AdAttributionKit is Apple's replacement for SKAdNetwork, shipping with iOS 17.4+ and becoming the default attribution framework in 2026. If you're still running SKAN 4.0 campaigns, you need a migration plan now. Here's exactly what changed, what improved, and what it means for your ad spend.
Page Contents
What is AdAttributionKit and why did Apple replace SKAdNetwork?
AdAttributionKit is Apple's privacy-preserving attribution framework that replaces SKAdNetwork starting in iOS 17.4. It adds re-engagement measurement, developer postbacks, and improved crowd anonymity thresholds. According to Apple's developer documentation, SKAN endpoints will be deprecated by late 2026.
Key insight: AdAttributionKit is not an upgrade to SKAN. It is a full replacement with a new API surface and new capabilities.
- Replaces SKAN entirely, not a version upgrade
- Developer postbacks solve the black-box problem
- Supports re-engagement attribution natively
- Works across App Store and alternative marketplaces
- Privacy model retained: crowd anonymity, no IDFA
Apple introduced SKAdNetwork in 2018 and iterated through four versions, each adding complexity while trying to preserve user privacy. The fundamental architecture had limitations that couldn't be patched, particularly around re-engagement and developer-side postback delivery.
AdAttributionKit addresses these structural gaps. It retains SKAN's core privacy model (no user-level data, crowd anonymity tiers, conversion value windows) but rebuilds the delivery mechanism. The biggest shift: postbacks now go to both the ad network AND the developer, eliminating the black-box problem that plagued SKAN's network-only postback model.
According to Apple's WWDC 2024 session on AdAttributionKit, the framework also supports both App Store distribution and alternative marketplaces under the EU's Digital Markets Act. This future-proofs attribution regardless of distribution channel.
How is AdAttributionKit different from SKAN 4.0?
The three biggest differences are re-engagement support, developer postbacks, and a cleaner conversion window model. According to Apple's AdAttributionKit documentation, developers now receive postbacks directly instead of relying solely on ad networks to forward them.
Key insight: Developer postbacks are the single most impactful change, giving app marketers direct access to attributed conversion data for the first time since ATT.
- Developer postbacks: direct attribution data to your servers
- Re-engagement: measures lapsed user return for first time
- Cleaner conversion windows replace SKAN's timer confusion
- Same crowd anonymity tiers with improved thresholds
- Supports alternative app marketplaces under DMA
| Feature | SKAN 4.0 | AdAttributionKit |
|---|---|---|
| Developer postbacks | No (network-only) | Yes (developer + network) |
| Re-engagement measurement | Not supported | Supported |
| Conversion windows | 3 windows (0-2d, 3-7d, 8-35d) | Similar windows, simplified triggers |
| Postback tiers | 3 tiers based on crowd anonymity | 3 tiers, improved threshold logic |
| Alternative marketplace support | No | Yes |
| Fine-grained conversion value | 6-bit (0-63) | 6-bit (0-63), coarse available |
Under SKAN, only the winning ad network received postbacks. Developers had to trust networks to share that data accurately. This created measurement gaps and discrepancies that made optimizing campaigns with limited SKAN data one of the hardest problems in mobile UA.
AdAttributionKit sends a copy of each postback to the developer's registered endpoint. This means your MMP and your internal BI systems can ingest attribution signals directly. According to AppsFlyer's SKAN benchmark data, network-reported conversions diverged from MMP-reported conversions by 15-30% on average under SKAN 4.0.
Developer postbacks should close that gap significantly.
Re-engagement support is the second major unlock. SKAN had no mechanism for measuring whether an ad drove a lapsed user back into an app.
AdAttributionKit introduces re-engagement postbacks with the same privacy protections, which matters enormously since AppsFlyer's retargeting report shows retargeted users convert at 2-3x the rate of new installs on iOS.
How do AdAttributionKit's privacy thresholds work compared to SKAN?
AdAttributionKit retains crowd anonymity tiers (Tier 0, 1, 2, 3) but adjusts threshold logic so more campaigns qualify for higher-fidelity data. Per Apple's data quality guidance, the system factors in install volume, privacy signals, and campaign diversity.
Key insight: More campaigns will hit Tier 2 and 3 under AdAttributionKit because Apple refined how thresholds account for campaign diversity.
