Alexa rankings losing accuracy, what’s to blame? Social media traffic
By Justin Montgomery
Seems a lot of people have noticed inconsistencies in Alexa’s traffic reporting since its most recent algorithm update. The changes seem to have downgraded several tech-centric websites with a lower ranking even though true traffic numbers are showing the opposite. What’s to blame for the changes?
According to Collective Thoughts, it looks like Alexa has stopped counting social media traffic all together. Alexa looks to have enforced a new way of filtering which docks sites that see a lot of traffic from popular sites like Digg, StumbleUpon, and even from feed readers. Why? The only possible answer is to appease those that use the Alexa data for advertising figures.
Huge traffic spikes are possible from the largest social sites that can lead to an imbalance in their trend data, especially for the most recent 3 month period that appears in their toolbar by default. This can alter the conception for an advertiser that a site is actually more popular than it is, but it seems to be having the opposite effect. Worst off, it’s enraging those that rely on the ranking to determine the reach of their site.
Other traffic-gauging sites like Google Trends and Compete.com seem to be giving more believable stats as of late, and something else will most likely become the de-facto source for traffic numbers, but Alexa has been so popular in the past- why change it and upset its users? If the sole reason was to appease the aforementioned advertisers that rely on its data, then I completely agree, but some are saying that wasn’t the reason.
Allen Stern, owner of CenterNetworks.com, had an interesting video-post outlining his thoughts on the recent changes, and stated that when he was once fielding presentations in an earlier career, and someone would show Alexa graphs to prove their points, they would be “immediately shown the door.” This disproves of the “advertising theory” all together, and again begs the question of why such drastic changes.
If you ask me, there’s really no other explanation as to why this would have happened, other than the social media theory. When sites like TechCrunch and CopyBlogger start showing a dip in Alexa rankings, there’s obviously something wrong. Since these sites, and many other tech-related sites, rely heavily on social media traffic- the lack of this data is the only logical explanation for such a decline.
Related:
Entry was posted
on Sunday, August 10th, 2008 at 8:52 pm
under 
Print This Post
Read more entries by Justin Montgomery.
Stumble It!
August 11th, 2008
Nice post, very informative. Lets see how they’ll remedy this problem.
August 11th, 2008
It will be interesting to see if this recent commotion will finally get some explanation out of Alexa, especially with the big voices joining.
But in general I believe we already need a tool that will represent actual influence instead of traffic to use its rankings when talking to advertisers because they often buy reputation for their brands instead of traffic itself.
August 12th, 2008
Luckily, I don’t think it’s anything quite so ominous.
Alexa is a panel based service (around a million I believe, out of 204′ish mm US internet users), meaning they take a small slice of internet users that they gather data on and try to extrapolate those visiting habits out to the internet population as a whole. As you can imagine, this is very difficult to do accurately, especially for smaller sites that don’t have a lot of reach.
Every panel also has inherit biases that must be corrected for. Perhaps Alexa’s panel happened to skew heavily towards tech enthusiasts, so tech sites were being over-represented. It’s actually likely that they simply improved their panel inference algorithm which better corrected for particular biases.
I’m no fan of Alexa and panel based measurement for the Internet, but I think Alexa is smart enough to know that any hint of favoritism towards any site would be its death blow.