When the MMP and the store console refuse to agree
App analytics teams in Southeast Asia often live with a quiet 6–12% gap between MMP installs and store consoles. Mobile attribution reporting starts when that gap is named, owned, and bounded — not when it disappears.
Finance does not need the two systems to match perfectly. They need to know whether this week’s drift is the same shape as last week’s, and who is responsible if it is not. Vendor chat threads that say “attribution is directional” do not survive a closing. A one-line tolerance does.
Pick a primary, then a band
Choose which figure may enter the signed weekly total. Many Thai retail apps use Play Console for Android volume and the MMP for paid/organic split. That is a valid rule if it is written down. What is not valid is switching primary source when the slide looks better. Around the primary, set a band — we often start at 3% for stable catalogues and widen it only with a cause. Inside the band: note the delta. Outside: stop publication until someone owns a cause.
Three causes we see constantly
First, time zones. MMP reports in UTC, store consoles in local Bangkok time, and a Sunday campaign bleeds into Monday. Second, uninstall and reinstall filters that one system applies and the other does not. Third, non-paid inflators: gift cards, preloads, employee devices, and agency test traffic sitting in “organic.” None of these is fraud. All of them move the delta.
Write the owner into the cell
A reconciliation sheet without a name is a hallway conversation. In The Attribution Ledger we add an owner column: analytics for definition drift, UA for test traffic, engineering for missing store credentials. The weekly memo quotes the delta and the owner in the same sentence.
If your gap is larger than 12% and stable, do not wait for a course to start the map. Download both CSVs for the same calendar week, in the same timezone, and list every filter you can find. Then send the studio a note if you want that map marked.