Apple announced App Tracking Transparency at WWDC in June, and the mobile advertising industry spent approximately 45 minutes convincing itself it would not be that bad. The past two weeks of additional clarity have made it clear that it will be exactly that bad — or possibly worse, depending on how iOS users respond to a consent prompt that is explicitly designed to make them think carefully before saying yes.

The mechanism is straightforward. Starting with iOS 14, apps that want to access the IDFA — the Identifier for Advertisers, the device-level identifier that powers most mobile programmatic advertising — will be required to show users the App Tracking Transparency prompt. The prompt reads, in Apple’s standard language: “[App name] would like permission to track you across apps and websites owned by other companies.” Users will be able to tap “Ask App Not to Track” or “Allow.”

Apple is not a neutral party here. Apple has competitive interests in mobile advertising — it operates its own advertising platform (the App Store search ads and Apple News ads). Apple has designed the ATT prompt to maximize user awareness of tracking, not to maximize advertiser access to tracking consent. The expected opt-in rates of 40 to 60 percent that the industry was discussing in June now look optimistic. Some analysts are modeling 20 to 30 percent opt-in rates in consumer applications.

What the IDFA Actually Powers

To understand the scale of the disruption, it helps to understand specifically what the IDFA enables in mobile programmatic campaigns.

Audience targeting: The IDFA is the persistent identifier that allows DSPs to match a user’s device to an audience segment. Attribution vendors, data providers, and identity graphs use the IDFA to build mobile audience profiles. Without IDFA, these targeting capabilities are materially limited.

Frequency capping: Limiting how many times a specific user sees a specific ad across multiple apps requires a persistent identifier. Without IDFA, frequency capping at the user level on iOS is no longer technically possible across different apps.

Retargeting and re-engagement: App advertisers running re-engagement campaigns — serving ads to users who have already installed the app but become inactive — depend on IDFA to identify those users in other app environments. Without IDFA, re-engagement at the user level is not possible for users who have not opted in.

Conversion attribution: The dominant mobile attribution model — last-touch attribution linking an impression or click to an install or in-app event — depends on IDFA to match the attributing signal to the converting user. Mobile measurement partners (Adjust, AppsFlyer, Branch, Kochava) built their entire attribution infrastructure on IDFA.

Each of these use cases is affected differently by the ATT prompt. For users who opt in, IDFA-based workflows continue as before. For users who opt out — the likely majority — alternatives are required.

Modeling the Opt-In Scenarios

The functional shape of post-ATT mobile programmatic depends heavily on where actual opt-in rates land. The range of scenarios being discussed in the industry is wide:

High opt-in (50-60%): If roughly half of iOS users allow tracking, the total addressable universe for IDFA-based campaigns shrinks by roughly half but remains commercially viable. Premium app verticals where users understand the value exchange — premium gaming, utility apps, subscription apps — may maintain higher opt-in rates. The targeting efficiency loss is significant but manageable.

Moderate opt-in (30-40%): At this level, most audience-based targeting on iOS shifts from deterministic to probabilistic. Attribution becomes modeled rather than measured for a significant portion of conversions. Campaign optimization signals are noisier. CPMs for IDFA-opted-in traffic surge as demand concentrates on a smaller available pool.

Low opt-in (under 25%): At this level, IDFA-based campaign models break. Retargeting at scale becomes commercially unviable. App install campaigns relying on deterministic attribution lose most of their signal. The advertising value of iOS inventory shifts dramatically toward contextual, cohort-based, and publisher first-party data approaches.

Most current industry analysis is converging on the moderate scenario as the likely near-term equilibrium, with considerable variation by app category. Gaming apps, where users expect to see ads as part of a free-to-play model, may achieve higher opt-in rates with well-designed prompt context. Finance and health apps, where users are most sensitive to tracking, may see rates below 20 percent.

What This Means for Mobile Video, Retargeting, and App Install Campaigns

Mobile video: Programmatic mobile video is primarily inventory-based rather than audience-based — the video unit’s value comes from its placement in a high-engagement environment, not from sophisticated audience targeting. CPM impacts will be less severe for in-app video than for performance-oriented formats. CTV-style quality video environments with strong contextual signals will hold value better than open exchange video.

