Attribution Across Surfaces
Reconcile SKAN, MMP logs, web UTMs, and warehouse rows without declaring a fake source of truth. Includes a real limitation: we will not build you an MMP.
Each programme is a bounded problem. The flagship, Attribution Across Surfaces, is the only one with a full public syllabus on this site. The others run in rotation — write to us if you need the next date.
Reconcile SKAN, MMP logs, web UTMs, and warehouse rows without declaring a fake source of truth. Includes a real limitation: we will not build you an MMP.
Catalogue every emitter, retire zombie properties, and write names that mean the same thing in iOS, Android, and the webview. Often paired with the Atlas.
Join what can be joined. Document the modelled remainder. Stop treating null conversions as a personal failure of the analyst.
Cohort charts that respect missing days, timezone seams, and the difference between “opened the app” and “did the job the app exists for”.
Holdouts that still mean something after ATT, OEM limits, and a banner that actually blocks a pixel. Kill tests before they ship empty.
See Field Ledger, Bench Seat, and Steward Circle before you write.