The problem is is that both of these methods miss a crucial (vast)
and growing Internet population, namely the power users. In the case of
Onestat with their online counter systems, the problem is that
generally only two types of clients use this type of service, newbies
and occasionally businesses. In the case of Websidestory, the big sites
they monitor are not those likely to be those visited by power Internet
users, if you are one yourself, do you remember the last time you
visited Disney or Liz Clairborne?
That skips a vast portion of
the Internet’s users, namely the techies and power users. These guys
know what they want online, they know where to go, and a vastly higher
number of them use the Firefox web browser. The Internet’s users are
slowly becoming more knowledgeable then they once were, and this group
of power users are growing all the time. Experienced Web developers are
likely to be using their own log file analysers for statistics and for
those without log file access, counters like our own statistical counter
that serve the same purpose. The only places I really see remote hit
counters is on newbie web pages and free blog services (the exact
places where one expects people to be using the default browser that
came with their computer.)
To get a real idea of the number of people using Firefox and other
alternative browsers, companies like Websidestory and Onestat need to
do deals with sites like Groklaw, Slashdot, TheRegister, Wired,
W3schools and other such sites and combine their results with their
existing figures. Then perhaps we’d get a real picture of the browser
landscape.
<p>I suspect there is probably more to it than even this. <a href="http://htmlfixit.com/?p=479#comment-91">One comment</a> on this article points out that WebSideStory's stats are based in actual web browser usage whereas the stats in the HTML Fix It article also take into consideration the fact that many people don't download <a href="http://www.spreadfirefox.com/?q=affiliates&amp;id=44329&amp;t=1">Firefox</a> because it is given to them through a distribution of Linux or friends who have it on disc. I think those stats should also be taken into account although regardless of where the browser was obtained from, you do need to look at the web stats. Here I agree with HTML Fix It in that you need to look at as broad a cross section of websites as is possible. Alternatively, look at sites almost everyone uses like <a href="http://www.google.com">Google</a>, <a href="http://www.yahoo.com">Yahoo!</a> and <a href="http://www.msn.com">MSN</a>.
What do you think?