Retargeting: This is the most severely affected use case. User-level re-engagement on iOS becomes essentially impossible for users who opt out of tracking. Some advertisers are exploring SKAdNetwork-based re-engagement concepts, but SKAdNetwork was not designed for retargeting. The honest answer is that iOS retargeting as a campaign type loses most of its functionality under ATT at any opt-in rate below roughly 50 percent.

App install campaigns: The performance signal for app install campaigns shifts from deterministic (this specific user installed the app after seeing this specific ad) to modeled (aggregated conversion data suggests this campaign drove installs at approximately this rate). SKAdNetwork, Apple’s privacy-preserving attribution framework, provides aggregate campaign-level data with strict signal limits. MMP integrations with SKAdNetwork are in development across Adjust, AppsFlyer, and others, but the attribution fidelity is materially lower.

Which DSPs Are Investing in Alternative Signals

The DSPs positioned better for the post-ATT environment are those investing in three areas: contextual signals for in-app environments, cohort-based targeting that does not depend on individual IDFA matching, and partnerships with publishers who have first-party logged-in user data.

The Trade Desk has been the most vocal among DSPs about its identity investment — UID2 (Unified ID 2.0) is designed to work on authenticated web inventory but has limited applicability to in-app mobile environments where most traffic is not logged in. Its contextual and first-party data capabilities are stronger.

Google’s UAC (Universal App Campaigns) for app install buying is already largely opaque and machine-learning driven rather than IDFA-dependent in the traditional sense — Google has been preparing for signal loss in its own campaigns for longer than the third-party DSP ecosystem. This gives Google’s app install product structural resilience that independent DSPs lack.

Facebook’s Audience Network — the primary revenue mechanism through which Facebook monetizes off-Facebook mobile inventory — depends heavily on IDFA for the cross-app targeting that makes it valuable. Audience Network is likely the biggest short-term casualty of ATT among the major ad tech platforms, which is part of why Facebook has been the most publicly critical of Apple’s implementation.

The practical advice for mobile programmatic buyers right now is to audit which campaigns are most IDFA-dependent and begin testing contextual and cohort-based targeting approaches while IDFA traffic is still available. The iOS 14 rollout will happen in the fall. The campaigns that have contextual alternatives tested and validated before then will have data to optimize against. The campaigns that have not will be optimizing blind.

FAQ

When exactly does iOS 14 and ATT go live? Apple announced iOS 14 at WWDC in June and clarified the ATT timing in August. The current guidance is that ATT will be required for all apps accessing IDFA starting with iOS 14, which will launch in the fall. Apple announced in late June that it would give the advertising industry additional time — a few months of warning after iOS 14 ships before enforcement becomes strict. Watch Apple’s developer documentation for specific dates as iOS 14 approaches general availability.

What is SKAdNetwork and does it actually replace IDFA for attribution? SKAdNetwork is Apple’s privacy-preserving attribution framework, available since 2018 but largely unused until ATT made it necessary. It allows apps to receive attribution data for installs without accessing IDFA — Apple’s servers sign the attribution and send an anonymized notification to the ad network after a conversion delay. What SKAdNetwork provides is aggregate install count data with a time delay and without user-level detail. It does not replace IDFA for user-level attribution, re-engagement, or audience targeting. It is a significantly degraded measurement signal compared to current IDFA-based attribution.

Should we shift budget from iOS to Android in response to ATT? Some shift is likely to make sense, particularly for app install campaigns where Android’s continued support for GAID (Google Advertising ID) maintains deterministic attribution capabilities. However, Android is not immune from similar changes — Google has indicated it is watching ATT’s rollout and Apple’s model. Concentrating entirely on Android creates its own risk if Google moves to a similar opt-in model, which some analysts consider likely within 18 to 24 months. The right response is diversification and alternative signal development, not complete platform reallocation.

How should we brief our clients about ATT? Start with what is certain: IDFA-dependent campaign types on iOS will have reduced targeting and attribution fidelity for users who opt out. Be specific about which campaign types — retargeting, deterministic app install attribution — are most affected and by how much at different opt-in rate scenarios. Avoid speculative reassurances that optimization will compensate for signal loss without data to support that claim. The clients who are prepared for a period of reduced measurement certainty will adapt better than those who receive false reassurance followed by performance shortfalls.