Fragmentation and segmentation

Android fragmentation is an old story, relatively speaking. Hardware manufacturers take a long time to create implementations of new versions of Android for their handsets, and for many handsets that have already been sold they may not even bother. Google has made repeated attempts to do something about this, all of which have failed. The end result is that 6m after the release of the latest version of Android, just 7% of the base has it, while 80% of the iOS base is running on the latest version, which was released about the same time. This is then worsened by the widely varying device specifications and the variations in precisely how each OEM implements notionally identical versions of Android on different devices. Just about everything one can say about this has been said.

Android fragmentation is a serious problem for developers, since they face a massively fragmented user base, though it is rather less of a problem for Google, since a forked fragmented screwed-up Android phone still has a web browser and a data package and so is still generating Adsense eyeballs.

Indeed, a helpful way to think about this is that Apple makes money from great apps by selling the hardware to run them, while Google loses money from great apps, since they lead to less web search.

iPhone fragmentation looks rather different. 80% of the base is on iOS5, and there are only 3 models and 2 screen resolutions on the market. On the other hand, there is an emerging tendency for Apple features only to be available on the newest hardware:

  • The new ‘Flyover’ 3D maps in iOS6 are only available on the 4S and iPad 2 or later, while Siri is only on the 4S or iPad 3
  • Misc new features including VIP List for email, the Offline Reading List, and Shared Photo Stream won’t work on the iPhone 3GS

Hence Apple is announcing, and promoting, features that are not available on iPhone models that it is still selling. Some of this is a pure hardware issue - the 3GS probably does not have the horsepower to run 3D maps or Siri, but some of the other exclusions seem purely arbitrary.


Perhaps one possible way to look at this is to pair it with one of the prevailing Apple rumours: to make a cheaper model. The iPhone factory gate price is an average of around $650 while no other smartphone OEM averages more than $350 and Android is now pushing well under $200 at quality and $100 at the bottom end. Hence, people ask if Apple should make a new cheaper model to address a larger market.

Of course, Apple already sells the old 3GS model for $375 ( NOT free). But that’s actually still a pretty high-end price in the smartphone market. And the fact that Apple’s reported iPhone ASP hasn’t fallen since it started discounting older models suggests that they’re not selling very strongly.

The obvious problem with a cheap iPhone is how to make it clearly different (and ‘less good’) than the current iPhone without fragmenting the underlying iOS platform and forcing developers to write for two different phones. How do you segment without fragmenting? Apple achieved this pretty easily with the iPod by varying the storage, but that wouldn’t be meaningful for the iPhone. The cheap one has to run the apps, but people still have to have a reason to buy the expensive one.

What you CAN do is vary the Apple supplied features, without varying the hardware and API platform that your third-party developers are targeting. Suppose, for example, in the autumn Apple releases the iPhone 5 at the usual $600 entry price but at the same time sells the 4 at $300 - without Siri flyover, and the 4S at $450 (numbers chosen out of the air). All the apps work on all phones, the developers remain happy, but you have a nice clear segmentation proposition to allow you to sell low-to-mid-price devices without cannibalising the high-end model:

  • Cheap iPhone: no Siri or 3D maps
  • Mid-price iPhone: all the features, no sexy new case
  • Full-price iPhone - everything

(edit) - Apple already sells old models, of course. This segmentation analysis may simply be a helpful way of looking at why it does this, or might point to new devices. We’ll see.