Splitting Huawei/Xiaomi/OPPO/vivo cohorts in a teaching export
Measurement lab with illustrative rows showing how to split China Android OEM store cohorts when OAID-based attribution is available. Uses AppsFlyer OAID definitions; figures are illustrative inputs, not industry averages.
Metric definition
AppsFlyer’s OAID article frames OAID as the identifier for attributing installs from third-party Android app stores where Google Play Services / GAID are absent. This lab does not invent match rates; it shows a illustrative export shape a UA analyst can reproduce.
Illustrative export (not live data)
| store_cohort | attributed_installs_synthetic | oaid_present | notes |
|---|---|---|---|
| huawei_appgallery | 1200 | true | HMS or MSA OAID path |
| xiaomi_getapps | 900 | true | MSA OAID path |
| oppo_gamecenter | 750 | true | MSA OAID path |
| vivo_igame | 680 | true | MSA OAID path |
| unknown_oem | 150 | false | Teaching residual — investigate SDK gaps |
All install counts above are synthetic labels for spreadsheet practice. They are not Sensor Tower scrapes, not MMP pull quotes, and not CPI benchmarks.
Operator steps
- Require
store_cohort(or equivalent media-source × agency site ID) in the China Android export. - Require an
oaid_presentboolean from the MMP. - Refuse to average Huawei+Xiaomi+OPPO+vivo into a single “China Android CPI” without the split.
- Send unknown_oem rows back to engineering against the MSA integration checklist.
