iOS attribution in 2026: what ATT actually broke (and what still works)
A credited summary of Attriqs’ practical guide to attribution after ATT: why click-level truth is gone for most iOS users, how to think about SKAN vs AdAttributionKit, and the resilient stack (first-party + server-side + incrementality).
Original source: Attriqs - “iOS Privacy Changes and What They Broke in Attribution” (May 15, 2026)
The blunt reality: your highest-value users are often the least measurable
Attriqs frames ATT as a structural shift, not a UI annoyance:
- tracking is off unless explicitly accepted,
- IDFA-based stitching across apps disappears for most users,
- iOS-heavy funnels become partially “dark” to traditional dashboards.
That is why ROAS often looks suspiciously good on iOS-heavy channels: the dashboard is increasingly filled with modelled or partial signal.
Web is different, but not “safe”
They point out Safari was already constrained by ITP, and Apple’s approach to web measurement trends toward privacy-preserving aggregation (e.g., Private Click Measurement) rather than user-level tracing.
Practical implication: “drop a pixel and trust it” is no longer a measurement strategy on iOS.
App attribution: SKAdNetwork vs AdAttributionKit
Their useful clarification:
- SKAdNetwork is workable but delayed and coarse by design.
- AdAttributionKit is where new iOS attribution feature work is landing (with configurable windows and re-engagement concepts), but it still lives in the same aggregated, privacy-first world.
If your organisation is still expecting the granularity of 2020, you are benchmarking against a past that is not coming back.
What actually works now (the resilient stack)
Their recommended shape is a layered approach:
- First-party tracking on your own domain (your own source of truth).
- Server-to-server conversion delivery (Meta CAPI, TikTok Events API, enhanced conversions, offline imports).
- Multi-touch attribution on your own data (where possible) so platform claims are inputs, not outputs.
- Incrementality / lift tests as the truth layer (periodic recalibration).
- Aggregated frameworks for the gaps (SKAN / AdAttributionKit / AEM / PCM) so the dark areas are at least visible.
The point is not “one magic fix”. It is reducing dependency on any single identifier or dashboard.
Tiny win
Pick your two biggest paid channels.
- Ensure every conversion is captured first-party.
- Send the same conversion server-side.
- Then run one simple incrementality test this quarter.
If your reported ROAS and your lift do not rhyme, the problem is not creative. It is the measurement layer.
Read the original: https://www.attriqs.com/blog/ios-att-attribution/
Want help with ASO?
If you want this implemented for your app, check out our services - or run your workflow in APPlyzer.