SKAdNetwork 5: what changed and what to ship this quarter

SKAdNetwork 5 has been generally available long enough that there's no excuse left for staying on SKAN 4. The headline changes are re-engagement attribution, coarse conversion values across three postback windows, and a richer crowd-anonymity tier. The migration is mostly mechanical — here's the checklist. (And everything below transfers to AdAttributionKit, which inherits SKAN's semantics on newer iOS.)

Update your conversion value schema

SKAN 5 keeps the fine conversion value (0–63) and the three coarse buckets across the three postback windows. Re-map your event values so 'trial started' lands in the right window, and audit the crowd-anonymity tier each event is likely to receive.

// Fine value 0–63; the coarse bucket is what later
// postback windows will actually report.
LinkTrail.shared?.updateConversionValue(42, coarseValue: .medium)

Handle re-engagement postbacks

Re-engagement is the most underused feature in SKAN 5. If your acquisition is mature and your reactivation budget is meaningful, you should be lighting this up — and you should be tagging postbacks accordingly in your warehouse.

Don't trust the deck — trust the postback

Ad networks publish their SKAN reporting in summary form, often with proprietary smoothing. Always ingest the raw postback yourself, store it indefinitely, and reconcile against the network's reports. If you're using LinkTrail, the postback endpoint is one route handler — no excuse not to.