Alexa.com is an Amazon-owned, Microsoft-powered Search Engine that measures traffic-popularity based on a sampling of about 10 million users.
Alexa was the first traffic-measurer and is often maligned, mostly unfairly
by people who do not have a firm grasp of statistics and how to account for sampling error inherent in any such system. Other traffic measurement tools now exist and we'll cover them shortly. When used together, you can get an excellent estimate of a site's traffic.
And this article will show you how to use Search It! to get five traffic estimates at five traffic-measuring sites later, including Alexa.
Alexa is the most widely known. No other site-tracking system has more users. We use Alexa's objective ranking scores to support the value and effectiveness of
Site Build It!. For example, see these sites...
No other site-tracking system has more users than Alexa. However, it does have certain limitations that you have to watch out for.
First of all, what is Alexa totally useless for?
Answer... measuring your own traffic. You have a far more reliable tool at your fingertips -- your own traffic stats tell you with great precision how many visitors you're getting. So don't be depressed by an Alexa ranking of 1,000,000 if you have 5,000 visitors per day.
What is Alexa great for?
Answer... Comparing how you are doing against your competitors. The odds are your visitors and their visitors are similar in nature so they likely use Alexa to a similar degree. That means not much bias when you stay within an industry -- but we'll get to bias in a second.
What about overall reliability?
Answer... Sites can get weird results, with a good Alexa ranking but with only a low number of visitors (and vice-versa) -- we call this "scatter" (more below). Often that can be due to something like the user having the Alexa toolbar and a dynamic ISP (i.e., your ISP address changes whenever you log in to use the Net) which can fool Alexa into thinking you are getting many more visitors than you really are -- especially, if you also delete cookies frequently.
Some marketing pros even manipulate their traffic ranking to appear better than it
really is. Why? To scam advertisers. It may work for a month, until advertisers realize they aren't getting the traffic they were promised. Not good.
Overall, the correlation of "visitor counts" vs. "Alexa ranking" (across thousands of sites) is good. SiteSell knows because we run scripts comparing thousands of SBI! sites.
Is the correlation perfect? No, there is a great deal of "scatter" (extreme examples -- an Alexa 150,000 ranking with only 20 visitors per day, or an Alexa 500,000 site with 5,000 visitors per day).
The scatter makes correlation less reliable after 100,000 and almost totally useless after a 1,000,000 ranking. But poor rankings still yield useful information...
Anyone with a ranking of 1,000,000 or worse (i.e., higher) is simply not getting much traffic, if any. Beyond 1,000,000 your ranking can bounce around substantially with just
a change of a few visitors to your site.
A site with an Alexa rank of 1,000,000 is still likely to have a bit more traffic than one with 5,000,000 or one that is not known to Alexa (meaning that there were no visits by the millions of folks using the toolbar during the prior 90 days). Think big picture.
Overall, if you plot those thousands of sites on an "Alexa vs. visitor-count" graph,
there is a good correlation and it is a very clearly a worthwhile tool, as long as you understand a little probability theory and how Alexa works. And you don't get
shaken up by "one-of" reports of how someone or other manipulated Alexa, etc.
One-of reports are useless. They are not statistically significant. And they may have their own agenda, reporting falsehoods. Only a study of large numbers of sites can count when studying the reliability of any product such as Alexa.
There is no point to manipulating Alexa since it's your own actual visitor count that matters. And who are you fooling, anyway? Only the cynical and dishonest would do that and claim that others do it. There is just no point, unless you're the same type of person who used to put a fake cell phone antenna on your car to "look good" (way back in the 80s!)
Here's a great "primer" on how Alexa works and calculates their numbers....
But the most interesting information (for this discussion) is their section headed Some Important Disclaimers.
Yes, there is bias, as there is for any sampling by any sampling device or service. The two bugbears of any sampling are statistical error and bias, and the key is to understand each and know how to account for it while using it. So let's look at each...
1) Statistical Error
Alexa has the lowest statistical error, since it has by far the largest sample base generating data for it (millions of toolbar users).
Alexa is very clear that stats on sites with traffic rankings greater than 100,000 are less statistically valid. SiteSell's own studies show "better and better" correlation with "better and better" Alexa rankings, "good" correlation down to the 300,000 mark, and "OK" to about the 500,000 level. It becomes pretty unreliable after 1,000,000.
