Attribution desk
Mobile Attribution Analytics, drawn as a route rather than a slogan.
This page is the school’s working map. Use it to see whether a Model Routehub course is the right next step, or simply to name the parts of the stack that keep colliding in your Monday meeting.
What the phrase actually covers
Mobile Attribution Analytics is the practice of deciding which paid or owned touch deserves credit for an app event, then living with the uncertainty that decision creates. It is not “the MMP dashboard.” The MMP is one ledger. The ad network is another. The store, SKAN, finance, and your product analytics tool each keep their own.
When those ledgers disagree — and they will — the work is to explain the disagreement without inventing precision. That is the craft we teach.
Five stretches of the route
- Identity Device identifiers, probabilistic matching, customer IDs after login, and the limits of stitching across iOS, Android, and web. Thailand traffic often arrives on dual-SIM Android devices; that changes how confident a match can be.
- Collection SDK placement, event taxonomy, consent flags, and the unglamorous versioning of names. If “purchase” means three different things, no downstream model will save you.
- Credit rules Last click, multi-touch folklore, view-through windows, reattribution, and partner priority. Write the rule down. The rule will still be wrong sometimes; at least it will be the same wrong.
- Privacy postbacks SKAN conversion values, coarse postbacks, and Android privacy changes that do not arrive on the same calendar as Apple’s. Treat buckets as buckets.
- Truth tests Incrementality, geo holdouts, and reconciliation against store and finance. This is where vanity credit usually dies, which is the point.
Where teams in Bangkok usually get stuck
LINE and other chat-led inventory create long, messy paths between click and install. COD-style checkout delays revenue events past the window that a US playbook assumes. Android share is high enough that an iOS-only SKAN story leaves most spend unexplained.
We also see MMP “out of the box” partner settings left untouched for a year, then blamed when finance cannot reproduce last-click revenue. Configuration drift is a measurement problem. It is also a documentation problem.
If any of that sounds like your week, the course list is the practical next page. If you need a human to look at a specific discrepancy, write to the desk with the three ledgers you are holding.
