I was reading Michael Arrington's TechCrunch yesterday, and I saw this. It reminded me of a pet peeve of mine - just how far off Alexa's web site statistics can be. Since reading the piece, I have been watching the comments roll in with amusement.

As recently as a year ago, I relied on Alexa rankings to give me a broad view of web site traffic. I've touted the charts stats on this blog and put stock in them personally as a useful metric. I've used it for competitive analysis, and to tell apart the windbags from the companies that are the real deal.

(Does anyone else even remember Techies.com?)

It was always obvious that the stats were not exact, but I did have enough faith in the numbers to use them as a guide.

Recently, I have been I have seen enough bullshit numbers to convince me that the Alexa numbers are completely useless.

The other day, Jason Davis and I were talking about this very subject. As I’ve said before, Jason’s done a spectacular job with Recruiting.com, and Alexa tracks the rapid rise of the site.

However, Alexa currently shows that ERE.net and Recruiting.com have comparable levels of traffic. When Jason and I compared the actual number of page views on each using our server logs (Google Analytics for ERE vs. I am not sure what Jobster uses over there) ERE.net had nearly four times as many.

Alexaaaaaaaaaaa!!!

Measuring the behavior of a small sample of Internet users who have downloaded a toolbar may have once been a decent way of measuring traffic, but it is clear to me that they are doing a disservice to the Internet community at this point.

There’s a great business opportunity out there for some entrepreneur to create an Alexa that actually works.