- Tiers 0-3 retained from SKAN 4.0 structure
- Threshold logic now factors campaign diversity
- Tier 3 unlocks full 6-bit conversion + 4-digit source ID
- Consolidating campaigns improves tier qualification
- Apple will never disclose exact threshold numbers
| Tier | Source ID Digits | Conversion Value | Practical Signal |
|---|---|---|---|
| Tier 0 | 0 digits | None | Install count only |
| Tier 1 | 2 digits | Coarse (low/medium/high) | Basic optimization possible |
| Tier 2 | 3 digits | Fine-grained (0-63) | Campaign-level optimization |
| Tier 3 | 4 digits | Fine-grained (0-63) | Full creative + channel attribution |
Under SKAN 4.0, many advertisers saw 40-60% of postbacks arrive at Tier 0 or Tier 1, according to Singular's SKAN benchmarks report. At those tiers, you receive only a coarse conversion value (or none at all) and limited source identifier granularity.
AdAttributionKit's threshold model considers additional signals beyond raw install count. Apple hasn't disclosed exact thresholds (and never will), but the documentation references "sufficient diversity in campaign identifiers" as a factor. Practically, this means consolidating campaigns into fewer, higher-volume ad sets improves your odds of reaching Tier 2 or 3.
For context, Tier 3 unlocks the full 6-bit fine-grained conversion value plus a 4-digit source identifier. At scale, this provides enough signal to run meaningful measurement frameworks and media mix models.
How does re-engagement attribution work in AdAttributionKit?
AdAttributionKit measures re-engagement by attributing app-open events from ads shown to existing users. This is distinct from install attribution and uses separate postback flows. According to Apple, re-engagement postbacks follow the same crowd anonymity tiers as install postbacks.
Key insight: For the first time since ATT, iOS marketers can measure whether retargeting ads actually re-activate lapsed users with Apple-sanctioned attribution.
- Separate postback flow from install attribution
- Supports unique conversion value schemas for re-engagement
- Follows same crowd anonymity tier system
- Both network and developer receive postbacks
- Eliminates need for probabilistic fingerprinting workarounds
Before AdAttributionKit, re-engagement measurement on iOS was essentially broken at the deterministic level. Marketers relied on probabilistic fingerprinting (which Apple has been cracking down on) or self-reported attribution hacks post-ATT.
AdAttributionKit's re-engagement flow works by recognizing that a user already has the app installed when they tap an ad. Instead of generating an install postback, it fires a re-engagement postback to both the network and developer endpoints.
The conversion value schema can differ from your install schema, letting you track re-engagement specific events.
This is a massive unlock for subscription apps. According to RevenueCat's State of Subscription Apps 2025, 28% of subscription revenue comes from reactivated lapsed subscribers. Without attribution, that spend was essentially flying blind.
What is the migration path from SKAN 4.0 to AdAttributionKit?
Apple supports running SKAN 4.0 and AdAttributionKit simultaneously during a transition period. Per Apple's documentation, SKAN endpoints will be deprecated in late 2026. Most MMPs (AppsFlyer, Adjust, Singular, Kochava) already support both frameworks as of Q1 2025.
Key insight: Run both frameworks in parallel now. Do not wait for SKAN deprecation to start your AdAttributionKit integration.
Need help scaling your mobile app growth? Talk to RocketShip HQ about how we apply these strategies for apps spending $50K+/month on UA.
- Both frameworks run in parallel during transition
- SKAN deprecation expected late 2026
- Conversion value schemas transfer directly
- Register developer postback endpoint first
- Watch for deduplication issues during overlap
The migration is less painful than SKAN 3.0 to 4.0 was because AdAttributionKit shares many architectural concepts. Your conversion value schema (the 6-bit fine-grained value and 3-level coarse value) transfers directly. The postback structure is similar but delivered to an additional endpoint.
Here's the practical migration order. First, register your developer postback endpoint with Apple. Second, update your MMP SDK to a version that supports AdAttributionKit (all major MMPs shipped this in 2024-2025). Third, configure your conversion value mapping in both SKAN and AdAttributionKit. Fourth, validate postback delivery in both systems.
The trickiest part is reconciling postbacks from both systems during the overlap period. You'll receive duplicate attribution signals for the same installs. Your MMP should handle deduplication, but verify this explicitly.
