According to Gartner, Android sold 144,720,300 units in the fourth quarter of 2012. But let me ask you this:
Does Samsung care how many “Android” units were sold? No, they do not. They only care about how many of their devices they sold.
Do the various Android manufacturers in China care how many “Android” units were sold? No, they do not. They only care about how many of their respective devices they sold.
Does Amazon care how many “Android” units were sold? No, they do not. They only care about how many Amazon devices are being used to direct traffic to their web site.
Do Android developers care how many “Android” units were sold? No, they do not. They only care about those Android units that their software can address and, even more specifically, they only care about that portion of the addressable market that is interested in purchasing their product or downloading their product and consuming their advertising.
Does the Google Play store care how many “Android” units were sold? No, they do not. They only care about how much is purchased from the store.
Why Do We Count Android As A Single Entity?
“Android” is not a single entity. So why do we add all of the “Android” numbers together? We do it because we assume that higher numbers mean a stronger platform. We use it as a proxy for the strength of the platform. But it just ain’t so. Total numbers mean nothing. The only numbers that matter are those that strengthen the platform.
And do you know who agrees with me? Google.
Android’s New Math
Google reported that the number of Android units using Android versions 4.1 to 4.2 jumped from 16 percent a month ago to 25 percent this month. Impressive, no?
The reason for the big jump was that Google changed the way they count the numbers. Previously, devices were counted when they checked in to Google’s servers. But Google is now only counting user vistits to the Google Play Store. Google argues that the data more accurately reflects users “who are most engaged in the Android and Google Play ecosystem.”
I agree. This is a better way to count the meaningful numbers rather than just the gross number of Android activations. However, did you notice the inconsistency in Google’s new math?
One Of These Is Not Like The Other
Google hasn’t recalculated and lowered the total number of Android activations.
In other words, when it comes to telling you how many activations they have, Google uses devices that check in to their servers. But when Google wants to tell you which versions of their operating system are in use, they only count user visit’s to the Google Play store.
Hmm. So we now know that versions of the Android pie are divided into different sized slices but what we don’t know is just how big that pie is. Exactly how many of the 144,720,300 units sold in the fourth quarter of 2012 are actually accessing the Google Play store?
We don’t know. Because Google isn’t saying. And until they do, those total unit sales and activation numbers have little meaning in determining the overall strength of Google’s portion of the Android platform.
Android’s Total Numbers Conceal Rather Than Reveal
“Android” shouldn’t be counted as a single operating system any more than Europe should be counted as a single country. Heck, Android doesn’t even have a “common market“.
If we’re going to use numbers as a proxy for determining the strength of various operating systems, then we have to use meaningful numbers. Perhaps we should be comparing the units running the latest version of iOS with the latest version of Android. Perhaps we should be counting the Amazon, Google, and the various Chinese portions of Android as distinct and separate entities. Perhaps we should even be counting that portion of the Android phones that run Facebook Home separately too.
What we most certainly should NOT be doing is lumping all Android sales and activations together and pretending that they’re one and the same and that their total numbers are advantageous to all of Android’s separate participants, such as Samsung, HTC, Amazon, Google, developers, etc. If an activation or a unit sale doesn’t count towards the strength of the whole operating system, then it shouldn’t be totaled. Totaling Android’s numbers together doesn’t make sense because there isn’t a single, unified Android platform.
Numbers should be used to reveal, not conceal. And Android’s numbers aren’t revealing its strengths, they’re concealing its weakness.
26 thoughts on “Google’s New Android Math Doesn’t Add Up”
Seems obvious, doesn’t it? Especially these days with so many forks of Android being the norm rather than the exception. Wonder when the press as well as the analytics will catch on? [edit to add: as well as you have.]
Given past history, they won’t catch on until after it’s already been determined that the real metric has already changed to something else,
Clear. Bold. Brilliant!
Thanks for this one. This column deserves to go viral.
Android dumbination detailed.
Kirk, your clarity of mind is enlightening. The trick is to bring the obvious out of the murky twilight and into the light of day. You do this so dang well.
stevesup, ‘dumbination’; I like it.
