The difficulty in analyzing the smartphone market lies in its extremely rapid growth. With the market growing at 90% (and 400% in some areas like China) the forces of demand and supply are disconnected. It’s impossible to discern whether a purchase decision is made from a choice of comparable alternatives or if it’s made from a choice between buying nothing and some alternative.
Furthermore, another problem lies in distribution inefficiencies. The global phone market is nominally serving 4.6 billion consumers but they are not all participating in the same market. Devices and the services they are bundled with are not fungible globally. You cannot buy a phone is one country and easily couple it with service in another. You can’t even transfer one bundle from one owner to another within the same country. Products are available only in some countries or some operators and not in others.
Commentators focusing on mobile OS platform shares assume a zero sum game and a liquid, efficient market. Both assumptions are false. They also assume that the platform is a basis for the choices made (rather than the device, the pricing illusion attached to the device, the service plan, availability, point of purchase and network switching costs, etc.)
There is data where competition for a single purchase decision by two comparable products with comparable specs and price plans can maybe show a preference. However that data is not public. Only an operator like AT&T who sells both Android and iPhone through the same channel has that data and they’re not sharing.
The closest I came to analyzing this was in a blog post last May. However, that data is obsolete. Since then Android growth has accelerated rapidly as vendors who were not making smartphones (Samsung, Asian second tier, Motorola and Sony Ericsson in that order) began making smartphones in volume. Their only choice in entering the market was Android. This supply-side view of the market is much more relevant than demand-side assumptions because the market is supply constrained.
Another way to see it is this: If there is a shortage of bananas, you can’t make a case for Costa Rican bananas being preferred over Honduran bananas. People will buy whatever bananas they can get (even if one may be preferred over the other). If Honduras can produce twice as many as Costa Rica then they will sell them all and it says nothing about their quality or end user preference or whether that volume advantage will be sustainable once the shortage abates.