Do not focus on getting a better Alexa score. An Alexa score is merely a good comparison tool. Focus on visitor count in your traffic stats and good things will follow.
The really good use of Alexa is to track your performance vs. your competitors. There's no point in having a high Alexa score just to have a high Alexa score -- it won't
bring you an ounce more of business.
Bias? Certainly. Two obvious examples existed for years...
i) Korea -- We don't know why (our guess is that the Korean browser may have Alexa installed), but Korean sites ranked too highly.
ii) Net marketing -- absolutely. A higher percentage of SiteSell visitors, just like PayPal or Linkshare users, have Alexa installed. They are more likely to be interested in this kind of data, to know about Alexa, and to install its toolbar.
Recently, Alexa has been correcting the bias. So you'll see many "Net marketing" companies with decreasing Traffic Ranking. For example, our own ranking has decreased substantially while traffic has been increasing. Compare with others and you'll see even bigger drops, all part of the algorithmic correction. So again...
If you know and understand statistics and sampling, how to use Alexa to compare against others in your industry, you can get valuable information out of it.
A Quick Test of Reasonableness
Scan through the Top 100 English sites at Alexa...
Surprise! They "make sense" and they are not all Net marketing sites. Other (non net-marketing) people use Alexa, too!
And as you can see... those sites are pretty well known. The better the ranking, the more reliable it is.
In any event, regardless of statistics and bias...
Alexa remains a great way for you to track how you are doing within your industry. For this, Alexa jumps from "good" to "excellent."
Why does SiteSell use it so much?
It establishes Site Build It!'s success-delivering credibility. SiteSell has grown quickly. That only happens for one reason. We offer a terrific product, Site Build It!, that really does build Web sites that work. Alexa is an objective way to demonstrate the success of SBI! sites. For example...
SBI! owners rank highly (see this site, results.sitesell.com) because they have the visitor count to back it up. If they are "scatter" (low traffic, good Alexa ranking), we exclude them as proof. And there are SBI! sites that have thousands of visitors per month that do not "make it into" results.sitesell.com because their Alexa ranking is not good enough to "make the Top 1% cut."
So results.sitesell.com is actually "better" than it looks (i.e., SBIers do better than this page suggests). But here's the important point...
It's just not possible to fake a page like that, as some cynics suggest. Quite the opposite -- each site has to have the traffic before it can make this page, so the page is not a self-fulfilling prophecy. And we don't "build up" sites to have high counts -- it
would be impossible to do that for the hundreds of sites on results.sitesell.com in any event, and simply not worth our time.
The bottom line is that this page really does indicate how fantastically well SBI! users do. And we are able to create this page because using Alexa...
So what's your take-home lesson?
Track your own internal traffic stats through your Traffic Stats software. Use Alexa to...
All the information that you need is on the Alexa site.
Bookmark it and use it often. The deeper you get into the advanced functionality, the more you'll get out of it.
P.S. People seek perfection in things that are not meant to deliver perfection. Alexa is a great "big picture" tool. Let's compare with the world-famous Nielsen ratings, which measure offline TV usage, supposedly for all of America.
Did you know...
- ... that the Nielsen Ratings system follows only 5,000 families (vs millions of Alexa toolbar users)? Now THAT is a small sample size. But advertisers spend billions based on the results, understanding that the sampling error is enormous, especially if you want to know how many women are watching TV at 11 PM in Cleveland.
...that during the famous quarterly Nielsen SWEEPS, only a minority of questionnaires are actually completed and returned, leading to sampling bias. "Memory bias" is also introduced since they are not completed in "real-time" (Alexa is real-time, of course). Yet networks place all their BLOCKBUSTER shows to boost those ratings artificially. And TV advertisers, yup, spend billions.
No tool is perfect. But if you know how to interpret through the weaknesses, you can get a world of useful info from Alexa.
P.P.S. Many years after Alexa launched, new services emerged that
also offer good ways to obtain traffic estimates of other sites. See the Popularity section of Search It!. Choose Popularity in Step 1 and then (Step 2) choose "5-in-1 Traffic It!" to get rankings/traffic counts for any site from Alexa, Compete, Quantcast, Google Site Trends and Ranking.com... all at the same time!
You'll find that no single service is perfect. But when you learn how to use all
five together, you'll get the best traffic estimate possible for a site (unless you have access to their actual traffic stats!).