Apple's App Tracking Transparency (ATT) framework reduced opt-in rates for cross-app tracking to 25-40% on average, with iOS now dominating SKAdNetwork-based aggregated attribution. The change broke the cost-per-install economics most app UA programmes were built around. CPIs spiked. Lookalike audiences degraded. Campaigns that worked for years stopped working overnight.
Three patterns show up in every app UA audit. One: still buying against install-based ROAS targets when post-install conversion measurement is fundamentally noisier than it was pre-iOS 14 — chasing yesterday's metrics with broken data. Two: SKAdNetwork conversion value setup that's either default (sending almost no signal back) or so complex that the team doesn't understand what they're optimising toward. Three: MMP (Mobile Measurement Partner) setup with leaking data — duplicate attribution, broken integrations with ad networks, postbacks not configured for the platforms that matter.
Doing this well in 2026 requires actually understanding how SKAdNetwork, App Tracking Transparency, and Privacy Manifests interact with paid acquisition. It means designing conversion value schemas that send meaningful post-install signal back to networks. It means using AppsFlyer, Adjust, or Branch for unified attribution while accepting the limitations of aggregated measurement.
It's a small specialist field. Most agencies that say they 'do app UA' are running mobile web Meta campaigns for app companies and calling it acquisition.