I knew Alexa traffic ranks were woefully inaccurate. We knew they were biased toward blogs and sites about SEO, marketing and monetizing. But only today did I realize just how pathetic Alexa’s rankings are. Today, I realized the biases are astronomical.
In fact, if you rely on Alexa to compare the relative traffic at blogs you may mis-gauge relative traffic by a factor of as much as 20,000! I said twenty thousand.
Don’t believe me? Well, read on and I’ll provide evidence by comparing the Alexa traffic ranks for three sites: Wendy Knits an A list knitting blog, my B list knitting blog and Big Bucks Blogger my two month old Z list “make money” monetizing blog.
First, let’s look at the graph to the right which shows the Alexa traffic “reach” for these three sites. The Knitting Fiend is shown in red; Big Bucks Blogger is shown in blue and Wendy Knits is shown in mustard. I selected The Knitting Fiend and BigBucksBlogger because I write both blogs. I selected Wendy Knits because I like and visit Wendy’s blog, she published a book on the strength of her blog visits, Wendy Knits also has zillions of comments.
Now look: According to Alexa, both my blogs have larger reach than than Wendy! Instant success! 
Let’s get a dose of reality. The chart says Big Bucks Blogger currently has a greater reach than The Knitting Fiend: Not true.
I write both and let me assure you, The Knitting Fiend gets way more traffic than Big Bucks Blogger by any metric. According to Sitemeter, over the last seven days, The Knitting Fiend averaged 1,416 daily page views; Big Bucks Blogger averaged 53 daily page views. In April, I estimated Wendy Knits’s daily page views as 45,000 - 60,000.1
If remotely accurate, Alexa should show that Wendy Knits gets the most traffic and BigBucksBlogger gets the least: Alexa shows exactly the opposite.
Below, I tabulated the statistics according to Alexa. In addition to my estimate of page views, the table shows Alexa’s figures for 1) “Reach”, which is the fraction of internet users who visit a site, 2) Page views per user and 3) Alexa traffic rank for last week. (Recall, #1 is the best possible Alexa rank.)
I calculated and added a value I call “Traffic” by taking the product of “Reach” and “Page views per user”; in principle, this should tell us the fraction internet traffic arriving at a particular site. This number should be linearly proportional to page views: that is, if one site has twice as many page views as another, it should have twice the “Traffic” as the other site.
| Actual | |||||
| Site | Page Views (daily) | Fractional Traffic | Page Views Per Visitor | Reach | Rank |
| WendyKnits | 45,000 - 60,000 | 0.0004% | 1.3 | 0.0003% | 421,742 |
| 1,416 | 0.0036% | 4.8 | 0.00075% | 135,852 | |
| Big Bucks Blogger | 53 | 0.0075% | 4.7 | 0.0016% | 71,187 |
If you examine the numbers and put on your thinking cap, you will start laughing! Recall: In reality, Wendy Knits gets 30-40 times the daily page views of The Knitting Fiend; The Knitting Fiend gets about 30 times the daily page as Big Bucks Blogger. That means that Wendy Knits gets approximately between 900 and 1200 times the page views of Big Bucks Blogger.
Yet, according to Alexa, last week Big Bucks Blogger had roughly 19 times the traffic Wendy Knits received!
So let’s do a little “what if?” If Alexa’s miscount (due to unevenness in Alexa toolbar use) were the same as it currently is, but BigBucksBlogger actually got the same amount of traffic as WendyKnits, what would Alexa tell us?
Alexa would tell us BigBucksBlogger got 16,000 to 22,000 times the traffic as WendyKnits!
That’s right: Relative to WendyKnits, Alexa over-counts BigBucksBlogger’s traffic by a factor of roughly twenty thousand
Now, we all sort of knew that Alexa was biased in this way: SEO / Monetizing / and blogs about blogging tend to be over-ranked. Personal hobby blogs tend to be underranked.
But did you really expect the relative measurements to be off by a factor of roughly twenty thousand?
Notes:
1. I based my estimate of Wendy’s traffic on the ability of her site to drive traffic to mine which was discussed here. A conservative estimate suggests Wendy Knits gets 30 - 40 times the page views that The Knitting Fiend receives, or ~30 * 1,500 = 45,000 page views a day. One days worth of statistics she mentioned in an email also suggest the factor of 40.
it’s a shame because so many sites base your blogs worth on Alexa ratings. They show that I only have 16 links yet my site went from 7,000,000 in ranking to 717,000 and I have counted the links in that I know of and they have still not changed the link in count on my blog. It’s infuriating.
I absolutely detest Alexa and the way that so many companies base your site’s popularity on a flawed algorithm.
It makes me crazy!!
- Martin Reed
You’re right, it’s depressing to see your traffic go up and Alexa rank go down and for all the webmaster-type sites to flaunt a high Alexa rank that is just an illusion. The sad part is many people put a lot of stock in that rank.