Alexa.com (owned by Amazon) provides information on web site popularity. They do this by people downloading the Alexa Toolbar. Alexa monitors where they go online. Based on the traffic they see going to a web page, they rank it. Typically, the top 100k sites show more data on their rankings page than others. Lower rankings are easy to manipulate by having some friends download Alexa's toolbar and then visit your web page every day.
As a result, Alexa's data is commonly thought to be unreliable. Our own log analyses show steady traffic increases but Alexa's data does not seem to reflect it. In fact, they showed that nearly a third of our traffic to our MailerMailer site was from Vietnam. We do have clients there but it is nowhere near a third of our traffic.
Even so, I periodically monitor our rankings just to see how Alexa thinks we are doing. A couple of days ago, I saw a rather dramatic spike. I went to Alexa's site to find out what the deal was. It turns out that they totally revamped their ranking algorithms and now supposedly rely on “more sources of data to give a better indicator of web site popularity” according to their announcement.
I'm not sure what this really means, but our rankings are higher than their previous numbers. They now show that 5% of our traffic is from Vietnam, 4% from the UK. That's still very inaccurate; we get far more UK traffic than Vietnam traffic. These third-party measurement sites can be really off base.