Our traffic is spiky, but not as much as the Alexa graph suggests. In reality our growth is much more steady; every now and then Alexa seems to "catch up". It's weird, and we're not sure what would cause that effect.

How close are Alexa traffic estimates to the actual traffic volumes? I've heard from numerous people that Alexa is almost always wrong and wrong by several orders of magnitude. It could be better for big sites, not sure. Just wondering what you guys see..

Alexa's numbers are crap. Total crap. They only reflect your traffic in the loosest way. There was a period of time where our traffic doubled and the alexa actually declined. It eventually catches up though....think of a random distribution around the true value.

Just so you don't get the idea I'm picking on Alexa, estimating this stuff is hard (or the people doing it are incompetent, but that seems unlikely with so many entrants). Quantcast and Compete.com are even worse, and ComScore is expensive but no better than Alexa.

I've always found quantcast to be much more accurate, although their number for justin.tv looks low: http://www.quantcast.com/justin.tv

what he's implying is that it's (American) football season -- both collegiate and professional -- as well as grand finale of regular season baseball, great football(soccer) league matches, tennis, etc. and any curious passer-by wonders how much of the traffic spikes are due to this subset of viewers

Alexa's totally crazy way of data collection?