I can think of a number of ways to interpret this data that don't support your conclusions: 1) Online device utilization is a lagging indicator, since the traf

I can think of a number of ways to interpret this data that don't support your conclusions: 1) Online device utilization is a lagging indicator, since the traffic data is representative of the smartphone market share numbers from about 2 years ago. This is consistent with observations that smartphone device use increases over the ownership period. 2) iOS users are engaging in more casual discovery on their devices, and as such hitting more of the reporting sites in these statistics. 3) iOS better serves data-hungry users, either through device selection, plans, legacy, or some other factor. 4) iOS is worse at managing data use, on-device cache, etc. 5) The "unique methodology for collecting this data" employed by NetMarketShare, which is without detailed explanation or put to scientific rigor could be misrepresenting this data. Any one of these could explain or contribute to a considerable amount of inaccurate data. A combination of them could exacerbate these effects. So unless you're arguing that people are buying Android devices and throwing them in the trash, the number of bits pulled down by devices isn't a terribly useful indicator of market penetration, just user behavior. If you want to know what platform to target for a particular service (assuming you can only target one), mass-analysis is pretty lousy. Get to know your target market and what devices they use. Unfortunately, this looks like the kind of "analysis" that starts with a belief and sets out to find data to support it.