Attribution Across Surfaces
SKAN postbacks, MMP logs, web UTMs, and the warehouse row that should have matched. You leave with a reconciliation method, not a single “true” number.
Each programme is built around a real multi-surface product problem, not a tour of vendor slides. The flagship remains Attribution Across Surfaces.
SKAN postbacks, MMP logs, web UTMs, and the warehouse row that should have matched. You leave with a reconciliation method, not a single “true” number.
Name what the client, the webview, and the backend actually emit. Kill zombie properties before they fossilise in BigQuery.
Holdout groups that still mean something after ATT, OEM restrictions, and a consent banner that actually blocks a pixel.
“The SKAN versus warehouse session in Attribution Across Surfaces is the first time our growth and data teams used the same vocabulary. We still argue about windows. We no longer argue about what a session is.”Pimchanok Rattanakul · Measurement lead, regional travel marketplace
Most teams buy another connector and hope the lines converge. We start with what the operating system, the store, and the SDK are allowed to tell you — then design reporting that survives that limit.
iOS, Android, web, webview, and server jobs each lie differently. We catalogue emitters before anyone opens Looker.
You will not get a perfect user spine. You will get rules for when to join, when to keep keys apart, and how to explain the gap to finance.
A measurement contract is a short document: definitions, windows, exclusions, and the sentence “this number will not match that number, and here is why.”
Cohorts meet in Ladyao and online. Case material draws on SEA super-apps, but the methods travel to any multi-surface product.
Tuition is published without checkout. Enrolment starts with a conversation.