Lately I'm seeing more and more bloggers including Alexa traffic graphs in their blog posts, usually to compare some sites' relative success. Here's an example post that has Alexa graphs.

Since they are the only statistic that most people can find on web traffic, people will naturally use Alexa. VCs do it, pundits do it. No one mentions that the Alexa sample can be very mis-representative of the traffic for certain types of sites.

Simply put, alexa traffic numbers are heavily skewed in odd ways, because they depend on toolbar data from an ancient toolbar. For example, did you know:

• Alexa doesn't count any Firefox users?
• Alexa doesn't count any Mac users?
• Alexa massively undercounts AOL users?

It's great of Alexa to publish all this data. But all the people that include an Alexa graph on their blog or in their pitch to investors are just reinforcing Alexa's position as the arbiter of what's successful on the web. Caveat emptor.

Now if Google or Yahoo cared, they have all this data from their toolbars and other sources, and they have a much, much more accurate sample of worldwide internet usage. 20% project anyone? Or are you gonna let Amazon be the only big internet company that actually gives some interesting data away?

¶ 4/16/2006