Unpacking the WWDC AdAttributionKit updates
Apple introduced a set of changes to the AdAttributionKit (AAK) marketing measurement framework at its WWDC event last week. The developer video outlines the updates, which fall into five new sets of functionality.
The first is multiple re-engagement conversion windows, which allows developers to use the conversion tag that was introduced in iOS 18.4 to set one re-engagement conversion window per tag. Prior to this change, any click on a re-engagement campaign ad overrode the conversion window that was active for the previous re-engagement campaign click. Now, if the conversion tags for the campaigns are different, both re-engagement conversion windows would remain live, meaning that conversions could be attributed to both clicks if the user completed the conversion actions for both. Apps must opt into this feature with a new key in the Info.plist file.
The second is custom attribution window settings per ad network. Previously, attribution windows in AAK were fixed at 30 days for clicks and 1 day for views. Now, a developer can set specific conversion window values globally and for each individual network. A developer can also exclude a specific conversion type (eg., ignore view-through) globally and for individual networks.
The third is the ability to set cooldown periods for engagements which apply across engagement types. These cooldown periods prevent subsequent conversions from being registered: any ad-driven conversion that takes place within a cooldown period will be ignored.
The fourth new piece of functionality is an optional countryCode field in the postback that is tied to the App Store storefront associated with the user’s Apple ID at the time of the original install (for installs from Alternative App Marketplaces, the countryCode field is set in the install verification token). The insertion of the countryCode parameter is moderated by Apple’s crowd anonymity standards (meaning: the field is left blank if too few conversions with the same attributes have been registered within some period of time).
And the last new area of functionality, which was introduced in iOS 18.4, is a set of testing features that allow developers to send development postbacks to validate AAK implementations absent real advertising events.
Broadly, I don’t view this new functionality as fundamentally changing the role that AAK plays in app install measurement: the developers that weren’t meaningfully relying on AAK prior to these features being introduced are not likely to prioritize it now. Of the features, the ability to operate multiple re-engagement windows in parallel strikes me as the most meaningful. But it’s important to note that AAK doesn’t allow for campaigns to be targeted specifically for re-engagement, only measured as such after the fact. And it seems bizarre to me that the countryCode field and the ability to set per-network attribution windows are only being introduced to AAK now, many years after SKAdNetwork was first introduced in 2018.
Although several exist, the fundamental dysfunction with SKAdNetwork (now, AdAttributionKit) is the random timer system that prevents postbacks from being received by a developer until crowd anonymity standards can be assessed. Ultimately, this is what caused SKAdNetwork to fail. Almost every other limitation of AAK can be navigated by advertisers, eg., the campaign limitations, the lack of creative keys in postbacks (which Apple’s attribution framework for Apple Ads, the AdServices Attribution API, allows), and coarse versus fine-grained conversion values. But without access to short-term feedback for campaign performance, AAK is relegated to secondary measurement use cases that don’t inform campaign optimization decisions, like media mix modeling.
My sense is that an annual drip-feed of new functionality can’t reverse the status quo, in which advertising networks very reluctantly and belatedly support updates to AAK. If Apple wants to reverse this dynamic, it needs a fundamental reset of AAK that addresses its core deficiency as well as some sort of incentive for advertisers to embrace it. Absent that, I don’t see AAK becoming an important part of the advertising measurement landscape.
Comments: