iOS 14.5 went live on Monday and App Tracking Transparency is now in effect. The opt-in prompt — “Allow [App Name] to track your activity across other companies’ apps and websites?” — is appearing for users across the iOS ecosystem, and the early data on consent rates is landing toward the bottom of the range that the industry had been modeling.

Initial reports from measurement partners and app publishers suggest opt-in rates are running between 20 and 30 percent across general consumer app categories, with some utility and productivity apps seeing rates in the 30-40 percent range and entertainment apps seeing rates at or below 20 percent. Gaming apps, many of which have tried to implement pre-permission prompts to contextualize the ATT ask, are seeing variable results.

The most immediate market signal has come from Facebook Audience Network — the off-Facebook programmatic advertising product that monetizes third-party app inventory using Facebook’s audience targeting. Multiple publishers and app developers reported this week that Audience Network CPMs dropped significantly within days of iOS 14.5 going live, in some cases by 50 percent or more on iOS inventory. Facebook previewed this impact months ago, publicly warning publishers that Audience Network’s iOS effectiveness would be “severely limited” under ATT.

The industry’s planning models for ATT were mostly built around two scenarios: a high-consent scenario (40-50% opt-in) and a low-consent scenario (under 30% opt-in). The early data is landing in or below the low-consent scenario for most app categories.

Why are consent rates so low? The ATT prompt is designed by Apple for user clarity, not for advertiser conversion. The language — “track your activity across other companies’ apps and websites” — is accurate and somewhat alarming to users who read it carefully. The placement of “Ask App Not to Track” as the first option (before “Allow”) follows a choice architecture that is favorable to opt-out.

App developers who implemented thoughtful pre-permission prompts — appearing before the system ATT dialog to explain why tracking benefits the user (free apps, personalized content) — are seeing meaningfully better opt-in rates than those who show the raw ATT dialog without context. The gap suggests that the floor for opt-in rates is not fixed; user communication about the value exchange can move consent rates by 10-15 percentage points in favorable conditions.

But even with well-designed pre-permission prompts, most apps are seeing the majority of users decline tracking. The structural reality is that when users are explicitly asked about cross-app tracking, most of them say no. This is not a UX problem to be optimized away. It is a signal about user preferences that the advertising industry needs to incorporate into its planning.

Facebook Audience Network: The First Casualty

Facebook’s Audience Network has been the most immediate and visible casualty of ATT. The product’s value proposition — extending Facebook’s audience targeting to third-party apps outside the Facebook environment — is built on IDFA-based cross-app identity. When that identity is unavailable for the majority of iOS users, the targeting accuracy that justified Audience Network’s premium CPMs disappears.

Facebook’s developer documentation had already indicated that Audience Network CPMs would drop significantly for iOS 14.5 users without ATT consent. What publishers are reporting this week confirms the magnitude of that impact. Apps with primarily iOS user bases that were monetizing heavily through Audience Network are experiencing revenue drops that in some cases exceed 50 percent of their pre-ATT iOS programmatic yield.

The Android portion of Audience Network inventory is unaffected — Google has not implemented a comparable ATT-style opt-in for Android’s GAID identifier, though industry analysts expect Google to move in that direction. Publishers with Android-heavy user bases are insulated from the immediate Audience Network impact, and there is evidence of some budget rotation from iOS to Android as buyers seek addressable mobile inventory.

App Category Impacts

The impact distribution across app categories is non-uniform, and the patterns are worth understanding for campaign planning.

App install campaigns: This is the most severely affected use case. App install campaigns on iOS through most programmatic channels are now operating with significantly degraded attribution signal — SKAdNetwork data is available but lacks the granularity and timeliness of IDFA-based attribution. Campaign optimization cycles that previously ran on near-real-time install data now run on SKAdNetwork conversion notifications with a multi-day delay. iOS app install CPIs are rising as buyers compete for the reduced pool of IDFA-opted-in inventory.

Retargeting / re-engagement: Effectively non-functional for opted-out users at scale. Re-engagement campaigns that targeted lapsed users by IDFA matching are no longer possible for users who have declined ATT. Some DSPs are exploring fingerprinting-adjacent methods — but Apple has indicated that using device signals as a proxy for tracking without consent is a policy violation that could result in App Store rejection.

Brand video and rewarded formats: Less affected, because these formats are less dependent on individual-level targeting for their value proposition. Contextual placement — ads in apps with audience characteristics that align to campaign goals — is a viable alternative for awareness-oriented mobile video.

E-commerce / retail: Mobile commerce apps with logged-in users can use first-party identity for retargeting within their own apps. Cross-app retargeting to drive users back to a retail app from other apps is affected by ATT. The impact on overall e-commerce performance is being partially absorbed by the growth of web-based shopping, where IDFA is not relevant.

The Android Opportunity and Its Risks

The ATT impact has already produced a partial rotation of mobile programmatic spend from iOS to Android. Android’s GAID (Google Advertising ID) is still available by default on most devices, making Android inventory more addressable for IDFA-dependent campaign types in the near term.

Buyers accelerating Android spend should do so with awareness that Android’s current position is temporary. Google has publicly stated that it is watching the ATT rollout and evaluating changes to its own mobile identifier approach. The timeline for any Google changes to GAID is not announced, but the direction of the industry — toward opt-in models for cross-app tracking — is consistent across platform operators.

FAQ

What is SKAdNetwork and how does it work for attribution post-ATT? SKAdNetwork is Apple’s privacy-preserving attribution framework for iOS app install campaigns. When a user sees an ad and installs an advertiser’s app, Apple’s servers send a signed attribution notification to the ad network — without revealing the user’s identity. The notification includes a campaign ID, source app, and a conversion value (a number between 0 and 63 that advertisers can configure to represent post-install events). Notifications are delayed by a minimum of 24 hours and can be delayed by up to several days. The aggregated, delayed, low-granularity data SKAdNetwork provides is significantly less useful for optimization than IDFA-based attribution but is the only deterministic attribution available for opted-out iOS users.

What is the current opt-in rate for ATT and where can I track it? Real-time ATT opt-in rate data is not publicly published in a centralized way. AppsFlyer, Adjust, and other mobile measurement partners are publishing aggregate opt-in rate data based on their client network data. Flurry Analytics, owned by Verizon Media, is publishing daily opt-in rate updates as a public resource. The early figures as of iOS 14.5 launch week suggest 20-30% opt-in across most consumer categories, with variation by app type and prompt design.

Should we be building for modeling and probabilistic attribution now? Yes. The days of deterministic last-touch attribution for iOS campaigns are over for the majority of users. Media mix modeling (MMM), incrementality testing, and probabilistic attribution models are now necessary complements to whatever deterministic signal survives. Companies that have invested in MMM infrastructure already are better positioned to navigate signal loss than those who have relied entirely on click-based attribution. This is also true for web-based campaigns as cookie deprecation approaches — the attribution methodology transition needed for mobile is a preview of what web programmatic will require.

Is Facebook Audience Network still worth running for iOS? For advertisers running awareness and brand campaigns that do not depend on individual-level targeting precision, Audience Network’s iOS inventory may still deliver reach at lower CPMs than alternatives, particularly during the current period of demand pullback. For performance advertisers who need measurable conversion attribution, the short-term recommendation is to reallocate iOS budgets away from Audience Network toward channels with better ATT-era attribution infrastructure, or toward Android. Facebook’s Aggregated Event Measurement is designed to partially restore some attribution signal for opted-in users and through modeled conversion estimates — evaluate your specific campaign type’s eligibility for AEM before making reallocation decisions.