iOS 17 · Privacy
What is left of app tracking?
Where We Stand in 2026
ATT opt-in is at 22 to 38%. Remaining users are estimated via SKAdNetwork 5 or probabilistic attribution. This means app performance reporting has a 60 to 78% estimation component. Optimizing on pure app data is risky.
01
22-38%
ATT Opt-In Rate
02
SKAd 5
Aggregated Data
03
Server-Side
Clean Alternative
The 4 Tracking Layers
- ATT opt-in: clear user IDs, highest data quality
- SKAdNetwork 5: aggregated conversion data without user ID
- Probabilistic attribution: estimates via device fingerprint
- Server-side with user match data: the cleanest layer
Example
A telehealth app built its server-side setup on user match data with ATT opt-in at 22%. EMQ exceeded 8.5 despite IDFA restrictions, and algorithm performance outperformed ATT-only tracking.
What Most Brands Get Wrong
Using ATT tracking alone without server-side. Treating probabilistic attribution as the primary data source. Reporting on SKAd 5 without a multi-touch layer. The result: inaccurate optimization and a reporting dispute between marketing and finance.
„iOS 17 is not the end of tracking. It is the end of easy tracking.”