Wierd Alexa numbers
So everyone knows that Alexa rankings are not necessarily the most reliable, especially when they’re large. It’s not controversial, Alexa itself admits it on it’s traffic page. It has a large selection bias and non-random distribution - so that unlike Nielsen which does all manner of statistical randomization fanciness so that it can sort of get away with a small sampling - Alexa’s user base is elective, so statistically non-random.
Anyhow, I added the Alexa toolbar because I was curious about it. When I installed it prior to Google supplying me with any love (Google as of a week or so ago must have taken me out of the sandbox and is now sending a good bit of traffic my way) this had an immediate and obvious impact on my Alexa rank, which made sense - my reach (% of the internet) went up as did my pages per visitor. So given the very little traffic I was getting, I was not surprised.
And very broadly as my traffic has risen so has my reach - generally in tune although sometimes Alexa would zig as my traffic zagged, but if you drew a best fit line through both graphs, they’d have a similarish slope. The thing, though, that is wierd is the pages per visitor. I can not fathom how Alexa is calculating this. As of this posting Alexa believes that the average number of unique page views per visitor per day (for yesterday) is 45.5 pages.
Whaaa?
Let’s just say that that is impossible. Not one person, including my obsessive self, has viewed 45.5 pages on my site. Especially as of this week where I’ve been spending a lot less time noodling around. Whereas with the reach number I can understand how and why it is wrong with this page views number I am genuinely baffled, especially so because my reach is rising. That means that Alexa is tracking more people to my site. If my reach was low and remained low such that perhaps Alexa thought only me and one other guy were hitting the site, I could understand it - but clearly there’s more than a few people with the toolbar that are arriving. I’m just completely at a loss to understand this number.
Anyone out there have insight into how the page views per user is calculated?
