I'm writing a CSS book

JavaScript archives Compatibility Thidrekssaga QuirksBlog Donations CSS book Politics Mobile About COH

contact

Platform smartness in the US

The release of US mobile usage statistics allows me to calculate the “smartness” of several platforms in the US market according to the method I proposed earlier. Better still, I can include non-smartphones in the calculation, which gives a more honest picture of the overall phone market.

iOS users turn out to surf about 4 times as much as the average phone owner, while BlackBerry and Android users surf about 2.5 times as much. The real surprises, however, come from Windows and Symbian.

Comscore released its calculations of April-to-June mobile phone usage in the US, and gives percentages for the five largest device vendors and the five largest smartphone platforms. Since Apple and RIM figure in both lists, we can calculate that 33.6% of Americans currently uses a smartphone, and from there we arrive at the overall market share of the other smartphone platforms.

StatCounter provides mobile OS surfing market share over the same period, allowing us to calculate a platform’s “smartness” by dividing browsing share by penetration share. I do not count the 0.6% PlayStation browsing share, since this device does not occur in the Comscore stats due to it not being a phone.

Surprising results. Not iOS but Symbian appears to be the smartest platform around, while Windows is basically a non-smartphone platform.

Windows

I’m at a loss to explain the lousy Windows result.

• Do Windows users hate IE so much that they won’t browse? (Users tend to use whatever comes to hand, so I don’t believe it.)
• Do they use Opera, which does not always pass platform information on to StatCounter? (Unlikely; they’d use Opera Mobile, which is pretty good in providing such information.)
• Or are people who buy Windows phones less inclined naturally to browse? (If so, why?)
• Or does StatCounter misdetect Windows phones as Windows desktops? (Possible, I suppose, though I haven’t yet found such a mistake in my tests of StatCounter.)
• Or is the Comcast sample skewed in favour of Windows phones? (Possible but unprovable.)

I just plain don’t know.

Symbian and S40

I’m also at a loss to explain the excellent Symbian result. I see three options:

1. The Comscore sample of 36,000 Americans is somehow skewed against Symbian users. Could be, but we’ll never be able to prove it.
2. The StatCounter Symbian numbers are incorrect. Now I know for a fact that currently S40 is detected as Symbian, so this could definitely be the case.
3. My calculation is correct and Symbian users surf about 30% more than iOS users.

If we want to preserve our world order and keep iOS at the top of the smartness chain, we must choose option 2 and decide a significant amount of “Symbian” browsing in fact comes from S40 and should be compared to the Other OS category. Thus Symbian’s smartness would fall, and the Others’ would rise.

So let’s say that half of the “Symbian” browsing share comes from S40. That would bring Symbian’s smartness down to Android/BlackBerry levels: still pretty good, but at least lower than iOS’s.

How many S40 phones are used in the US? No study will ever answer that question because S40 is not hip and happening. Still, we can make an estimate. The last time Tomi Ahonen gave general phone sales numbers, for 2009, Nokia sold 432 million phones of which 68 million were Symbian. Assuming that all other phones are S40 we arrive at a 1:5.35 ratio. That would mean 4.3% of all phones used in the US are S40.

Let’s re-calculate:

Now S40 users surf more than Windows users! It’s almost enough to start to doubt the entire methodology.

So let’s wait a while, hope StatCounter implements my suggestion to separate Symbian and S40, and see what happens next.

One thing is certain: these surprising US numbers, combined with my earlier research, where Symbian (+ S40) ended a good second, mean that Nokia’s platforms aren’t as dumb as people seem to think.

Unless you disagree with this entire method of calculation, of course.

Non-smartphones and app downloads

The Comcast research contains two more factoids that should be studied. In the April-June period, 40.1% of phone users used their browser, while 39.5% used a downloaded app.

The point here is that only 33.6% of Americans uses a smartphone at all. So 6.5% of Americans browsed while not being on a smartphone, and 5.9% downloaded apps while not being on a smartphone.

Do all smartphone users browse and download apps? I’m not certain they do. I’m especially thinking of youth users of BlackBerry, who don’t care about browsers or apps, but only about BBM. The non-smartphone numbers may be even larger.

So let’s change our conclusion: at least 6.5% of Americans browsed while not being on a smartphone, and at least 5.9% downloaded apps while not being on a smartphone.

In other words, at least 9.7% of non-smartphone users browsed, and at least 8.8% downloaded apps. Now browsing is not a function that’s necessarily restricted to smartphones, but downloading apps is. After all, the very definition of a smartphone is the fact that you can install apps on it (and that it runs on a recognizable OS).

This means that the definition of smartphones in studies such as Comcast’s should be extended. Right now there’s an arbitrary fixed set of platforms that is deemed a smartphone, and everything else isn’t.

That state of affairs should change. Several former feature phone platforms have already made the jump to smartphone and should be counted as such. I’m especially thinking of S40, but the new Brew MP may also be a candidate. And we have to go through the countless Java ME platforms and figure out which of them are still important.

We need better stats; stats that pay attention to platforms deemed non-smartphone, as well as to the long smartphone tail. With the current crop we’re never going to figure out how the mobile market works.