According to Adjust's iOS attribution guide, 12-18% of advertisers experienced postback deduplication issues when running SKAN 3.0 and 4.0 simultaneously.
What should I do if my MMP doesn't support AdAttributionKit yet?
Every major MMP (AppsFlyer, Adjust, Singular, Kochava, Branch) shipped AdAttributionKit support by early 2025. If you're on a smaller MMP that hasn't updated, escalate immediately. You need at minimum 6 months of parallel running before SKAN deprecation to validate data parity.
Also confirm that your MMP handles developer postback ingestion properly. This is a net-new data flow that didn't exist under SKAN. Some MMPs route these through different dashboards or require separate configuration.
How should I set up conversion value schemas for AdAttributionKit?
Your conversion value strategy should prioritize revenue signals for subscription apps and engagement depth for ad-monetized apps. The 6-bit fine-grained value (0-63) remains the same as SKAN 4.0. According to Apple's guidance, mapping revenue buckets to conversion values produces the highest-quality optimization signals.
Key insight: Map your conversion values to predicted LTV buckets, not raw event counts. Algorithms optimize toward what you measure.
- 6-bit fine-grained value (0-63) unchanged from SKAN 4.0
- Map values to revenue/LTV buckets, not raw events
- Coarse values critical at Tier 1 (most common tier)
- Separate schemas allowed for re-engagement vs. install
- Test schema changes during parallel SKAN overlap period
The best-performing schemas we've seen in the industry use a tiered revenue approach. For subscription apps, the conversion value encodes trial start, subscription start, and revenue bucket. For gaming apps, it typically encodes engagement milestones that correlate with retention.
A wellness subscription app case study demonstrated that optimizing SKAN conversion schemas contributed to achieving 150%+ year 1 ROAS alongside rigorous creative testing and multi-channel expansion. The same schema principles apply directly to AdAttributionKit.
The coarse conversion value (low/medium/high) matters more than most teams realize. At Tier 1 (the most common tier for smaller advertisers), coarse values are all you get. Map them carefully: "low" should mean the user is clearly non-monetizing, "medium" should indicate engagement potential, and "high" should signal a revenue event.
This gives ad network algorithms enough signal to optimize even with limited data. For deeper guidance, see how to optimize campaigns with limited SKAN data.
What is the AdAttributionKit adoption timeline for 2025-2026?
AdAttributionKit shipped in iOS 17.4 (March 2024) and has been production-ready since iOS 18. Apple is expected to deprecate SKAN endpoints by late 2026. According to StatCounter data, 82%+ of active iOS devices run iOS 17 or higher as of early 2025.
Key insight: With 82%+ of iOS devices on iOS 17+, AdAttributionKit already has the install base needed for full deployment.
- iOS 17.4+ required (March 2024 launch)
- 82%+ device penetration already achieved
- Major ad networks support it (Meta, Google, TikTok, ASA)
- SKAN deprecation expected H2 2026
- Prioritize AdAttributionKit postbacks by H2 2025
| Timeline | Milestone | Action Required |
|---|---|---|
| H1 2025 | Parallel running | Register developer endpoint, validate parity |
| H2 2025 | Primary migration | Shift optimization to AdAttributionKit postbacks |
| H1 2026 | AdAttributionKit primary | Phase out SKAN dependency in dashboards |
| H2 2026 | Expected SKAN deprecation | Full cutover, remove SKAN code paths |
The adoption curve for AdAttributionKit is faster than SKAN 4.0's was. SKAN 4.0 required iOS 16.1, and it took over a year for sufficient device penetration. AdAttributionKit benefits from Apple's aggressive iOS update push and the fact that most active purchasers run newer OS versions.
Here's the practical timeline. In H1 2025, run both frameworks in parallel and validate data parity. In H2 2025, begin shifting optimization decisions to AdAttributionKit postbacks. In H1 2026, AdAttributionKit should be your primary attribution source. By H2 2026, expect Apple to formally deprecate SKAN.
Ad networks are also moving. Meta, Google, TikTok, and Apple Search Ads all support AdAttributionKit postbacks. Smaller networks are the laggards, so check your long-tail partners specifically.
How does AdAttributionKit affect Apple Search Ads measurement?
