Home > Statistics > Browser market share statistics skewed, according to Opera

Browser market share statistics skewed, according to Opera

January 9, 2006

The recent browser market share statistics released by NetApplications showed Firefox and Safari making gains, while Opera’s share stayed virtually untouched.

We all anticipated Opera’s market share to increase, even slightly, after it decided to drop the ads last September. Opera has doubled its average daily download rate, since going ad-free, so why hasn’t it been reflected in the statistics?

According to Opera these statistics are skewed. In an email to Opera Watch Opera spokesman Thomas Ford explained that it’s tough to get an accurate and precise picture of the browser market share since most firms that measure it, skew to U.S. sites. “Our traditional geographic stronghold – Europe – may not always reflect fully. Germany, Poland, Russia have been very strong markets for us.”

There are in fact multiple explanations for Opera’s poor showing in these statistics.

Firefox’s link prefetching and Opera’s enhanced caching mechanism is perhaps the most obvious. In fact Opera’s CEO, Jon von Tetzchner, believes that Firefox’s figures are inflated due to their link prefetching. “Sadly the statistics are undercounting Opera and over counting Firefox. Opera has a better caching mechanism so it doesn’t access Web sites as often as other browsers. Firefox has added a pre-loading feature that Google has made use of. This inflates the numbers on the statistics,” von Tetzchner said.

Another reason why the statistics are skewed is that different sites attract different audiences. Some web sites attract advanced users with technical skills, which more often use secure Browsers; other sites attract hobbyists using older low spec browsers. Firefox has made some great advances with non-technical-savvy users as well, while Opera’s main clientele remains mostly with the technical-savvy. Opera has much work to do in this area.

Not properly counting Opera in these statistics could also be due to the fact that Opera, by default, identifies itself as Internet Explorer (IE) with its User Agent (UA). There are some sites that specifically lookout for the Opera browser, and supply it with broken HTML code for the webpage. To counter this, Opera identifies itself as IE. Opera 9 will by default identify itself as Opera.

Advertisements
Categories: Statistics
  1. January 9, 2006 at 11:48 pm

    Blah blah blah… Again? Same arguments repeated over and over. Anything new?

  2. January 10, 2006 at 12:34 am

    Facts dont change every fortnight.So naturally they would remain the same.The reasons put forward by daneil are quite valid indeed.

  3. January 10, 2006 at 2:02 am

    The article is very legitimate and informative in light of what would be expected gains in Opera usage as reflected in the statistics.

    I still wonder, given the reasons are the same as always for Opera’s underreported usage, shouldn’t an increase in usage statistics be expected somewhat still if the new free Opera is actually being used more than was the ad version of Opera?

    Opera is just awesome software, regardless. It’s still unbelievable to me that it is not more widely acclaimed.

  4. Anonymous
    January 10, 2006 at 3:47 am

    Regardless of the specific issues with any particluar methodology, if one follows a single study over a period of time and that study doesn’t change its methodology, then one would expect the _trend_ to represent some _real_ change.

    That Opera’s numbers on this particular study show a gain of only .03% (not 3%, 0.03% — three one hundredths of one percent) in total market share during 2005, and that Opera actually ended the year down 0.07% from its high in August, suggests strongly that Opera did not make any significant (or even insignificant) improvements in browser market share during the last year.

    I’d also like to shoot down, right now, any bullsh*t about Opera’s cache having a negative impact on its maketshare. No web analytics firm worth its salt would care whether or not a browser does a good job of caching visited content. These stats packages track “users” not “hits” so it doesn’t matter if an Opera user hits the page 1 time and some other browser user hits the page 10 times. It’s not hits they’re tracking.

    Also, on Firefox prefetching, Firefox sends along a header when that happens and any stats package worth its salt can easily identify the few users that come from prefetching and exclude them. None of this is the rocket science some people are making it out to be.

    With the millions of downloads that happened after the release of Opera 8 and the millions more in the month after Opera ceased being adware, shouldn’t there have been a significant impact on the market share trendline?

    The gains that this blog predicted from the vastly improved Opera 8 never materialized. The gains predicted when Opera stopped being adware never materialized. Meanwhile, Safari and Firefox are making consistent gains at the direct expense of Internet Explorer. There’s just no getting around it. Opera is not making any serious gains.

