Some folks have been asking for my opinions of Hitwise & Alexa data recently, and I figured the best I could do would be to illustrate the sampling problems of both with some real life data.

First off, let's look at SEOmoz's traffic levels over the last few months. This comes courtesy of our Indextools stats program + Feedburner data for RSS. You can see more on our advertising page.

SEOmoz Traffic from July to Present Day

The graph shows the days we had big linkbait, the linkbait bump and the relative levels of increase as we went from around 5,000 daily uniques to 7,500 daily uniques (July to August).

Feedburner Stats for SEOmoz's RSS Feed

The Feedburner data for "all time" says our average number of subscribers per day is ~3000, with a majority of those arriving on days when we blog a lot (they're less sensitive to linkbait than the visitor traffic numbers).

Alexa Data for SEOmoz over the summer

Alexa is estimating that for every million Internet users, SEOmoz is read by 1070 of them. We're also nearly into the top 1000 most popular websites in the world. Obviously, this data is completely bogus - our traffic rank is probably somewhere between 80-100,000 (or lower) and our reach is probably around 5-10 users per million.

Hitwise's Rank for SEOmoz among all "Web Development" focused sites

Hitwise's data shows us a few positions above Flock (the alternative browser) and below zen-cart.com. We're the 611th most popular site in the web development category, and while I think this data might be closer to accurate, it's worth looking at who Hitwise thinks is at the top of this chart:

Hitwise on the Top 20 Sites in Web Development

Real traffic data was tough to come by, but one of these sites - Pimpyourpro.com - had this to say on their advertising page:

Would you like your site listed on one of the most popular MySpace Help websites on the net? We average 1.4 million page loads and about 185,000 unique visitors each day.

I'm inclined to trust those numbers to a relative degree (though I don't doubt they could be somewhat inflated). Basically, we're looking at a site that gets approximately 25X the number of visitors per month as SEOmoz. Let's see what Alexa has to say about that:

This is where you start to see the massive disconnect that Alexa can sometimes provide. Hitwise isn't always better, especially as they don't provide any real estimation of traffic level, just comparisons. What is illustrated above is the need to go beyond a reporting service and get the data yourself to be sure you're looking at something accurate. When Alexa or Hitwise gives you numbers, compare those to what similar sites are reporting on their advertising pages or in public stats they provide. Using those figures, you can generally get a good sense of who's under or over-reporting and to what degree.

p.s. With Hitwise reporting that Pimpyourpro gets 2.07% of traffic in that market and SEOmoz.org gets less than 0.01%, they're underreporting our traffic by an order of magnitude of ~10X. I.e.

Imagine that all the webdev category sites receive 9 million uniques per day.
Pimpyourpro.com - 9 mil x 2.07% = 186,300 visitors per day
SEOmoz.org - 9 mil x 0.01% = 900 visitors per day

In reality it should be something like:
SEOmoz.org - 9 mil x 0.085% = 7650 visitors per day