WebRevenue.org | Increase your Website Revenue

I can’t tell you the number of conversations I’ve had with fellow web publishers who cite their Alexa ranking with pride. I usually make a bet with them that I can start a site with any random name, and break into the top 10,000 ranks within 1 week. Not that this is a practice I engage in — I’m merely saying this to illustrate my point: that Alexa is severely broken, and will never be an accurate source of web traffic information.

The basic problem with Alexa is that it gathers data by using the “Alexa Toolbar”. The toolbar is a browser plugin that monitors your surfing habits, and reports those statistics back to Alexa where they are compiled with the data from other users into the pretty charts that you see on Alexa’s homepage.

A flawed methodology

There are fundamental flaws with this methodology of data gathering — the most important one being that the number of users who have the Alexa toolbar installed is too small to make a reliable statistical sample for most sites on the Internet. Alexa attempts to take this minor subset of Internet users and mathematically extrapolate greater Internet traffic trends. Not only is the subset of Alexa toolbar users a non representative sample, but we all know that web publishers surf their own site constantly. Since web publishers are more likely to install the toolbar (we’re one of the groups who most wants traffic data) we tend to blow the traffic trends out of whack just by installing the toolbar and surfing our own sites.

Ease of spoofing

Which brings me to the next problem: Spoofing. Since Alexa is based upon a small subset of Internet users who actually have the toolbar installed, it’s very easy to install Alexa on 5 or 6 machines, hammer away at a site and blow the Alexa rankings to the moon. Getting everyone in your office (or entire company) to do the same and you can appear (in Alexa rankings, at least) hundreds or thousands of times more successful than you are.

Clearly Alexa is busted.

So what are the options?

Many other statistical packages have arisen to fill this void in reliability — but none are perfect. For one’s own site, Google Analytics is probably the most accurate. ( See my recent article on Google Analytics vs. AWstats ). If you’re looking for bragging rights (as in, ‘you want to show the world how much traffic you get’) I’d recommend using Quantcast. Both Google Analytics and Quantcast have similar systems which involve embedding a short javascript into the footer of every web page. The javascript pings a central server and gathers an enormous amount of data from visitors’ geographic information, to pageviews, referring pages, etc.

You can follow any responses to this entry through the RSS 2.0 feed. Both comments and pings are currently closed.