Browser Stats For The BBC Homepage 260
Lord_Scrumptious writes "An interesting article titled 'The software used to access the BBC homepage' has recently been published on a blog by a BBC employee. It's all about the different browsers and operating systems accessing the BBC's homepage. The analysis is from a week of page requests in September 2005. Not surprisingly, Internet Explorer accounted for 85% of site visits, but Firefox had a very respectable 9.7% share. Even requests from Sony's handheld PSP device were recorded, but interestingly there's no mention of mobile phone devices."
Mobile devices (Score:5, Informative)
Many people go to BBC, CNN and other major sites through their mobile service provider's front pages. These would naturally point to the dedicated mobile editions too.
Representative of Overall Market Share (Score:5, Informative)
The BBC's numbers are simply representative of this, as any large web site would be.
mobile devices (Score:4, Informative)
Re:Opera (Score:2, Informative)
Mirror (Score:4, Informative)
How about Slashdot generating a mirror link via a neat little "mirror" icon next to the links?
Re:Mirror (Score:4, Informative)
Only nyud.net links [nyud.net] may help then, although my experiences with those aren't the best and why I tried to avoid it in the first case.
Re:Opera (Score:5, Informative)
Yes, they are.
Old versions of Opera that identify themselves as IE by default use a user agent string like this:
So the "Opera" string is here and easily identifiable.
New versions should simply use the proper Opera UA string by default [slashdot.org].
If you use Opera I suggest to check that it sends the "correct" Opera UA string: the sky will (mostly) not fall down.
All I could get of the article (page 1 and 2) (Score:5, Informative)
Our old stats reporting tool at the BBC gives a breakdown of requests from different user agent strings, which is where the browsers and operating systems people use to navigate around the web leave their digital fingerprints. It is about to be phased out in favour of a new solution, but I'm not sure that the new system gives the same granularity of data, so once I'd started, I thought I'd look at the figures in some detail before the old system gives up the ghost.
Now if you've never looked at user agent strings, they are rather dull and geeky, and full of lots of technical gubbins like these examples:
* Mozilla/5.0 (Windows; U; Windows NT 5.2; en-GB; rv:1.7.10) Gecko/20050717 Firefox/1.0.6
* Mozilla/5.0 (Macintosh; U; PPC Mac OS X; en-us) AppleWebKit/85.7 (KHTML, like Gecko) Safari/85.5
* Mozilla/4.0 (compatible; MSIE 6.0; America Online Browser 1.1; Windows NT 5.1; SV1;
* Mozilla/4.0 NETIKUS.NET GetHttp v1.0
* Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; Hotbar 4.5.1.0)
* Mozilla/4.0 (compatible; MSIE 6.0; Windows CE)
* Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.5) Gecko/20031007 Firebird/0.7
* Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.0; T312461; BT [build 60A])
There are of course some caveats around the figures I'm about to talk about.
User agent strings aren't an exact science. Or rather, they ought to be, but in the real world the come out a right mess. I've done my best to untangle them, but I still ended up with a significant number of user agents that I could not identify properly. And that is before we get started on the corporate networks that use the UA string to broadcast their corporate branding to the world whilst masking their operating system. Or requests claiming to come from both Internet Explorer 6 and Internet Explorer 5.5. Or that claim to be from a particular Linux distribution and Windows 98 at the same time. Or the plain weird like the inadvisably named KummClient from Hungary that proudly proclaims 'Linux rulez' to anyone like me dull enough to be delving through their logfiles.
User agent statistics on something as big as the BBC homepage could almost be the very definition of the long tail. The most popular user agent string - IE6 on Windows XP - clocked up nearly 6 million requests. I only counted user agents that had made more than 50 requests, but between 6 million and 50 requests there were nearly 11,000 different user agents to look at. Examining that number of requests accounted for 95% of the reported traffic, but only around 1/3 of the stats report. I initially suspected that counting the whole of the tail was likely to increase the market share I derived for the quirkier set-ups, but a random sample showed that a large proportion of the tail consisted of the most popular browsers and operating systems, but with different installed toolbars or corporate network messages that distinguished them as a unique string.
And I must stress again, these figures don't represent the breakdown of visitors to the BBC site as a whole, they are based on requests to the homepage alone, over the course of one week in September. Nevertheless I think they provide an interesting snapshot of web activity.
In total I've examined around 32 million requests to the BBC servers - although some of these have been discounted as 'unknowns' and some originate from crawlers and spiders.
The complete dominance of Windows XP and Internet Explor
As always, defaults play a role (Score:3, Informative)
Thecounter.com has stats as well. (Score:1, Informative)
http://www.thecounter.com/stats/2005/Sep
1. MSIE 6.x 62223734 (83%)
2. FireFox 5806423 (8%)
3. MSIE 5.x 3170911 (4%)
4. Safari 1311540 (2%)
...
