All the projects I have worked on over the last few years haven’t been for paid apps. My clients have tended to give them away as part of some other service. Hence, it was with interest I read Russell Ivanovic’s post on "How New Versions of Android Work". He sells Pocket Casts for Android and has some interesting observations.
Russell says "People confuse overall numbers, with actual numbers of people who buy apps". The people who run the newer versions of Android are more likely to buy apps. For example, Russell says "Android 5.0 has less than 1% adoption in the overall Android eco-system, 23% of our customers already run it". For paid apps targeting Android 4.1 and above will get you the paying users. There’s no need to support users running older versions of Android.
If you are selling an app, you will want it to be the best of its genre. Often, supporting older Android versions dilutes the effort available to make the app the best. Also, supporting older screen idioms and APIs can also drag the app user experience down to the lowest common denominator. Russell’s insights suggest it’s possible for a paid app to focus on a better, latest, user experience for the paying subset of users.
It seems to be research release season and ABI has also reported numbers for smartphone shipments. However, this time we have a breakdown of what proportion of Android is forked (AOSP). That is, Android devices that aren’t sanctioned by Google, haven’t passed compatibility testing and don’t have Google Apps and Google Play Services (at least not legally anyway).
UPDATE: Changed value from 41% to 29% as the first row in the table, despite its generic naming ‘Android’, doesn’t include the second.