Last week on Twitter I shared some browser and OS statistics from a site I manage. These turned out to be quite popular, so I’ve decided to expand on them a little further, and also add the stats from another site I manage, to broaden the base numbers a little. I’m not trying to make any point here, just sharing a little bit of analytics data. If there’s any interest in my doing so, I’ll provide further updates in the future; leave a comment if there’s anything in particular you’d like to know.
Stats are taken from Google Analytics from the period 1st January 2013 to 26th April 2013 (it took me a few days to write this post). I’m going to give the mean average of stats between the two sites, noting where there is significant variation — and I’ll speculate on the variation at the end of the post.
I decided not to use the stats from this blog as it tends to skew towards a very non‐typical audience, consisting in large part of the tech‐savvy and early adopters. Instead I’m using two sites I manage which I think have a more ‘normal’ set of visitors (no offence intended, dear reader).
To keep the sites anonymous I’ll refer to them as Site A and Site B. Site A is for news and opinions on Arsenal football club. It’s London‐centric but has an international audience, and had an average of 1,935 unique visitors per day in the selected period. Site B is a hyper‐local blog for a London neighbourhood, and its audience reflects that. It had an average of 120 unique visitors per day. Both sites are updated semi‐frequently but irregularly.
The percentage split between mobile (including tablet) and non‐mobile is 31.2 to 68.9. Of those mobile visitors, iOS accounted for 68.4 percent and Android for 23.3. Blackberry was a distant third, with other mobile OSes barely registering.
In overall OS stats, Windows is still by far the dominant platform, with more than double the visits of the second‐placed iOS. Some strong variation to note here: Site A had 12.2 percent of visitors using OS X, with Site B on 27.1.
|Operating system||% of total|
In overall browser stats, Safari is the perhaps surprising leader, thanks to the large share that iOS holds. Some very strong variance to note: Site A actually had Opera Mini in fourth place with 7 percent share, beating Android into sixth, while on Site B there were no hits at all from Opera Mini.
|Browser||% of total|
Breaking Internet Explorer into the different versions, IE9 is the most popular followed closely by IE8. Very gratifyingly, IE6 is barely existent and soon to be gone altogether. There was a point of strong variance here: Site A had IE8 on 42.1 percent of the IE share and IE9 on 41.6, while Site B had the same versions on 35.3 and 50.3 percent respectively.
|IE version||% of IE share||% of all browsers|
Finally, in my original tweets I noted that over 98 percent of visitors running iOS devices used their phone in portrait mode, but this is incorrect; in fact Safari for iOS only shows dimensions for the portait mode, even when in landscape, so there’s no way to get this figure. Android devices don’t follow the same pattern, however, so there we can see that an average of 92 percent of visits were made using a device in portrait mode. I can’t be sure where the false iOS figures came from; perhaps other browsers on iOS, or jailbroken devices.
Speculation on variance
There are a number of points of strong variation: in visitors using OS X; in the use of Opera Mini; and in IE8 vs IE9. My broad theory to explain this is that Site A has a global audience, including a not insignificant number in emerging markets across Africa and Asia, whereas Site B is almost exclusively visited by (comparatively) wealthy Londoners, which explains the higher number of Macs in Site B, and the use of Opera Mini on Site A (from areas with poorer network coverage and a higher incidence of low‐end handsets). As for IE, I’d speculate that you need a genuine copy of Windows 7 to run it, and I believe that Site A’s global popularity would mean that it’s used on pirated Windows which couldn’t install it.
Of course, this is all speculation, and with the limited sample base I have, especially on Site B, it could just be an aberration in the data. As always, it pays to use many different data sets.