The Guardian newspaper blog has a thought provoking post on why an 80% market share might only represent half of smartphone users. The main insight is that the latest sales don’t necessarily reflect the installed base. This is something I have previously mentioned in the context of stats, fragmentation and as long ago as 2009 and 2008. The same reasoning might also be applied to PCs.
The best indication you can obtain is one that measures use, in the geographic region you are targeting, for the type of app you intend to create. Obviously, this might be a chicken and egg situation as you might not have created your app yet. However, if you are re-writing an existing app you might actually have the numbers and that’s where it get more interesting.
On a recent project, in the UK, we has usage across smartphone devices prior to a re-write. What was interesting was the use was very similar to the current reported sales shares. Why, I don’t know. This was just one case, for one type of app so it probably doesn’t prove much statistically. However, it did make me question if, when people buy a new phone, they tend to experiment more with new apps that might cause the installed base to be less significant.