Photo by William Hook on Unsplash
SKAdNetwork's quiet retirement is one of the underreported attribution shifts of the year. Apple introduced AdAttributionKit (AAK) in 2024 and used WWDC 2025 to ship major updates: configurable attribution windows per network, configurable cooldowns, country codes in postbacks, and improved testing tools (Singular WWDC 2025 recap). AAK is the successor to whatever SKAN 5 was going to be.
SKAN 4 itself never crossed mainstream adoption. As of 2025, SKAN 4 adoption sits below 50% across the major platforms; TikTok runs SKAN 4, but Meta, Google, and Snap remain on SKAN 3 (SplitMetrics, 2025). Apple Search Ads, in a notable shift, joined SKAN's measurement framework in April 2025 after years of using its own attribution data (AppsFlyer, 2025). Despite the standard's age, the deterministic floor under iOS attribution is still ATT opt-in, which Adjust measured at 35% globally in Q2 2025 (SplitMetrics, 2025).
The pragmatic stack most teams actually use is probabilistic modelling on top of SKAN, not deterministic SKAN. AppsFlyer's modelled SKAN data, which uses Bayesian inference across null conversion values, reports an average 15% ROAS improvement versus un-modelled revenue and extends in-app event tracking to 7 days post-install (AppsFlyer Help Center, 2024-2025). 67% of app marketers reported at least some familiarity with AAK in AppsFlyer's April 2025 survey (AppsFlyer 2025 App Marketing Outlook).
The migration path: get the measurement vendor onto AAK in test mode now while SKAN 3 still carries production traffic, validate the postback contents against your existing attribution windows, and plan a six-month parallel-run before switching the source of truth. The platforms have not enforced an end date yet, but they will, and the teams that started in 2025 will not be the ones scrambling.