How it works
The ad network signs each ad with campaign information. If the user installs and opens the app, iOS itself attributes the install to the last qualifying ad. The app can then set a conversion value to describe what the user did. SKAN 4 sends up to three postbacks, covering days 0–2, 3–7 and 8–35 after install. The first can carry a fine conversion value from 0 to 63 or a coarse one (low, medium or high); the second and third are coarse only. How much detail a postback includes depends on crowd anonymity, Apple's term for how many installs stand behind it. Postbacks are sent after a random delay: 24 to 48 hours after the first window closes, and 24 to 144 hours after the second and third.
Why it matters for an app team
For iOS users who have not allowed tracking through App Tracking Transparency, SKAN is the main way to see which campaigns produced installs. It works without the IDFA and needs no permission prompt, but the data is delayed, limited and cannot be joined to individual users.
In practice
Example with made-up numbers: a campaign drives 900 installs in a week. The first postbacks arrive a few days after each install. 700 of them carry a fine conversion value, and 200 carry a coarse value or none because those installs fell below Apple's privacy thresholds.
Common confusion
SKAN figures do not match user-level attribution and should not be added to it without deduplication, or the same install is counted twice. SKAN reports installs, not people: there is no device ID, no exact install time and no way to follow one user's revenue. Apple's newer framework, AdAttributionKit, builds on SKAN and works alongside it.