Glossary
What is App Tracking Transparency (ATT)?
App Tracking Transparency is Apple's iOS framework requiring apps to request explicit permission before tracking a user's activity across other companies' apps and websites.
What ATT actually requires
The prompt appears once per app and asks a single question: may this app track you across other companies' apps and websites? Decline, and the app cannot access the IDFA or use other identifiers for cross-party tracking.
The scope is narrower than it is often described. ATT governs tracking *across* companies. It does not restrict an app measuring its own users inside its own product, and it does not restrict contextual advertising — which is why first-party analytics survived the change essentially untouched.
What it changed
- Deterministic mobile attribution largely ended, replaced by SKAdNetwork aggregate reporting.
- Audience targeting on third-party data collapsed on iOS, pushing budget toward contextual and first-party signals.
- Retargeting became far harder, since the pool of addressable lapsed users shrank sharply.
- Measurement became slower and coarser, with delayed, threshold-gated postbacks replacing real-time user-level events.
What it did not change
Three things survived that are worth naming, because plans built on the assumption that iOS became unmeasurable are wrong.
First-party measurement. Behaviour inside your own product is yours to observe, and lifecycle analytics work as they always did.
Contextual advertising. Matching an ad to what a user is currently doing needs no identifier at all — which is why conversational surfaces are structurally unaffected by identifier policy.
Causal measurement. Incrementality and media mix modeling never depended on identifying anyone, and both became more valuable rather than less.
The direction it set
ATT was the first large-scale demonstration that a platform can withdraw a measurement primitive unilaterally, and that the industry adapts rather than collapses. Everything since — Privacy Sandbox on Android, browser restrictions, aggregated reporting APIs — follows the same pattern: less user-level data, more aggregation, thresholds before disclosure.
The teams that handled it best treated it as permanent immediately rather than waiting for a workaround, and reinvested in measurement that does not require identity.
Common questions
What is a typical ATT opt-in rate?
It varies widely by category and by how well the pre-prompt explains the value exchange. Most consumer apps see a minority opt in; utility and finance apps that explain the benefit clearly do noticeably better.
Does ATT apply to advertising inside AI assistants?
Not in the way it applies to app install attribution. Conversational targeting works on stated intent rather than on cross-app identifiers, so it does not depend on the permission ATT governs.
More in privacy and compliance
The identifiers, consent signals and rules that bound targeting.