[sesqu]

Regarding Alexa and Compate, you're half right. They're wrong in useful ways, but not the way you suggest. What's important to realize is why they're wrong - sampling bias. So they give an accurate representation of a certain demographic, but the usefulness of that depends on the target audience of the site in question. I don't have any examples at hand, but I've come across plenty of big sites that pretty much flatline on Alexa et al - because their clientele is not welcoming to such measurements, for cultural or geographic reasons. So you can (usually) use them to contrast direct competitors, but I would be wary about quora vs. stackexchange.