What this bench is for
Cross-Platform App Analytics is usually taught as a tour of tools. This programme is a tour of disagreements. You bring one journey — signup or first purchase is enough — and we spend eight weeks forcing iOS, Android, web, and server logs to sit in the same conversation without a fantasy identity graph.
You will not leave with a single number finance, growth, and product all accept. You will leave with a reconciliation method, a surface inventory, and a short contract that names the gaps. That limitation is intentional. Teams that want a vendor to “just align it” should not sit this bench.
Modules
-
01
Surface inventory
Package names, bundle IDs, web hosts, webviews, widgets, instant apps, and the cron jobs nobody listed. If it can emit, it gets a row on the Atlas before any dashboard is opened.
-
02
Identity without a fantasy graph
When to join IDFV, GAID, email hash, and first-party user id — and when joining is a lie. Probabilistic matching is discussed only as a risk, not as a feature.
-
03
Windows that survive the store and the OS
Click-through, view-through, SKAN conversion values, and web UTMs that die in a webview. You will write windows as sentences, then as filters.
-
04
SKAN, MMP, warehouse
Postbacks beside MMP exports beside ledger callbacks. Modelled events stay labelled. Null conversions stop being treated as analyst error.
-
05
Consent, ATT, and the missing day-zero
What the first twenty-four hours look like when a prompt delays, a banner blocks, or an OEM quietly drops the session end.
-
06
The contract product will sign
Definitions, exclusions, dual reporting (“media-readable” vs “ledger-readable”), and the paragraph that prevents the next all-hands argument.
Learning outcomes
- Produce a surface map for one live product, including forgotten emitters.
- Write identity rules that a lawyer and an analyst can both initial.
- Reconcile at least one week of SKAN / MMP / warehouse exports with documented remainder.
- Publish a one-journey measurement contract with explicit non-goals.
- Explain, in plain English, why two official reports are allowed to diverge.
Instructor
Waranya “Ann” Srisuk
Former measurement lead at a Bangkok marketplace super-app. Ann ran the weekly reconciliation between SKAN, an MMP, and a warehouse that finance actually trusted. She teaches modules 03–06 and marks every contract for wishful joining.
Guest clinics are occasionally led by alumni now sitting inside banks and travel marketplaces. They do not pitch tools.
Informational tuition
As a Bench Seat programme: ฿28,900 per person for the eight-week cohort, including peer review and twelve months on the alumni reading list. Field Ledger access alone does not include live marking. Steward Circle companies may sit the same syllabus privately — write to the studio. There is no checkout on this page.
FAQ
Do you teach us to build or replace an MMP?
No. That is the real limitation. We do not cover building an MMP from scratch, migrating vendors as a professional service, or promising a single number that finance, growth, and product will all accept. You leave with a reconciliation method and a contract, not a magic dashboard or an implementation team.
How technical do I need to be?
You should be able to read a CSV export and a simple SQL select. You do not need to be the person who owns Airflow. Product managers who cannot query still succeed if they write the contract language and pair with someone who can run the filter.
Can I bring production data?
Yes, with names and emails stripped. We never ask for a live production credential. If your security team forbids even sampled exports, we work from public synthetic sets and you apply the method off-camera.
What if I miss a Bangkok evening session?
Recordings exist for Bench Seat, but peer review still has a deadline. Two missed reviews and we convert the remainder of your fee to Field Ledger access rather than pretend you sat the cohort.