iOS Attribution in 2026: How to Stop Losing Data After ATT and Why You Need gBraid

Since iOS 14.5 shipped, mobile marketing has split into a "before" and an "after." Most teams still operate in partial blindness — they see only 30–50% of real conversions and make budget decisions on incomplete data.

Why ATT broke attribution as we knew it

Before ATT (App Tracking Transparency), attribution was built on the IDFA — a unique device identifier available to any ad SDK. Since iOS 14.5 the user has to grant tracking permission explicitly. The real consent rate in most apps is 25–40%, and in fintech it drops to 20–35%. That means for 60–80% of users, no ad channel gets deterministic attribution at all.

How gBraid works and why it's critical for Google Ads

gBraid is a parameter Google passes through the AdServices Framework (Apple's own attribution API), bypassing user consent. It works at the operating system level and doesn't require ATT permission. To put it to work, all you need is to add AdServices.framework in Xcode (Required vs Optional — choose Optional so you don't break deployment on older iOS versions). gBraid recovers attribution for ~40% of the traffic that would otherwise land in direct/none.

SKAdNetwork: what to read and what to ignore

SKAN 4.0 is the current version of Apple's privacy-preserving attribution. Key characteristics: the postback arrives with a delay of anywhere from 24 hours to 35 days depending on the crowd anonymity threshold. The fine conversion value gives you 6 bits (64 values) to pass in-app events — use them to encode LTV segments, not just install/no-install. For most teams SKAN is a source of trends, not exact numbers: don't optimize campaigns on SKAN data directly.

The right iOS attribution stack for 2025

The minimum viable stack: an MMP SDK (AppsFlyer or Adjust) + a Server-to-Server API for sending events from the backend + BigQuery as the reporting layer. The MMP gives you probabilistic matching and aggregated attribution. The S2S API lets you send purchases and subscriptions independently of the client SDK — critical on iOS, where the SDK can be blocked. BigQuery is what you need for cross-channel analysis without sampling.

What to do right now

First, check the timing of your ATT prompt: the optimal moment is after onboarding, once the user has already understood the app's value. The right timing adds +10–15 pp to consent rate. Second, configure your SKAN conversion value scheme around your product: for subscription apps, encode trial start and first payment. Third, connect gBraid if you haven't yet: it's literally 30 minutes of developer work and it restores 30–40% of Google Ads attribution.