SKAdNetwork and SKAN 4, Explained for Gaming UA

Apple’s privacy-preserving attribution framework, what changed in SKAN 4, and what it means for how you buy users.

SKAdNetwork (often shortened to SKAN) is Apple’s framework for attributing app installs to ad campaigns without identifying individual users. It lets you measure which campaigns drive installs on iOS while keeping user data private.

Since Apple’s privacy changes, SKAN carries much of iOS attribution. Understanding how it works, and what SKAN 4 added, is now part of the job for anyone buying users on iOS.

Why does SKAdNetwork exist?

Before 2021, iOS attribution relied on the IDFA, a device identifier that let networks match an ad click to an install at the user level. App Tracking Transparency (ATT) made that identifier opt-in, and most users decline. SKAN is Apple’s replacement: it reports campaign performance in aggregate, so you learn what worked without tracking who did it.

How does SKAdNetwork work?

Simplified: when a user installs your app from an ad, the ad network registers the install with Apple. Your app records a conversion value (an encoded signal about what the user did after installing). After a privacy delay, Apple sends a postback to the ad network confirming the install and the conversion value, without revealing the user’s identity.

The conversion value is the key lever. It’s a small amount of information you get to define, so studios encode early signals of value, like a first purchase or a level reached, to judge campaign quality beyond the raw install.

What’s new in SKAN 4?

• Multiple postbacks. Up to three postbacks over a longer window, instead of one. You get an early signal and later signals, closer to real LTV.
• Coarse and fine conversion values. A fine value (detailed) when data volume is high enough, and a coarse value (low/medium/high) when it isn’t, so smaller campaigns still return a usable signal.
• Crowd anonymity tiers. Apple releases more detail only when enough installs protect anonymity. More scale means more granular data.
• Hierarchical source IDs and web-to-app. More room to encode campaign structure, and support for measuring web-to-app journeys.

What does SKAN mean for gaming UA?

Two practical realities. First, SKAN data is aggregated and delayed, so it can’t fully replace the user-level view studios once had; you plan your conversion-value schema carefully and accept less granularity. Second, SKAN is one signal among several. Modern measurement blends SKAN with the deterministic data you do have (from consenting users) and probabilistic modelling to fill gaps. A good MMP stitches these together so you can still make budget decisions. Turn SKAN, deterministic and modelled signals into one view you can act on. That’s what an MMP built for gaming is for.