Apple Search Ads continues to have privileged access to deterministic attribution via its own framework, separate from AdAttributionKit. However, custom product pages paired with Apple Search Ads campaigns now interact with AdAttributionKit's postback system for cross-channel deduplication.
Key insight: Apple Search Ads still gets its own deterministic attribution, but AdAttributionKit improves cross-channel deduplication with other paid sources.
- ASA keeps its own deterministic attribution system
- Developer postbacks improve cross-channel deduplication
- Custom product pages get better attribution pairing
- ASA + AdAttributionKit gives most complete picture
- Product page A/B tests correlate with conversion values
Apple Search Ads has always existed in a measurement silo. It provides campaign-level attribution through its own API without relying on SKAN or now AdAttributionKit. This won't change.
The advantage of running ASA alongside other channels is that you get both ASA's own deterministic data and AdAttributionKit postbacks from other networks.
Where it gets interesting is deduplication. Under SKAN, if a user saw a Meta ad and then searched and installed via ASA, the attribution could go to either source depending on timing. AdAttributionKit's developer postbacks give you visibility into both touchpoints, enabling better measurement framework design.
Pairing A/B testing on App Store product pages with AdAttributionKit conversion data also opens new optimization paths. You can now correlate product page variants with downstream conversion values at the aggregate level.
How should I monitor campaign performance during the SKAN to AdAttributionKit transition?
During parallel running, compare postback volumes and conversion value distributions between SKAN and AdAttributionKit daily. Expect 5-15% variance initially due to timing differences in postback delivery, per industry observations from early adopters reported by Singular.
Key insight: Use RocketShip HQ's Weighted Anomaly Scoring approach to flag real performance changes vs. attribution framework noise during migration.
- Compare postback volumes between both frameworks daily
- Expect 5-15% variance during parallel running
- Use weighted anomaly scoring to avoid false alarms
- Watch for systematic directional biases by channel
- Monitor fraud signals during framework transition
The transition period will be noisy. You'll receive postbacks from both systems, and the numbers won't match perfectly. This is normal. The risk is overreacting to discrepancies that are framework artifacts, not actual performance changes.
RocketShip HQ's Weighted Anomaly Scoring methodology helps here. Weight metric changes by business impact, not just percentage: abs(% change) x sqrt(spend). A 15% ROAS drop on $5K/day spend scores higher than a 40% drop on $200/day spend.
This eliminates 70%+ of false alarms, which is critical when your attribution data is inherently noisier during migration.
Also set up privacy-first attribution dashboards that display both SKAN and AdAttributionKit data side by side. Watch for systematic directional differences, not exact number matching. If one framework consistently shows higher conversion rates for a specific channel, investigate the postback delivery configuration for that network.
Be especially vigilant about ad fraud during the transition. Framework changes historically create windows where fraudulent installs can slip through validation gaps.
Does AdAttributionKit work with Privacy Sandbox on Android?
No, AdAttributionKit is Apple-only. Google's Privacy Sandbox for Android uses a completely separate Attribution Reporting API. However, both frameworks share philosophical alignment: aggregate measurement, no user-level tracking, crowd anonymity thresholds.
Key insight: Plan for two parallel attribution frameworks: AdAttributionKit on iOS and Attribution Reporting API on Android.
- AdAttributionKit is iOS-only, not cross-platform
- Android uses Privacy Sandbox Attribution Reporting API
- Both use aggregate measurement, no user-level tracking
- MMPs abstract differences in unified dashboards
- Debug platform-specific when troubleshooting discrepancies
Cross-platform marketers now manage two privacy-preserving attribution systems that work differently but solve the same problem. The practical impact is that your conversion value strategy, postback infrastructure, and measurement dashboards need platform-specific configurations.
According to Google's Privacy Sandbox documentation, the Attribution Reporting API supports both event-level and aggregate reports. This is structurally different from AdAttributionKit's tier system. Android gives you more granular event-level data (with noise added) while Apple gives you clean data at reduced granularity.
The good news is that most MMPs abstract this complexity. Your dashboard shows unified metrics regardless of which underlying attribution framework generated the signal. But understanding the differences matters when you're debugging discrepancies or building incrementality-based measurement frameworks.
What impact will AdAttributionKit have on Meta and Google Ads campaigns for apps?
