Recently, I've tried to focus my blog entries on Analytics and more specifically, analyzing website Analytics data. In Part 1, I wrote about a simple process for determining the traffic value of your SEO campaigns. In Part 2, I want to talk about 3rd party data providers that tend to steer people, including web marketing professionals, in the wrong direction. Today's post is geared toward understanding the value (or lack thereof) of traffic estimation websites like Alexa.com, Compete.com and other similar services.
To begin, it's important to understand where the data for these services comes from. To quickly summarize, these services monitor the Internet use of a given segment of internet users and make estimates based on these segments. Click Alexa or Compete if you're interested in learning more about their methodologies. I don't want to get into the strengths and weaknesses of each as, at the end of the day, I think that accuracy is the only thing that is important. The following a less than statistically-significant experiment that I performed to determine the accuracy of Compete.com.
Comparing Compete.com to Website Analytics
For the following experiment, I took 6 websites, each of which have over 10,000 uniques visitors per month, and compared them to compete.com estimates to get a rough idea of the accuracy of this service.
For the following summary, I compared Compete's unique visitors with that of the given web analytics solution used (either Google Analytics or SiteCatalyst). The % in each box represents the difference between the two data sources. For example, 2% means that Compete data was almost dead-on-target and 98% would mean that data wasn't even in the ballpark. I've also included information on the industry from which the sample website is in (technology, travel, hotel, real estate and manufacturing).
% Difference Between Onsite Analytics and Compete Data
In general, compete.com unique visitor traffic estimates are no where close to actual analytics data collected from 6 websites reviewed. Based on this analysis, I have no reason to believe that compete.com is capable of accurately estimating website unique visitors.
Is Bad Data Better than NO Data?
I regularly speak with clients and marketers (some of which are fairly Web saavy) who regularly use data from sources like Compete and Alexa to make comparisons. While many of them understand that there are flaws with these data sources, they still seem to rely on this data at some level. I guess they feel that bad data Is better than no data at all.
Me, I say that bad data is dangerous and is in fact worse than no data. Bad data is harmful in situations where decisions are made based on it. Unfortunately, services such as Compete and Alexa are also effective marketers and lead the average user to believe that the data is accurate when in fact, they should know better.
Click for other posts about understanding your website analytics data.
Post your findings to comments below!