John Cook posted on his blog a link to us. As always, we learn about new companies (BigOven) and add to our list of Startups in Seattle. We have more than 150 companies so far.
I just want to remember everyone that the Alexa Ranking is just a fun subject, and it shouldn't be taken seriously, seriously.
For those that don't know, Alexa works by gathering statistics data from its toolbar users. You install the Alexa toolbar and every site you visit the toolbar notify the Alexa servers. Based on that they compute how many page views and how many unique visitors a site received in a day, and that drives the Alexa Traffic Ranking.
There are an estimated 2 million users of the Alexa Toolbar, and about 598 million of users that don't have it.
The first catch is that Alexa is using 0.3% of the Internet population as the sample. That would be great if they applied demographic data (age, gender, region, etc.) and normalized the results, but they don't.
The results of the lack of demographic data is that Alexa Ranking is skewed towards technically savvy, US-based people. Which means that Web 2.0 sites rank higher on Alexa than, let's say a site for a Wenatchee newspaper (rank #201,132), despite the fact that it might have more visitors every day.
The problem becomes even more palpable when you talk about Web 2.0 sites that might use AJAX and Flash to improve user experience. Recently, we realized that we were about to achieve more than 1 million page views on Sampa (excluding bots, crawlers and other critters), but the Alexa ranking was completely off for us. We know for a fact that other companies with 1 million page views per month are in the 10,000-20,000 range. After just a little investigation it became very clear. Alexa doesn't count page views inside a frame, and Sampa uses that a lot to improve client-side caching on the browser. We also use AJAX at a lot of places and Alexa also ignores that.
Other sites that might make heavey use of AJAX and Flash are also under-counted.
The conclusion is to use Alexa responsibly. 