SKAN
Apple’s privacy-preserving attribution framework for delayed iOS campaign measurement.
Staan vir
SKAdNetwork
Formele definisie
SKAN stands for SKAdNetwork, Apple’s privacy-preserving attribution framework that lets ad networks and measurement systems receive delayed postbacks with campaign and conversion-value signals instead of user-level identity.
Eenvoudige verduideliking
SKAN is Apple’s way of saying: you can still measure iOS campaign outcomes, but the signal is delayed, aggregated, and encoded into a small conversion schema.
Voorbeelde
- An iOS install campaign is measured through SKAN postbacks rather than device-level identifiers.
- A team compares creative performance using SKAN conversion values instead of raw user-level install logs.
- A game encodes day-1 retention, ad-show volume, and cumulative USD purchase value into the six fine-value bits available in the first postback window.
Hoe dit werk
- A user interacts with an ad on iOS and later installs or re-engages with the app.
- During postback window 1, the app can update a fine conversion value from 0 to 63. Attriax treats that as six bits and lets each bit range represent a group of ranked events.
- During postback windows 2 and 3, Apple only exposes coarse values: low, medium, or high. Attriax maps those values per event, not per group.
- Each event can optionally lock the SKAN window when that signal is valuable enough to stop waiting for later updates.
- Apple sends delayed postbacks, and Attriax interprets the saved schema so teams can compare campaign performance without exposing the full user journey.
Hoe Attriax dit gebruik
- Attriax stores one dashboard-managed schema with three panels: window 1 bit groups, window 2 coarse events, and window 3 coarse events.
- Window 1 events can carry their own coarse value and lock-window flag while later events in the same bit group encode higher ranked values.
- Window 2 and window 3 are coarse-only event lists; each event owns its coarse value, lock-window flag, and optional parameter conditions.
- The Flutter and Unity SDKs evaluate special SKAN signals locally: retention day, cumulative USD revenue, purchase count, and ad show count.
- For purchase revenue, the SDKs treat the `revenue` parameter as USD equivalent for SKAN rules. Non-USD purchase events can use the SDK revenue conversion endpoint before local matching.
- The schema editor includes templates for common app types so teams can start from a reasonable bit/coarse allocation and then tune it to their own event model.