What SKAdNetwork is
SKAdNetwork (SKAN) is Apple's native framework for attributing app installs and a limited set of post-install events to the ad network that drove them, without exposing device- or user-level identifiers to advertisers or ad networks. It's the mechanism that replaced device-level IDFA attribution once App Tracking Transparency made that identifier opt-in - see the broader context on our mobile attribution page.
How it actually works
- Ad network registration. Apple, not the advertiser, tracks which ad network showed which ad.
- Conversion value. The app can encode a limited number of post-install signals (a 6-bit value, expanded in later SKAN versions) representing things like "user completed onboarding" or "user made a purchase in tier X" - not the actual event or amount.
- Delayed, randomized postback. Apple sends the attribution result back to the ad network after a randomized delay (historically up to 24-48 hours per conversion window), specifically to prevent the timing itself from being used to re-identify a user.
- Aggregated reporting. Results are only reported once a minimum install-count threshold ("crowd anonymity") is met for a given campaign - small campaigns may get no data back at all.
What this costs you, precisely
| SKAdNetwork | Deterministic (pre-ATT) attribution |
|---|---|
| Aggregated, delayed postbacks | Real-time, per-user event data |
| A handful of encoded conversion-value bits | Full event names, properties, and revenue amounts |
| No reporting below a minimum install threshold | Every install and event reported individually |
Where an MMP still adds value on top of SKAdNetwork
SKAdNetwork alone gives you Apple's raw postback, network by network, with no cross-network deduplication or dashboarding. A mobile measurement partner (AppsFlyer, Adjust, Branch - compared on our mobile attribution page) normalizes postbacks from every ad network into one consistent report, models conversion-value schemas so the limited bits map to meaningful business events, and layers in probabilistic modeling to fill some of the gap SKAdNetwork's aggregation leaves behind.