comparison. iOS against Android at a point in time is not.

Wait before judging iOS. Reported iOS performance improves as delayed postbacks arrive. Reading a recent window as final produces a pessimism that corrects itself.

Use your own revenue data as the arbiter. Your backend does not care which platform attributed a sale. Blended revenue against blended spend sidesteps the attribution asymmetry entirely.

Track the size of the gap rather than closing it. The gap between platform-reported, MMP-reported, and your own numbers is itself information about how much reported performance is claimed rather than observed. Teams that try to reconcile the numbers into agreement lose that signal.

Patterns worth recognizing

iOS conversions far below Android at similar spend. Usually measurement. Check the privacy-preserving path before concluding the channel is weak.

iOS performance appearing to drop after an app release. Check whether permission prompt timing or the priming screen changed before investigating campaigns.

Platform and MMP iOS numbers diverging further over time. Expected as the measured share shifts. Not a fault.

iOS improving retroactively. Normal. It is delayed data arriving, not a change in performance.

The decision this actually affects

Budget allocation between platforms. If you allocate on reported performance, you will systematically underfund iOS, because iOS reporting is systematically incomplete.

Allocate on blended business results where you can, and treat platform-reported iOS numbers as a floor rather than an estimate.

Practical rule

Never compare iOS to Android at a point in time. Compare each platform to itself over time, and use your own revenue data when you need to decide between them.