SKAdNetwork (SKAN)

SKAdNetwork, usually shortened to SKAN, was introduced by Apple in iOS 14.5 (released April 2021) as the replacement for IDFAGlossaryIdentifier for Advertisers (IDFA)Apple's per-device advertising identifier — historically used for deterministic ad attribution, now only available when the user explicitly opts in via App Tracking...-based attribution after App Tracking Transparency (ATT)GlossaryApp Tracking Transparency (ATT)Apple's iOS 14.5+ framework requiring apps to display a system prompt and obtain user consent before accessing the IDFA or tracking the user across other apps and we... took effect. When a user taps a SKAN-enabled ad and installs the app, Apple — not the ad networkGlossaryAd NetworkA platform that aggregates ad demand from many advertisers and matches it to publisher inventory through real-time auctions. — signs a cryptographic postback that credits the install to the right campaign, 24-48 hours later, with aggregated and fuzzed conversion data.

The SKAN era now effectively covers the entire iOS measurement landscape. 88% of iOS apps in MWM's catalog have been updated since ATT launched — meaning they ship in a world where the IDFA is opt-in by default. 67% of iOS apps in the catalog launched fresh in the post-ATT era, having never relied on deterministic IDFA-based attribution at all. The pre-ATT measurement playbook (cookie-style identifiers, deterministic match) is a legacy practice that applies to a shrinking minority of installed apps.

SKAN 4.0 (current version) supports three postback "windows" that fire at different times after install — typically 0-2 days, 3-7 days, and 8-35 days — allowing advertisers to see early, mid, and late conversion signals. Each postback includes a conversion value (0-63 on older versions, coarse/fine pairs on SKAN 4) that advertisers encode post-install events into. The conversion-value schema is the most important UA configuration decision: it permanently determines what behaviors you can measure.

Trade-offs vs. classic device-level attribution: SKAN is privacy-respectful and works regardless of the user's ATT opt-in status, but it gives you much less granularity. There's no user-level LTVGlossaryLifetime Value (LTV)The total revenue (or gross profit) you expect a single user to generate over their entire relationship with the app — the denominator of every UA decision., cohort size has to clear a privacy-suppression threshold before postbacks fire, and the 24-48 hour delay makes intra-day optimization impossible. Most iOS UA teams now run SKAN and any available probabilistic or — for the 20-40% of users who opt in — device-level data in parallel.

Android has no direct SKAN equivalent today. The Google Play Services Advertising IDGlossaryDevice IDA unique identifier assigned to a mobile device (or its advertising profile) used to recognize the same device across apps and sessions — the backbone of determinist... (GAIDGlossaryGAID (Google Advertising ID)Google's per-device advertising identifier on Android — the equivalent of Apple's IDFA. Still broadly available in 2026, but Privacy Sandbox for Android will eventua...) is still broadly available, though Google's Privacy SandboxGlossaryPrivacy Sandbox (Android)Google's multi-year initiative to introduce privacy-preserving alternatives to GAID-based ad tracking on Android — the platform's eventual SKAN equivalent. for Android (in phased rollout) will introduce SKAN-like aggregated attribution by ~2027. Until then, Android attribution remains deterministic via GAID for the bulk of users — a contrast that's now central to how cross-platform UA teams allocate budget and design measurement.