ATT (App Tracking Transparency) is the iOS framework (introduced in iOS 14.5, April 2021) that requires apps to display a system prompt and obtain explicit user consent before accessing the IDFA or tracking the user across other apps and websites. Apple controls the prompt — apps cannot modify the wording, cannot ask twice in the same install, cannot use dark-pattern preambles to coerce consent.
The reframe: pre-ATT, iOS advertising operated on an opt-OUT model — the IDFA was available to every installed app unless the user dug into Settings → Privacy and disabled it (almost no one did, opt-out rate <5%). Post-ATT, it's an opt-IN model — most users decline by default. The flip was one of the largest privacy shifts in digital advertising history, and the most consequential mobile UA event of the past decade.
Opt-in rates by app category (2026 anchors):
- Consumer apps with clear personalization value: 40-50%.
- Productivity / utility: 25-40%.
- Casual games: 20-30%.
- Hyper-casual games: 15-25%.
- Strongly-branded consumer apps: 50%+ — trust-driven.
Lifting opt-in: apps that show a "pre-prompt" screen (explaining what tracking enables in concrete user-benefit terms) lift opt-in 10-20 percentage points vs cold prompts. The pre-prompt is not Apple-restricted (you control the wording) — but the system prompt itself, when triggered, must follow Apple's rules exactly.
Economic impact: iOS CPIs roughly doubled across many verticals in the 6-12 months after ATT rolled out. ROAS dropped 30-50% on iOS-heavy paid UA programs. Subscription apps and games were hit hardest; brand-driven and broad-reach apps less so. The industry adapted by (a) shifting more budget to Android (where GAID-based deterministic attribution still works), (b) building SKAN-aware conversion-value schemas, (c) investing heavily in first-party data (CRM, web-to-app, email).
Probabilistic / fingerprinting attribution: was widely deployed as an ATT-era workaround in 2021-2023, attempting to identify users via device + IP + browser fingerprint rather than the IDFA. Apple progressively restricted these methods (App Store Review Guideline 5.1.2 explicitly prohibits fingerprinting for advertising purposes), and most reputable MMPs have de-emphasized them. Surviving alternatives are SKAN (Apple-blessed) and first-party signal (CRM, owned-channel).
ATT opt-in rates by app category (2026)
| Category | Cold prompt | With pre-prompt |
|---|---|---|
| Consumer apps with personalization promise (streaming, dating, finance) | 30-40% | 45-55% |
| Productivity / utility | 20-30% | 30-45% |
| Casual games | 15-25% | 25-35% |
| Hyper-casual games | 10-20% | 15-25% |
| Strongly-branded consumer apps (brand-trust users) | 40-50% | 55-65% |
Pre-prompt — a custom screen explaining what tracking enables in concrete user-benefit terms before the iOS system prompt fires — is the single highest-impact ATT optimization. Apple does not restrict the pre-prompt; only the system prompt that follows.