For platform providers it is important to not just understand your customers needs but also how they behave while on your platform. Most companies I come across know a fair bit about their customers but not much about how their customers behave while using their solutions. It is with this thinking that I want to dive into some observations on Android version fragmentation.
Those of you who follow us know that we cover this quite a bit. This subject, is in fact, very important to understand from an industry viewpoint, which is one we are paid to take. Understanding Android at a deeper level than just statistics is something that effects the industry and in particular those who will employ Android to help them sell hardware.
Google released some updated statistics this month showing the version percentages of customers on each flavor of Android. Here is the latest chart.
What the chart shows is that Jelly Bean surpassed Gingerbread to grab the highest percentage flavor being run in the market. Let’s first make some points about Gingerbread.
Gingerbread ver 2.3 was released on Dec 6 2010. Subsequent updates to Gingerbread ver 2.3.1 through 2.3.7 were released up until September 21st 2011. Devices running Gingerbread were shipped until approximately March of 2012 if not slightly longer. Based on Google released Android activation numbers and during the approximate time period Gingerbread devices were largely in market, we can conclude approximately 135 million devices were shipped worldwide running Gingerbread.
It is important to note that this time period represented the beginning of the aggressive Android growth on a worldwide level. Gingerbread devices also spanned all of 2011 and into the second quarter of 2012.
Some quick points on Honeycomb. This was largely developed as a tablet solution, to which we all recognize the failure of Android tablets at large. Gingerbread releases overlapped with Honeycomb and Ice Cream Sandwich came quickly. Most OEMs skipped shipping Honeycomb phones entirely. This explains the minuscule percentage of Honeycomb on devices.
Now let’s look at Ice Cream Sandwich. This flavor of Android ver 4.0 was released on October 19th 2011. Subsequent versions 4.0.1-4.0.4 were released through March 29th 2012. This version did not play into the 2011 holiday shopping period.
The market share boost of Ice Cream Sandwich and the reason it is still on 23.3 % of devices is largely thanks to the Samsung GS3 which represented approximately 110 million units worldwide during peak Ice Cream Sandwich time frame.
Next let’s look at Jelly Bean. Jelly Bean ver 4.1 was released on July 9th 2012 with updates 4.1.1-4.2.2 going through February 11 2013.
Here is a chart looking at Android activations from 2010 up until March of 2013. ((this chart’s data is using announced monthly activation reports from Google, paired with quarterly smartphone growth rates as reported by IDC, Gartner, and our own estimates from Creative Strategies, Inc.))
In understanding platform fragmentation for Android, it is important to look at the degree in which upgrades are happening from legacy devices to current OS updates and developer APIs. My conviction is that the overwhelming majority of devices are not being upgraded frequently by the end user and thus the platform share from Gingerbread, Ice Cream Sandwich, and Jelly Bean have been acquired largely from organic device sales rather than end user OS upgrades.
If most cell phone device upgrades are happening on a two year average cycle in subsidy markets, then the big growth we saw with Gingerbread in 2010 would have translated to 2012, which would explain why these two platforms make up the majority share.
The primary difference that I am pointing out is that with iOS, Apple has a track record of getting a pull of demand to upgrade for their consumers. Where with Android, the evidence is that the upgrades are not a pull from consumer but more of a push from Google, carriers, and OEMs. This is highly informative about Android customers, and more evidence to suggest that the bulk of Android device sales are to the low-end.
This all matters to developers. If the majority of installed base is not upgrading in a timely fashion, but rather largely getting upgrades when they buy new devices, then there is little reason to invest time and effort into the new APIs launched with each new OS upgrade.
Interestingly, Google is trying to thwart this with the release of their new Play framework which is initially targeted at games. Google’s Play framework tied to the new Android studio development software release appears to be more heavily targeted at fragmented OS development. As developers tell us, developing Android still takes more time and resources to develop for than iOS and Google is taking small steps to address this.
I also need to make mention about China in the story of Android’s growth. There has been much speculation about whether or not non-Play activated devices in China are factoring into Google’s activation numbers. Google’s has recently cleared this up with recent stats by saying they are only for devices that have Google services installed. That is good to know about activations from here on out but what about the last few years? Based on the numbers, I have a hard time believing Google’s activation numbers up to this point did not include non-Play devices in China.
The industry sold 700 million smartphones in 2012. Of the Android device sales in that year, approximately 400 million, 33% were in China. Looking again at my previous chart plotting Android growth, there is no way we can get to that reported number without including China. This would also explain the spike in my chart toward the second half of 2012 when we know in particular Android took off like a rocket in China. To further look at Android device sales to date, here is chart showing overall worldwide device sales to date by year. According the pace shown we may very well hit 800 million Android device shipments in 2013, mostly in the low-end.
Despite the growth demonstrated in these charts, I am bearish on Android. I will remain so until I see a different picture of the install base’s behavior emerge. I am don’t believe that Android goes away entirely. Rather, I am bearish on its value to the ecosystem. I remain convinced that Motorola will become Google’s Android solution. There is no doubt in my mind that Google has taken full control of their hardware destiny with Motorola.
Time will tell if we will see a more valuable set of habits emerge from the general Android installed base. Until that time, I remain convinced that although large, there remains a great deal of ecosystem unhealthiness related to Android. Device OEMs and developers need to tread water lightly and strategically in the Android sea of sameness.