Google updated its latest Android version distribution this week, which led to the usual dull accounts of slight shifts in version counts and lamentations over Android fragmentation. Undeniably, the Android market is fragmented, because, as an open source OS, it runs on smartphones, tablets, cameras, vending machines, thermostats, remote controls for electrical switches and outlets, and other devices not listed here or thought of yet. But this fragmentation does not prevent segmentation within a device category.
Fragmentation defines the problem as a software development problem surrounding cost, whereas segmentation defines the problem as a design and product management challenge involving return on investment. Identifying the smartphone devices used by target customers is the first task. For example, a general mobile travel site likely needs to run on most versions of Android, Gingerbread (Android 2.3x) through Jelly Bean (Android 4.1/4.2). A comparative case could be made for Nike’s rumored Android app that will be used with its $149 Nike Fuel Band, which will communicate quantitative measurements of the wearer’s physical activity to an Android smartphone. That would likely only require building the app for the most recent Android versions (4.1/4.2) and Ice Cream Sandwich (Android 4.0x) because these versions represent the more expensive smartphones. When developing any app in Brazil, where the conversion from featurephones to smartphones is just beginning and most smartphones are low-cost Android 2.3x phones, an app developer would waste its development budget building a 4.0x or 4.1/4.2 app specifically for the Brazilian market. Jelly Bean (4.1/4.2) can be thought of as a refinement of Ice Cream Sandwich (4.0x).
The total worldwide distribution of Android represented above does not take into account distribution within a given geography. One can’t use Google’s distribution statistics to make product decisions. There are many tools to help segment the device. One such tools is Mobile Detect, an open source module that can run on a website that will read the HTTP header from a mobile browser and detect the attributes of a smartphone used to browse the website. The manufacturer, model, operating system version, and language and other attributes can be identified, recorded and analyzed.
A hypothetical case is a desktop ecommerce merchant planning its first release of a mobile commerce (mcommerce) app. After installing a mobile detection capability on its ecommerce site and collecting the mobile device data for a sufficient period of time, the merchant would be able to rank Android vs. iOS, Android version vs. Android version, and tablets vs. smartphones. If the mobile app is built without a reference to a website, there are other methods of interacting with potential customers to determine the distribution of mobile device types. Search words could be purchased that channel mobile users to a landing page where these attributes can be collected. The secret to mobile app product management is not adding features in, but taking them out to focus on the app’s function and the user’s intent. The target device models and operating system versions helps do this.
For the product team that has analyzed the data and determined its target customers and decided it must build for Android versions 2.3x – 4.2, it will be more expensive, more difficult to test, and the final app will have a less sophisticated UI that required more work to build. Android 4.0x is a leap beyond 2.3x. It has a new style of tabs and other UI features and refinements that simplify development, such as fragments (unrelated to fragmentation) and better memory management for images. Developing for Android 2.3x and older versions means working around various issues and using third-party libraries, such as Actionbar Sherlock, that add complexity.
One would hope that Android 2.3x would recede more quickly given that its last release date was almost two years ago. But Android 2.3x is here to stay. It is tied to some very low-cost smartphone and tablet hardware platforms. In developed markets, Android 2.3x will decline, but for the near term, it will appear as a “free” smartphone with new carrier subscriptions and renewals or as an inexpensive prepaid smartphone carrier option, especially among the price-competitive tier-2 mobile operators such as Metro PCS, Cricket, Boost Mobile, Virgin Mobile, and Straight Talk. There are still 5 million mobile phone subscribers yet to move to a smartphone in the world, and many will choose an economic Android 2.3x, especially in developing economies.