What Is SKAN? What SKAdNetwork Gives and Withholds

In one line

SKAN (SKAdNetwork) is the framework Apple built so ad-driven installs can still be counted without a user identifier. It returns counts at the campaign level, aggregated, and days after the fact.

What it gives and what it does not

It gives you how many installs a campaign produced and one compressed value describing their early behavior. What it does not give is a user-level link — you cannot build the sentence "this person saw this ad and installed" from SKAN data.

Under SKAN 4 the measurement windows are days 0–2, 3–7, and 8–35 after install. The precise conversion value (0–63) arrives only in the first window, and only above the privacy threshold. The other windows return low, medium, or high — three levels, no more.

Where it bites in practice

Campaign structure. On other channels finer structure means better analysis; under SKAN each slice carries less install volume, drops below the threshold, and the data you get back gets coarser instead.

Go deeper

The relationship to ATT, window design, and conversion-value mapping are covered in the iOS ATT and SKAN measurement guide.

Frequently asked questions

How does SKAN differ from an MMP?

SKAN is anonymous aggregate signal returned by Apple itself; an MMP is a third-party tool that collects data from many networks and settles attribution. Most teams receive SKAN postbacks through their MMP and read them next to other data.

Why does SKAN data arrive late?

Apple adds a randomised delay after the measurement window closes, so an individual cannot be re-identified by timing. That makes SKAN unusable for real-time optimization.