How Apple's App Tracking Transparency Changed Mobile Ad Revenue

In April 2021, Apple shipped iOS 14.5 with a single new permission prompt, and mobile advertising has not looked the same since. If you’re trying to understand why iOS ad revenue behaves the way it does today — why some published eCPM figures still look dated, why targeting feels blunter than it used to, why Android and iOS diverged in new ways — the story runs through that one prompt. This is what actually happened, and what it means for an app launching today rather than in 2020.

Before: the IDFA era

For most of the 2010s, every iOS device carried an IDFA (Identifier for Advertisers) — a persistent, resettable ID any app could read freely. Ad networks used it to stitch together a single user’s behavior across many apps: what you installed, what you clicked, what converted. That cross-app picture is what let networks target precisely and attribute a specific install to a specific ad, which in turn let them bid confidently and pay advertisers-grade prices for iOS inventory. Precise targeting is what iOS’s historical eCPM premium was substantially built on, not just higher iOS spending power on its own.

The change: App Tracking Transparency

App Tracking Transparency (ATT) requires every app to show a system prompt asking permission before it can access the IDFA or otherwise track a user across other companies’ apps and sites. Users could always say no — and overwhelmingly, they did. Industry-wide opt-in rates settled in a rough 20–30% range, meaning the majority of iOS users became untrackable by the old method almost overnight, across every app that updated to comply.

Apple also shipped a replacement measurement system, SKAdNetwork, alongside ATT. It can confirm that a campaign led to an install and report highly aggregated, delayed conversion data, but by design it cannot tie a specific ad to a specific user the way IDFA-based attribution could. That trade-off — privacy-preserving but structurally coarser — is the crux of everything that changed downstream.

What happened to revenue: an immediate dip, not a permanent collapse

In the months after ATT rolled out, iOS eCPMs across most ad networks fell noticeably — advertisers, suddenly unable to target and measure as precisely, cut bids while they recalibrated. Some publishers saw double-digit percentage drops in iOS revenue through mid-to-late 2021. That dip is the event a lot of older “iOS pays more” benchmark data was written around, and it’s why figures from that window undersell where things landed.

The market did not stay there. Ad networks and mediation platforms rebuilt targeting around aggregated, probabilistic signals instead of per-user tracking — contextual signals, on-device modeling, and SKAdNetwork’s own postback data, imperfect as it is. Advertiser demand for iOS users recovered as measurement matured, and the iOS eCPM premium over Android returned, though the mechanism behind it shifted from precise cross-app targeting to iOS users’ still-higher underlying spending behavior. That recovered premium is what the calculator’s 1.3× iOS multiplier reflects today.

What it means for an app launching now

Three practical consequences carry forward, years after the initial shock:

• Expect a same-format iOS premium, but earned differently. iOS still out-earns Android per impression, but campaigns lean more on aggregated modeling than on individual tracking — you cannot assume the hyper-precise retargeting playbooks written before 2021 still work as described.
• Rewarded and interstitial matter more, not less. Formats where the ad itself carries the value — a rewarded video a user opts into — are less dependent on precise targeting than a banner hoping to catch the right person. See the rewarded vs interstitial guide for why that format already earns more regardless.
• Mediation and in-app bidding matter more. With per-network targeting less precise, an auction across many networks (see AppLovin MAX vs AdMob) captures more of the available value than betting on a single network’s targeting to work well.
• Don’t benchmark against pre-2021 numbers. Any eCPM figure, case study, or rule of thumb dated before mid-2021 reflects a measurement environment that no longer exists. Treat it as historical context, not a current baseline.

Android didn’t escape unchanged either

Google has moved in a similar direction with its own Privacy Sandbox initiative on Android, gradually reducing reliance on cross-app identifiers in favor of on-device, aggregated measurement. The timeline and mechanics differ from ATT, but the broader industry direction — less precise individual tracking, more aggregated modeling — is the same on both platforms. iOS just got there first, abruptly, which is why it makes the clearer case study.

The takeaway

ATT didn’t end iOS as a strong monetization platform — it changed how that value gets captured, from precise per-user targeting to aggregated modeling and format choice doing more of the work. An app built today should expect the current iOS premium the calculator models, not the pre-2021 or the 2021-dip numbers still floating around older articles.

See the current iOS vs Android estimate for your app →