Here's a link to their OS stats for the same month:
http://www.thecounter.com/stats/2005/Se
1. Windows XP 54945908 (73%)
2. Win 2000 8468339 (11%)
3. Win 98 6806316 (9%)
4. Mac 2317188 (3%)
5. Unknown 1132090 (1%)
6. Win NT 473897 (0%)
7. Linux 322362 (0%)
...
Probably more accurate since they count more hits on a wide variety of different types of web sites.
Re:errr (Score:1, Informative)
If it wasn't for that, I'm sure that the stats would be even better.
... and this is page 3 (Score:4, Informative)
Windows Operating System Share
Concentrating on just Windows alone we can see that Microsoft have done a very thorough job of converting their user base to the most recent iteration of the software. Windows XP accounts for just under 70.5% of the Windows requests, and Windows 2000 a further 17.4%. That means in total around 88% of users of Microsoft Operating System products are using the two most recent consumer releases.
Windows 98 features in 7% of requests made from a computer running a version of Windows, and after that the figures are very small in terms of market share. In fact the next largest figures is a clump of 'Windows other' including Windows CE, and various unspecific Windows NT user-agents that I couldn't pin down to a precise version.
Mac Operating System Share
I was frustrated in my attempts to similarly breakdown the different versions of the Mac OS that people were using to request the BBC homepage. I established that from the requests we saw I could identify Panther as supplying 31%, Tiger supplying 21%, with Jaguar lagging behind at 3%. However there were 41% of requests where I could identify that the computer was a Mac, but not the specific version. That is because Safari helpfully supplies in the user agent string the WebKit build, allowing the precise version of the OS to be identified, but most other browsers do not.
Linux Requests To The BBC Homepage
The number of Linux requests to the BBC homepage was very small, representing only 0.41% - less than 100,000 - of the 32 million requests included in this study. With such a comparatively low number I didn't take the time to delve into which different distributions were driving the requests.
The figures may, however, mask a slightly higher use of Linux. Since the user agents generated are more likely to be unique, they are more likely to have fallen into the statistical long tail. However I should add that my random samples of the tail did not show that it consisted entirely of Linux, in fact as I mentioned earlier, a lot of corporate-branded Windows networks show up in the tail.
Legacy OS Systems
We have some fairly strict standards for supporting legacy technology at the BBC on the client-side - but the long tail of older OS software visiting the BBC homepage is amazing. We still saw over 300 requests for the BBC homepage coming from machines claiming to be running Windows 3.1, and around 200 requests from machines claiming to be persevering with 0S/2 Warp.
Re:No MSI build for Firefox - no mass deployment (Score:5, Informative)
http://www.frontmotion.com/Firefox/ [frontmotion.com]
MSI installers for Mozilla Firefox! Useful for installing Firefox on a single computer for the home user or deploy across thousands of computers automatically with Microsoft's Active Directory. Use Firefox on your corporate computers to decrease virus incidents and increase overall security. Save time and frustration with our installer that is targeted toward the corporate IT administrator with manageability and upgradeability in mind. This is not just a wrapper around the exe installer nor is it another half baked 'captured' install.
Re:Finally.... (Score:5, Informative)
However, given the headers returned by the BBC site, caches should NOT cache the HTML, as the headers say the content expires immediately:
Expires: Mon, 24 Oct 2005 11:57:59 GMT
Date: Mon, 24 Oct 2005 11:57:59 GMT
Content-Type: text/html
Server: Zeus/4.2
Cache-Control: max-age=0
So, the BBC figures may be more accurate than you think.
Re:Finally.... (Score:1, Informative)
Re:Finally.... (Score:2, Informative)
older stats (Score:2, Informative)
Then it got to about 95% IE, so 85% is quick a marked drop in IE support.
Do remember that the BBC is hardly a generic site for your average Internet user, it attracts a significant quantity of beginners and is dull for anyone technical (there are a higher proportion of technical users on the Internet than you'd meet on a street). So these stats are quite good.
I know the way they are worked out should be quite fair.
Re:Representative of Overall Market Share (Score:2, Informative)
I assume after it went free, flood of newbie user got the browser and they didn't change ID String yet. I can't blame anyone, you can still live problems with identifying as Opera to some sites. While I was using Windows, I was identifying as Mozilla browser.
Its kind of stupid to identify browser features with their names in fact. What they (big sites) need is to check capabilities of the browser, not its name.
Apple got a very good article on that:
http://developer.apple.com/internet/javascript/ob
Re:User Agent Switcher (Score:3, Informative)
The googlebot UA string currently is (remove the space in "ht tp"):
or:
Yahoo:
MSN: