The IDFA is gone for most users. Mobile analytics had to reinvent itself — and the new playbook is finally settled.
When Apple shipped App Tracking Transparency in 2021, opt-in rates dropped to under 25% almost immediately. The advertising and analytics industry that ran on the IDFA had to find a new model. Five years in, the patterns have stabilized.
SKAdNetwork: aggregate attribution
Apple's SKAdNetwork (now SKAN 5) gives advertisers aggregated install attribution without exposing per-user identifiers. Conversion values are coarse, the data arrives delayed, and you cannot retarget — but you can still measure campaigns and optimize spend.
On-device analytics is back
Apps now do more analytics on-device. Compute the metric locally, transmit only aggregated counts to the server. Apple's on-device speech, intelligence, and intelligence frameworks all use this pattern. So do well-designed product analytics.
Privacy-preserving attribution APIs
Apple's PCM and the W3C's Private Click Measurement provide cryptographic attribution that does not rely on identifiers. Coverage is still building, but for top-of-funnel measurement it works.
First-party identity matters more than ever
The cross-app identifier is gone. The within-app identifier — your user ID, tied to email, tied to a logged-in session — is the new ground truth. Build your analytics around that, not around fragile device identifiers.
What to stop doing
Stop relying on pre-2021 attribution SDKs that pretend the IDFA is still available. Stop using SDKs that fingerprint device characteristics — Apple is increasingly hostile to this and has rejected apps for it. Stop assuming user-level data is available; design as if it is not.