"There are three kinds of lies: lies, damned lies and statistics." --Disraeli
In a recent post, I ended up getting into some off -topic comment flaming with a reader who it turns out (IMHO) really wanted to play the "my site is more popular than your site" game. Really, this kind of "mental masturbation" is sort of childish. But, it did make me think, and I thank him for that.
In the traffic measurement game, there are some upstarts (like Quantcast, Compete and others) that don't seem too different from Alexa.
Based on the evidence I've looked at from some of the pundits, there isn't much indication to suggest that Compete or Quantcast are better than Alexa. And we all know that Alexa's data has flaws. If you analyze on toolbar installs, these new services certainly have less data. They come up with "ISP relationships", "Panels" and other inventions to show that they can compete, but I don't necessarily buy it.
Some pundits have given Compete.com positive comments. But Matt Cutts of Google says no.
Regarding third-party traffic estimates and actual traffic, Technorati links win as the best predictor. Personally, I have found Google Analytics traffic stats to be very close to actual IIS logs in the small comparisons I've made. And that's a free service. It's just that you can't see my stats - only I can.
I think the bottom line here is that some services that aren't really supposed to be traffic measurement services are ending up predicting actual traffic better than those who make the claim that they "are" traffic measurement services.
Personally, I think Google is the best measurement of site popularity (traffic or not). They have the biggest installed toolbar base, and they index more "stuff" (backlinks, etc.) than anybody else. And, they have a proven measurement system. It's called PageRank. All the hot-shot SEO people hate them, but I think PageRank is about the best indicator out there. It's objective, it has been tuned to reject SEO "spam", and it has withstood the test of time.
And with Google dominating SERPs with up to 90% of my search engine traffic, I want my PageRank to be just as high as I can get it, thank you!
Where I work, at a NYSE listed IT consultancy, we're building a Search Engine Marketing (SEM) and Search Engine Optimization (SEO) practice and I get to use my programming skills to build some exceptionally cool measurement tools. Business is literally flying in the door.
It took Robbe Morris and I since the year 2004 (when PageRank first became known) to get Eggheadcafe.com up to a PR 6. That's not bad for a niche site that only deals with developer topics. This UnBlog, which is much younger, is a PR 5 currently. But you have to understand that PageRank is "Logarithmic" - there is a huge difference between a 5 and a 6.
That's my 2 cents.