  5. January 10, 2006 at 4:30 am

    Not properly counting Opera in these statistics could also be due to the fact that Opera, by default, identifies itself as Internet Explorer

    Not this AGAIN?!?

    I can see the different target audiences. When W3schools grants both Opera and Firefox 2.5 times the marketshare that Netapplications does, that’s pretty obvious.

    I can kind of see the prefetch thing making a small difference to hit-based stats (but not to visitor-based stats), but people have this strange tendency to paint it as a conspiracy to deceptively inflate Firefox’s marketshare instead of a tool for developers to make their sites load faster by making use of idle time.

    But the UA spoofing? Still?

    I swear, every two weeks someone brings up this myth. Opera IDs itself as Internet Explorer in the same sense that Internet Explorer IDs itself as Netscape.

    Here’s my copy of Opera identifying itself as Internet Explorer: “Mozilla/4.0 (compatible; MSIE 6.0; X11; Linux i686; en) Opera 8.51” On Windows XP it would use “Windows NT 5.1” instead of “X11; Linux i686”. (I wrote a huge article on these strings a few summers ago.) You may have noticed that even when identifying itself as Internet Explorer, the fact that it is Opera is still clearly visible. If a stats package can’t figure this out, the stats package is just plain broken.

    I mentioned that Internet Explorer pretends to be Netscape. Way back when, Netscape was the dominant browser and identified itself using its code name, Mozilla. Netscape added new features to its browser, and site authors started looking for Netscape visitors in order to show them the nifty new features. Microsoft wanted to make sure that Internet Explorer would get these features too, which is why MSIE’s identification string starts with “Mozilla.” And yet somehow stats purveyors manage to figure that one out.

    What Opera basically does is wear a badge saying “I’m MSIE (actually I’m Opera).” It fools the casual glance, but the actual ID is clear to anyone paying attention.

    There is a very small list of sites on which Opera actually hides itself when identifying itself as MSIE. You can find these in ua.ini. Sites set to 5 will pretend to be MSIE and hide the fact that they are Opera. Sites set to 4 will pretend to be Mozilla and hide the fact that they are Opera. Right now we’re looking at 8 sites — that’s 8 sites in the world — where Opera really, fully pretends to be MSIE: Slate.com, the Japanese version of MSN, and six banks. There are 14 sites where it pretends to be Mozilla and completely hides itself: Ebay, Getty Images, Mac.com, and a bunch of banks. That’s a total of 22 websites (up from 1 a few months ago) that can’t tell you’re using Opera.

    Every other site in the world has the ability to distinguish an Opera browser identifying itself as IE from a copy of IE. Unless NetApplications is getting most of its data from Slate, Ebay, and banks, I don’t think it’s impacted much.

  6. January 10, 2006 at 5:17 am

    >href=”http://www.doommovie.com “>Doom the Movie DVD – At Olduvai Research Station, Mars, something has gone terribly wrong. The now deceased researchers at the station have quarantined the site and its up to Sarge (Dwayne “The Rock” Johnson) and his crew to clean up every hellish creature they find. Get your href=http://www.doommovie.com >Doom the Movie DVD at http://www.doommovie.com

  7. Anonymous
    January 10, 2006 at 7:55 am

    I’d like to add one thing, but I don’t know how relevant it is…

    Some proxy servers change the UA. I’m typing this behind an ISA server, fully identified as Opera, and yet sites think I’m “Mozilla/4.0 (compatible; MSIE 5.5; Windows 98)”.

  8. Anonymous
    January 10, 2006 at 8:03 am

    Whoops… 😦

    I’m sorry. It’s not ISA that is to blame, it’s NTLM APS (a local proxy I’m running).

  9. January 10, 2006 at 9:24 am

    http://davidnaylor.org/test/results.htm

    The only statistics service not to recognize Opera when ID’d as IE is S-Tracking. Ever read a public stats report from them? Didn’t think so.

    The day you do I’ll accept the claim that statistics services don’t recognize Opera properly.

  10. January 10, 2006 at 9:24 am
  11. Anonymous
    January 10, 2006 at 10:49 am

    I don’t know how they count, but I have site with ~300 people every day. My result (last 30 days = ~4700 different people):

    1. Internet Explorer engine 74,79%
    2. Mozilla engine 15,33%
    3. Opera 9,17%

  1. No trackbacks yet.
Comments are closed.
%d bloggers like this: