How it works
The model is the same as in SKAN: the ad network registers a signed impression, the device attributes the install, the app updates a conversion value and Apple sends delayed postbacks with no user-level identifiers. AdAttributionKit uses the same three conversion windows and the same fine and coarse conversion values as SKAN 4. On top of that it can attribute installs from alternative app marketplaces as well as from the App Store, and it can measure re-engagement, where a user who already has the app opens it after tapping an ad. Impressions registered through SKAN and through AdAttributionKit compete with each other, and only one of them wins the install.
Why it matters for an app team
Newer capabilities have been added to AdAttributionKit rather than to SKAN, so a team planning iOS measurement needs both on its radar. An app that already has SKAN conversion values in place can reuse the same schema, which keeps the change small.
In practice
Example: a user who installed your app three months ago taps a retargeting ad, and the app opens on a product page. With AdAttributionKit the ad network can receive a re-engagement postback for that open. SKAN alone would report nothing, because no new install took place.
Common confusion
AdAttributionKit does not bring back user-level data, and it does not remove the need for App Tracking Transparency if you want the IDFA. It is also not a switch you flip instead of SKAN. Ad networks adopt it at their own pace, so both frameworks deliver postbacks side by side.