When we launched Firefox for mobile on Android we were facing new challenges with regards to device compatibility. This was new for Mozilla, because on the desktop you do not need to test the whole range of hardware which ships with the platform you run on. Also our previous mobile efforts, for Maemo, included only 2 device models. But on Android, we had dozens of different phones. And testing all of them would not be an easy task.
So I wanted to see exactly what would be a decent amount of phones that we need to test in order to make sure that we cover a good amount of all Androids that are in people’s hands.
I looked at data of Android devices in market for the US, Western Europe (UK, France, Germany, Spain, Italy) and Japan and come to the conclusions that:
there is an average of 57 Android models per country
17 (35%) of them cover 75% of all of the phones in the market
It turns out that at one given time there are a few top devices which everybody has and then a long tail of others. The other are made up of not so successful models that will never rise to be popular ones and some which were recenly launched and haven’t had time to get owners. You can usually identify the ones in the long tail which are likely to become top devices by looking at how widely they are marketed, or success in other markets.