Amen. Excellent elucidation. It is National Alliteration Day today, isn’t it?
I see your point but this seems a little harsh. They just made the change to the OS version share chart a few days ago — and the first since Andy Rubin left — and they haven’t said anything about total activations or anything else yet since early March. I’d bet that the next time Sundar Pichai talks about this, he’ll either be using the new definition or explaining the difference.
Also, Google has previously explained that the activations total does not include devices that don’t run Google services, such as the various Chinese variants or the Amazon Kindle.
He’s not just talking about Google. It is very common for analysts of all kinds to magically pretend all Android-based devices are one platform.
If they’d wanted to, Google could have cleared this up when they changed the way they reported Google Play numbers, or just reported actual Google Play numbers. Clearly this is the way they intend the information be presented.
Fine insight and a fine post. To be even finer, let me note,
Until you specify what “strength” of an operating system really means, I don’t think you’ll have a very good way of choosing between the different measures.
I face this all the time in my econometrics work… should I use housing starts, the average price of homes or the unsold inventory to best assess the “strength” of the housing industry? And in each case, the best measure depends on what you’re trying to get at (plus, something that’s directly relevant here, availability, reliability, consistency and timeliness of the data).
AFAICT, the current question of Android strength is all about bragging rights, and for that purpose, the stats that are most important are the ones that show Android with the fastest growth rate or the biggest share difference between Android and iOS. Heck, to the extent that that’s the name of the game, why bother with stats at all? Why not just use individuals’ sense of entitlement or testosterone?
Entitlememt and testosterone are key features of tech analysts and the tech press. It always devolves into a game of measure-the-penis. The Mac has had 90% of PC sales above $999 for about 5 years, driving Windows systems down to a $400 average price point and crippling the profitability of Windows PC makers, yet that story was told in the tech press as “Windows has 90% market share” and “the rise of the netbook.”
Even with this conglomerated definition of “Android”, the most recent numbers show that the iPhone’s sales were 4 out of every 10 smartphones sold in the USA last quarter! The other 6 out of 10 includes ALL of the Android phones sold by ALL of the other companies, plus ALL of the Windows phones sold by ALL of the other companies, plus ALL of the Blackberry phones, plus all other smartphones!
But worse than that for Android (and for other smartphone operating systems) is that Apple is taking close to 75% of the profits from the sales of all smartphones sold worldwide!
And all of this is just about smartphones. When you start comparing the iPad sales and profits to those of Android tablets, it’s downright embarrassing for the Android faction.
exactly, JK. “Android” with a capital A is a myth. there isn’t even a real android family anymore.
“ecosystems” under one company’s control are the real platforms that matter now. Google does have one, but it’s only a chunk of the big android pie. Apple’s famous “walled garden” is of course the most fully realized, and hence the revenue leader by far. but it’s certaily not complete. MS is trying desperately to set up its own. Amazon has carved out its own chunk of Android too, and Samsung is headed in that direction. while Face Book has chosen to simply and brazenly steal Google’s outright!
Counting all Android as one platform is like counting all WebKit as one browser. The browsers are Apple Safari, Google Chrome, BlackBerry Browser, and so on. WebKit is just the open source project they are all based on. Yet you can’t trust that the same Web app runs in all WebKit-based browsers, same as you can’t trust that one Java app runs on all Android-based devices.
The reason Android is misrepresented is that it is magically imbued with the properties of either iOS or Windows, depending on whichever makes it look better in any particular situation. Android is nothing like either iOS or Windows. Android is a 2005-style Java phone stack with an iOS-like touch interface added in 2008. Manufacturers use Android to build their own stack, customizing the interface and features dramatically — nothing like Windows. The best analogy is Linux, which manufacturers customize to build their own server stacks.
“Counting all Android as one platform is like counting all WebKit as one browser.” – JohnDoey
I love that analogy and I will quite probably be “appropriating” it for future use. 🙂
Oh wait, they don’t? Wait, you’re telling me that browsers are about a million times closer to being the same thing REGARDLESS of the layout engine they use, because they are actually standards-based? You mean the analogy doesn’t work at all, and you didn’t even notice, and instead claim that you’re going to use a broken analogy in the future? Huh. How about that.
