Android DAU/MAU Charts:

As of June 2019 we have an archived source of validated DAU and MAU metrics generated by a script that pulls counts from MobileWikiAppSessions. Reporting for that dataset is available here.

Data Sources: MobileWikiAppSession and archive wmf.mobile_apps_unique_daily, wmf.mobile_apps_unique_monthly


Android Data Validation Observations:

  • Our prior mobile user counts for MAU are accurate but DAU is quite low which is likely due to a scripting issue when data is written to the DAU table.

  • For archived data we have user count datasets in tables:
    wmf.mobile_apps_uniques_daily
    wmf.mobile_apps_uniques_monthly
    This data is for in-app opt-in users only, so not an ideal source.

  • The DAU/MAU counts from Google Play are not a source of truth because of how they aggregate data. It’s still useful to have visibility to see growth or shrinkage trends.

  • Because we are building the MEP Modern Event Platform to replace current analytics it is not a priority to backwards engineer user counts, we are in good shape for current Android metrics, that data has been validated and serves it’s purpose for several stakeholders.




iOS DAU/MAU Charts:

Data Sources: iOS App Store


iOS Data Validation Observations:

  • There is a large discrepancy between our internal user counts and what we are seeing on the iOS App Store. Historically we have used the App Store data but in the future when the MEP platform comes online we will use that as a source for DAU/MAU reporting.
  • According to Apple, 33% of users have opted-in, within the app, to share data, those are the users counted in wmf.mobile_apps_unique_daily and wmf.mobile_apps_unique_monthly.
  • The counts reported above from App Store are also lower than reality, counting only users who have opted-in to share data with Apple on their device. See below for more in-depth information)
  • Consideration should be given to whether troubleshooting these issues with internal data collection is worth pursuing given available resources. An issue with moving from appinistallid value to app_install_id is likely worth investigating.




App Distribution By Country:





App Store Conversion Rates:


Android:


iOS:




App Store Pages:

Conversion rate for iOS app is ~50%, where Android is ~28%, from a marketing standpoint it might be worth trying some A/B testing on the Play Store app page to see if there is an increase in conversion. (The average app page conversion rate across all apps in the Apple App Store in the US is 34.34%, Google Play Store in the US is 24.13%).

Google Play Store:

 

Apple App Store:

Additional Information:

Android DAU/MAU monthly report:

https://analytics.wikimedia.org/published/datasets/wikipedia-android-app-reports/metrics/index.html

Further information on iOS user count issues:

iOS User Count Inaccuracy: https://phabricator.wikimedia.org/T193917

Future Data Collection: https://phabricator.wikimedia.org/T202664

mobile_apps_unique_monthly data discrepancy issue, sample data table:

Android MAU counts in wmf_mobile_users are as expected, iOS DAU and MAU counts are lower than App Store counts by a significant amount. Since the App Store counts aggregate, and are only device level opt-in users, relying on their DAU/MAU counts is not ideal.

Internal vs App Store Data Comparison:

Date app_mau wmf_mau percent_diff
2019-01-01 1042784 342977 67.11%
2019-02-01 1038961 327175 68.51%
2019-03-01 1035776 328942 68.24%
2019-04-01 1034456 320085 69.06%
2019-05-01 1047711 314038 70.03%
2019-06-01 1032449 302177 70.73%
2019-07-01 1037308 297768 71.29%
2019-08-01 1067346 296855 72.19%
2019-09-01 1108712 296576 73.25%
2019-10-01 1175330 294085 74.98%
2019-11-01 1196587 282912 76.36%
2019-12-01 1204490 282845 76.52%