2008 has been the year of web analytics. In the world of “Web 2.0,” we’ve read a literal flood of stats, analytics and reports. Articles written about everything from statistics in social media to website traffic and usage projections for 2009 and beyond. Stats, numbers and more stats everywhere. What do they mean? Should we believe them? I’d like to put it out there that you really need to take any analytics stat with a grain of salt. Here’s why…
Web Analytics Tracking Differ Greatly from One Program to Another
Google Analytics, Hitwise, Omniture, Woopra, Urchin, the list of proprietary tracking programs goes on and on and on. Every program tracks differently from one another, which is why a lot of professional web analysts will tell you that they use two or three different programs when they cite a problem. While some platforms have more inherent trust for writers to cite stats, there is little to no standardization across the board. Additionally, many tracking platforms require customization so that all of the data they track is “good” data. (Typically, customized so they ignore your interaction with your own website, block spam visits, etc.) In some cases, two tracking programs that are utilized on the same service could report a twenty to thirty percent difference in numbers.
Even though there is a Web Analytics Association, the way that analytics platforms define a “visit” may also vary. Sometimes, the same data point could be referenced using different words. This makes web analytics very challenging to pinpoint for people that are new to the field.
Web Analytics Data Does Not Have a One-to-One Relationship
What does web analytics measure? Just a visit? I argue that web analytics really measures human behavior on any given website or page. That one visit could be comprised of several pieces of data: point of origin, browser, platform, time on site, click path behavior, etc. So a simple “visit” could have six or seven data point components to it that may or may not be tracked each and every step of the way, depending upon the platform.
Should you spend hours tying click path from one user to another? Or how about tracking down misspellings of individual keywords? Web analytics is not like financial analysis. It won’t always add up and it wasn’t created “to” add up. One “mistake” here or there (i.e. lost data) isn’t as important as the overall trend of human behavior, which is why multiple platforms are so essential to understanding “why” things go wrong. Sometimes, there could be major errors in something as simple as a little tracking code. Other times, there could be no errors at all and you might find that you were looking at the data the wrong way. The bottom line, is that web analytics may be accurate, but it is not a “precise” science.
Web Analytics Comparisons Don’t Always Compare Like with Like
What do I mean by saying that web analytics doesn’t always compare “like with like?” Well, for example a lot of the social media sites talk about the activity on that platform or about which social media app is more popular like this one, Report Says Twitter Would Take 36 Years to Catch Facebook – If Facebook Stopped Growing Today.
I have a really hard time with articles like these because even if the data is accurate the question is: Why are apples and oranges being compared? Twitter is a microblogging service, where you can share links, interact with other people on Twitter and talk about your day. Even the Direct Message function has a limited character count. Many marketers, writers and other professionals use it to push little bits of content out there, too — like sales or writing opportunities.
Facebook, on the other hand, can be used in any number of ways. It can be used for advertising, to aggregate your existing content (including Twitter, btw), email, play games, connect with other people through a one-on-one or group basis, and the list goes on and on and on.
Twitter does not equal Facebook, MySpace does not equal Facebook either, but the comparisons keep flooding in. Do not hang your hat on these reports and statistics, please. Human behavior is a moving target, especially on the web. Web analytics can, does and will provide excellent insight, but sometimes we need to understand what is being compared and why. Instead, look at what tools are useful for you and use those. Just because one social media app is more popular than another, doesn’t mean that it may be useful for you. When a platform is deemed to be popular (like Facebook), it just means that the application in question may have more uses for more people. Hence, more visitors.
Also, depending upon what platforms are used the comparison data could have a larger margin of error. In financial analysis, the common denominator is a dollar, right? As I mentioned earlier in this post, even though the common denominator is a visit
The More Web Data an Analytics Platform Takes In, the More Room for Error
Ever use an Excel spreadsheet with 20,000 rows? Remember how easy it is to miss something? Web analytics should have a system of checks and balances, just like financial analysis does, but it’s really difficult. Especially since a “one-to-one” relationship is next to impossible. The more data you have — the bigger the room for human error. Even if every web analytics platform reported the same data in the same manner, when you have a site that has millions of visits every day there is bound to be a margin of error that may occur due to a technical glitch or due to a human error.
In other types of scientific analysis, this margin-of-error is typically cited as part of the report or article. Unfortunately, of the many “white papers” I’ve read it seems like there are no standards for citing web analytics notations.
Learn More about Web Analytics
Remember, that the field of web analytics is still pretty new. Two people that I recommend reading Avinash Kaushik and attending a seminar to watch both him and Matt Bailey from SiteLogic speak regarding web analytics.
You can also read a lot about web analytics through the platforms that you may be using. Here is the Google Analytics blog and the Omniture and Web Analytics for example. Last but not least, you can also read about web analytics on several news sites including: MarketingProfs and Search Engine Land.
Thoughts? Questions? Post ’em here!