What is Wrong With Alexa Traffic Ranking?
January 3rd, 2007 by Andrew Chen
Both John and Tyler replied my post. I am really proud. When I read john’s post I didn’t believe a company backed by Amazon is running such a low quality traffic ranking system until I compare the traffic of the company I am working for to John’s blog. I didn’t think of doing that before. I am working for Local.com as database developer currently as I told you in my about page. According to our financial report. Local.com attracted more than 10 million visitors each month and is in the top 100 web sites in the US. When you look at Alexa traffic, here is it:
Its daily reach is even less John chow’s blog. However John’s blog has less then 100,000 visits last month according to his post. So John’s blog is only one percent of the traffic of Local.com. How can it have an average daily reach more than
Suppose Alexa technically competent to build a web traffic ranking system. Then the problem would be on its tool bar’s download population. I myself didn’t download Alexa tool bars and never use it. But according to Alexa, the tool bar was downloaded over 10 million times. I would say that download base should be able to give statistically meaningful results. But 100,000 visit ranks the same as 10 million? It is just doesn’t make any sense. Would it be that the 10 million Alexa tool bar users are more likely reading blog? It might be.
If Amazon uses its fame and financial strength to strike a deal with firefox or even a smaller company like Algoco, it can broaden its user base and increase the accuracy of it ranking system. If one day Alexa became the ultimate authority of web traffic ranking in the world, what kind of money it is going to make. I guess we have to ask Alexa, what is going wrong?
Related Posts:
(No Ratings Yet)
Loading ...
Alexa rank is not based only on traffic from their company toolbar. Many small websites create their own toolbars via Alexa’s service and people who use it on that way (even if they maybe know it) create traffic too. In Serbia (where I live), it’s popular that kind of creating toolbars.
Also, Alexa give information of how their system works, so anybody can create his own toolbar or any add-on which sends traffic to Alexa.
I use one kind of that. It’s Search Status - excellent Firefox SEO Toolbar Extension. It’s excellent tool for any web developer. Before it, I never use any Alexa toolbar, but now I send traffic via Search Status.
Hi Andrew. It is pretty hard to do head-to-head comparisons like this without looking at all the factors. In your analysis you made one big dunder-headed mistake and at least one other major incorrect assumption.
Let’s start with the first big mistake. What is local.com? When they claim they get 10 million visitors per month, what pages are those visitors visiting? I’d bet that Local’s financial report is counting traffic to all pages that have the Local brand on them, whether or not they are on the local.com domain. These would include oodle.com, indeed.com, whitepages.com, and judysbook.com, among others. Alexa is only counting the traffic to local.com. All those other domains are not counted at all.
Now, let’s get to one of the major incorrect assumptions. What is a visitor? How does Local choose to count its visitors when bragging on their financial statement? Do you suppose they use a liberal counting method or a conservative method? What if a user visits several times a day… is he counted several times? What if he visits dozens of times per month, is he counted over and over again? Many sites count this way because it is an easy way to inflate their stats to make their investors happy and their advertisers eager to sign up. I can’t say for sure how conservative John Chow is when counting his users and/or whether he uses cookies to keep track of them. But I can say this, he most definitely is NOT using the same method as local.com and he probably doesn’t have as much incentive to exaggerate his visitors numbers.
That, in essence, is the value of a 3rd party rating service. You can be sure that we are counting both of them the same way.
So, is Alexa right? I don’t know. But it is definitely not nearly as wrong as you suggest.
I didn’t notice that they have other way of collecting data. But seems downloaded tool bar is till their major data source. This is what it says on the tool bar download page:
And again if they have other ways of callecting data. The sampling base should be more diversify and the result should be more accurate. Maybe their engineer can dig into the traffic data of Johnchow.com and Local.com to find out what is wrong.
I am pretty sure Local.com is only couting the traffic to the page the company host becuase these add on services, oodle.com, indeed.com and whitepages.com were not there at the time when the financial report released.
Take a step back treat the 10 million as page views and compare to JohnChow.com’s 184,640 page views it is still a difference of 50 times. How much better Alexa looks like? It still spells something very wrong with Alexa, either in its ranking algorithm or in its sampling base.
[…] started working on our Alexa ranking a few months ago until I read about how flawed their data is. Alexa admits that their data is far from perfect, and does not seem to have anything in the […]
HE HE!
I too am very confused by Alexa’s rankings. I found this out when using the Pralexa feature of Drupal’s Weblinks module.
I found that I (www.coofercat.com) have an Alexa traffic rank of about 1.9 million. I have no idea if that’s good or bad, so I compared against a couple of major websites. It seems my little ‘blog, running on the end of my domestic broadband connections outstrips slashdot.org (traffic rank: 939) and bbc.co.uk (traffic rank: 44).
Whilst I’m not sure about your comparisons to/from local.com, I’m absolutely sure the BBC pulls more traffic than Coofer Cat (although of course, I would love the reverse to be true
I see here: http://awis.blogspot.com/2008/04/alexa-ranking-system-has-been-changed.html that they state “it was with the goal of showing Alexa Toolbar users how popular any given site was within the Alexa community”, so I guess the Alexa community prefers coofercat.com to the BBC and slashdot. Sadly, it seems thats about all Alexa can tell me though.
How can I say this plainly.
As a student of statistics, it’s simple.
Bad sample = Bad Data.
Period.
If that’s too hard to understand, then maybe even finger painting, explaining it really slow still wont help.
Alexa has a biased sample, so the results are biased. Period.
Alexa data is only valid for the alexa community. They clearly have a data sample that does NOT match the rest of the world.
The author of this blog is right. The data they present is no good, unless you’re just interested in analyzing the Alexa community, or getting ranked high in their community. You can be #1 on google, but no where to be seen on Alexa. It happened to me with serveral of my own sites. These are big businesses, with big traffic, and millions in revenue.
Anyway, I’m making this comment because the author’s statement is astute, and poignant. You people that claim Alexa is accurate as a representation of the world need to pull your head out of whatever hole you’re stuck in….
Bravo! Well said!
Alexa.com should put a disclaimer on its website warning people that their data is “For Amusement / Entertainment Use Only” as Alexa rankings CAN be easily manipulated.
First of all, you need to understand that, there’s no anybody own the whole Internet. So it is impossible for other website to get your website traffic unless he or she is a illegal hacker.
Alexa traffic is based on it’s own toolbar traffic and made estimation, so it would have biased if your country doesn’t like Alexa toolbar.
If Microsoft and Mozilla would release the Internet traffic report, the data would be much more accurate because most Internet users use MS Internet Explorer and Firefox browser, but it would draw a security and privacy problems.
I would not defend alexa at all. Their data is flowed. Our website according to google webmaster tools has more than 500 incoming links but according to alexa, it has 54. I can trust google on this because I followed most of the links and I could see our site linked. The numbers of visitors to http://www.thezimbabwemail.com/ alexa gives is very much different to what all the analytics tools we have running. I have lost faith in Alexa as it is very unreliable.
Recently I started working on Alexa rankings. From your blog i m thinking i have to study more about that..thanks friend for sharing this information.