Bryan Bishop at The Verge:
As outlined on the Android Developers site, Google now uses the data collected when users visit the Google Play Store; under the previous system, any check-in to the store by the device would have been incorporated into the results, user-generated or not. The new system went into effect starting with this month’s results.
The change essentially skews the results towards those users who are actively visiting the Play Store. Google says as much on the page itself, noting that the new system “more accurately reflects those users who are most engaged in the Android and Google Play ecosystem.”
The Verge’s headline for this article: “Google changes how it measures Android version adoption, sees uptick in Jelly Bean devices”.
A more accurate title would be: “Google changes how it measures Android version adoption to show an uptick in Jelly Bean devices”.
You could argue that it better reflects the breakdown of OS versions among active buyers, but that means that Google’s statistics on version adoption can no longer be trusted to represent Android as a whole. Given that Play Store engagement seems about as disproportionally low as web-browsing marketshare on Android, these numbers now have far less relevance to the real Android market and aren’t useful for much.
It’s hard to see this change as anything but a desperate move by Google to attempt to hide Android’s poor update-adoption rates.