Meta and Google both support AdAttributionKit postbacks, and developer postbacks should improve the data available for their optimization algorithms. According to Meta's App Ads documentation, campaigns using higher-quality conversion signals see 15-25% lower CPA compared to campaigns optimized on coarse data.
Key insight: Developer postbacks create a feedback loop that should improve algorithmic optimization on Meta and Google over time.
- Meta and Google both support AdAttributionKit
- Developer postbacks reduce trust gaps with networks
- Better conversion signals improve algorithmic targeting
- Lookalike modeling benefits from cleaner data
- Ensure MMP forwards postbacks to all networks
Under SKAN, Meta and Google received postbacks but developers couldn't verify them independently. This created a trust gap. With AdAttributionKit, both parties see the same data, which should improve signal quality and reduce discrepancy-driven budget misallocation.
Lookalike audiences on Meta benefit particularly because the platform receives cleaner conversion signals to model from. The better the conversion data, the better the algorithmic targeting.
Google's App campaigns (formerly UAC) will similarly benefit from improved signal flow. For both platforms, the practical advice is the same: optimize your conversion value schema for the signals that matter most to your business, and ensure your MMP is properly forwarding AdAttributionKit postbacks to both networks.
AdAttributionKit is the future of iOS attribution. Start parallel running today, register your developer postback endpoint, and audit your conversion value schemas. The advertisers who migrate early and build measurement infrastructure around developer postbacks will have a significant optimization advantage by H2 2026 when SKAN goes dark.
Frequently Asked Questions
Can I still use SKAdNetwork after AdAttributionKit launches?
Yes, during the transition period through late 2026. Apple supports parallel running of both frameworks. However, per Apple's developer documentation, SKAN endpoints are expected to be deprecated in H2 2026, so plan your migration now.
Do I need to change my MMP setup for AdAttributionKit?
Yes. You need to register a developer postback endpoint and update your MMP SDK to a version supporting AdAttributionKit. All major MMPs (AppsFlyer, Adjust, Singular, Kochava) shipped support by early 2025. Check your specific SDK version requirements.
Does AdAttributionKit support view-through attribution?
Yes. AdAttributionKit supports both click-through and view-through attribution, similar to SKAN 4.0. View-through postbacks follow the same crowd anonymity tier system but, according to Apple's docs, typically receive lower tier data due to weaker intent signals.
Will AdAttributionKit solve the SKAN data delay problem?
Partially. Postback delivery windows remain in place for privacy (24-48 hours for the first postback). However, developer postbacks eliminate the additional delay caused by waiting for ad networks to forward data. Industry observation suggests this can reduce effective reporting lag by 12-24 hours.
How does AdAttributionKit handle web-to-app attribution?
AdAttributionKit includes support for attributing installs that originate from web ads directing users to the App Store. Per Apple's WWDC 2024 presentation, web-to-app flows use the same postback mechanism and privacy tiers. This is an improvement over SKAN 4.0, which had limited web attribution support.
Is AdAttributionKit relevant for apps with small install volumes?
Yes, but expect most postbacks at Tier 0 or 1. Small-volume apps rarely hit the crowd anonymity thresholds needed for Tier 2-3 data. Focus your conversion value schema on coarse values (low/medium/high) since that's likely all you'll receive. ASO and keyword optimization can complement paid attribution for smaller apps.
Does AdAttributionKit work with ad creatives testing and optimization?
Yes. At Tier 3, the 4-digit source identifier provides enough granularity to attribute performance to specific creative variants. According to Singular's benchmarks, campaigns reaching Tier 3 can distinguish between up to 10,000 combinations of campaign, ad group, and creative identifiers.
Looking to scale your mobile app growth with performance creative that delivers results? Talk to RocketShip HQ to learn how our frameworks can work for your app.
Not ready yet? Get strategies and tips from the leading edge of mobile growth in a generative AI world: subscribe to our newsletter.
Related Reading
- Privacy-first attribution and measurement for mobile apps (comprehensive guide)
- AppsFlyer App Retargeting Report: Benchmarks and Post-ATT Strategies (2026)
- AppsFlyer Mobile Ad Fraud Report: Fraud Rates and Protection Benchmarks (2026)
- How Has ATT Changed Mobile Advertising? (2026)
- How to Use Lookalike Audiences for Mobile App UA on Meta