Hell, developers have to test on every browser, nobody is denying that. But counting all Android as one platform is more like counting all BROWSERS as one platform… WHICH PEOPLE DO! See, all the platforms use the same language, and things work basically the same on all of them. You have to tweak here and there, but you’ve basically written the same thing for all of them… Just like Android. How about that! An analogy that works! Don’t worry, you can “appropriate” this one; I don’t mind.
I also like how you take offense at Google using “the number of activations” to count “the number of activations”, but using “the number of people who buy apps on the Google Play Store” to count “the number of people who are using a particular Android version and will buy your app if you write it for Android 4.0+”. Gosh, it’s almost like the different data is being used for different reasons. It’s almost as if Google is communicating the relevant data — number of activations for people who want to know the number of activations, and the number of people on an OS version who get apps from the Play Store for the devs who need to choose an OS API for people who are buying their app on the Play Store! But no, that can’t be right… Google must be fudging the numbers because they are really so scared of people finding out that Android really has a 2% market share! IT’S A CONSPIRACY!! IT’S ALL A LIE!!! REMEMBER TO TWEET MY ARTICLE!!!!
I think this is overstating the fragmentation issue by quite a bit. I use an Android phone running gingerbread and I have yet to run into an app that will not run on my phone. So, while individual hardware vendors don’t care about the total Android market, developers certainly do.
Well, sort of. Developers are more interested in making money. So far that is happening more on iOS than Android with a few exceptions. All you have to do is look at the number of iOS first or only apps vs Android. I can only think of one app that considers itself Android first, and I can’t even remember what they do. If that total Android market were actually translating to dollars, I’m sure the developers would care more.
Why only Gingerbread? Can you not upgrade to at least 3.x?
“when it comes to telling you how many activations they have, Google uses devices that check in to their servers. But when Google wants to tell you which versions of their operating system are in use, they only count user visit’s to the Google Play store.”
Good catch. As long as financial analysts, tech pundits, and other talking heads give companies like Google, Amazon, Samsung and others the benefit of the doubt, these companies will attempt to game the system by releasing as little hard data as possible.
It will take a skeptical eye and a more negative assessment of these pronouncements before their approach will change. It used to be analysts would punish companies for being so opaque and cryptic, now they’re pretty much paid lapdogs.
It seems like this is a religious war. Is this Christianity vs. Islam or is it Baptist vs. Islam? In that vein, why can’t we simply look at this as Apple vs. non-Apple? In that light the trend is more telling no matter how you skew the numbers.
Considering the average Android device has 2 major operating system updates in it’s life, then yes… they should be added together. Just as the same Apple devices have gone from iOS 4, 5, 6, etc. I refuse to accept the basic premise of this article. When everyone does the same thing, that is the standard. I don’t see Apple separated, so why should Android?
When Apple releases a new OS version, the great majority of iOS devices very quickly move to it. The only notable exception of a widely used device not on the latest OS is the original iPad, which cannot upgrade to iOS 6.x. But this is very much an exception in the iOS world.
“I refuse to accept the basic premise of this article. When everyone does the same thing, that is the standard. I don’t see Apple separated, so why should Android? – Denny Craneftw
60% of iOS users were on iOS 6.0 within weeks of its introduction. If you don’t see the difference, it’s because you’re refusing to look.
Just picking up on your point about Android being likened to Europe, you might also say that the iPhone is like America. Android & Europe benefit from a rich diversity of cultures whereas America benefits from a common currency, language and economy. Both have their benefits and suit different people, but nobody tries to say that Europe is better than America because we have more people or make more money so why are we bothering with measuring iPhones against Androids? Leave that to the Apple/Google marketing people, just choose the phone you like on the platform that suits you and don’t worry which is better for everyone else.
It’s a mess out there for Android phone users who want to stay current with the OS, apps and computer-phone connectivity. T-Mobile and Samsung can’t get their act together for the GB to ICS upgrade for the Galaxy Blaze, Google hasn’t made a version of Chrome for GB and Samsung’s KIES is a piece of crap.