« Bass-ackwards once again | Main | Brief notes »
Confused about Alexa
By Wolfie | May 8, 2008
I’ve never really understood why Alexa rankings are used as a guide to how popular a website is. As they themselves say that their data “may not be a representative sample of the global Internet population. To the extent that our sample of users differs from the set of all Internet users, our traffic estimates may over- or under-estimate the actual traffic to any particular site.” Yet everyone wants a really high Alexa rank.
Recently, Alexa changed the way that they gather data and arrive at the rank for a website. It used to be that you had to have the Alexa toolbar loaded for your visit to a site to count. But the ranks that were arrived at gave no indication of how many people were using the toolbar (as a percentage of total internet users) and gave no indication of how many sites were indexed - you might be ranked 544,232, but how do you know how good that is? If they only index 544,233 sites then it’s really bad. If, though, they index 54,423,200 then it’s pretty good. Relevance is what we’re talking about.
And that was a system that could be “gamed” quite effectively. I’ve got the Alexa toolbar loaded in Flock (it’s not available for Safari) and have managed to dramatically improve the Alexa ranking for all the sites that I maintain at home and at work just from my own visits. Being able to improve a site’s ranking by over 2,000,000 in less than 10 days from one person’s visits seems like there’s something wrong.
With the changes, though, they use the toolbar data backed up with “data from multiple sources to give you a better indication of website popularity among the entire population of Internet users.” Helpfully, they don’t give an indication of what those sources are. This should mean that Alexa rankings will be more accurate and more relevant, but I’m not sure that it does. The changes were made in mid-April (see this blog entry from the Alexa site), and immediately a lot of people were upset because their ranking got worse (see comments to the same blog post). The rank for this site improved by about 50,000 at that time, but since then hasn’t changed at all. Given that its been about three weeks, that seems odd (the old ranking system used to change the ranks every three or four days).
What’s also odd is that the current rank doesn’t seem to tie-up with Alexa’s own figures. On the traffic details page for wolfshowl.com, they state that all but 0.4% of the traffic comes from UK, US, Australia, Canada or Indoensia. The individual ranks for these countries range from 20,306 for the UK to 224,386 for the US - yet the overall rank for the site is 364,952. How does that work? If it’s done on averages, then the ranking in the 0.4% that comes from other countries must be really, really bad because the average of the countries they list is 140,002. Whichever way it’s done, how can a site have a ranking 120,000 worse than the result you get if you add together the ranks of the two countries that make up over 86% of the total traffic?
I’ve always wondered about the relevance of Alexa and lamented its use a measure for advertising spend, but now I’m also starting to wonder about its ability to add up.
Update (20 May 08): after going up slightly (to 303,195) on the very day I wrote this post, the Alexa rank for The New Wolfs Howl has this evening completely disappeared. This could be a glitch with their system, or it could be as a result of the changes they’ve made. Only time will tell.
Categories: Blogging, Internet |
Tags: Alexa, Relevance
Subscribe by: Email or RSS :: Read more Howling at The Wolf's Howl @ BlogSpot :: Copyright 2008. All rights reserved. No reproduction without prior permission. :: Disclaimer: All opinions in this blog are either my own or completely made up; you decide.
No comments yet.