What Your Web Statistics Didn't Tell You

To continue the conversation from Mo's great post last week titled 'How Many Site Hits? Depends Who’s Counting', I wanted to share some additional statistics and secrets from the world of web statistics and analytics.

Thoughts// Secret #1 - Web statistics companies don't track website visitation numbers the same way.

What? Really, they don't?

No, they don't. You could run the same exact set of log files through two different web statistics tools and come up with two completely different answers.

Think of it in terms of a search engine. Yahoo! and Google both have them, but they use slightly different formulas (industry folks like to say algorithms) to present you with the best results. More often that not, those results are slightly different, even when searching for the same word or phrase. Web statistics companies follow a similar pattern.

The Arizona Office of Tourism just recently experienced this exact situation...but using the same web statistics company. We had a set of log files and analyzed them with version 7 and version 8 of a popular web statistics tool. Keep in mind, this was software from the same exact company. The results of upgrading to the newest software version? A sizable difference in page views and unique visitors when comparing the two versions.

For another example, read the summary report from the 2007 Web Analytics Shoot Out, which produces a side by side comparison of 7 analytics packages. Web statistics tools were installed by experts on 4 different sites where they recorded the same data. The report highlighted several scenarios, but for this post let's just focus on the results of unique visitors on the CityTownInfo.com website (by web statistics company):

- Clicktracks - 645,380
- Google Analytics - 600,545
- IndexTools - 614,600
- Unica Affinium NetInsight- 607,475
- Visual Sciences HBX Analytics (or HitBox) - 524,055

As the report shows, the results are similar, but not exact.

So, what does this all mean. Well, to start you can spend less time worrying about what the other guy is reporting as his/her web visitation numbers, because unless they are using the exact same software package (and even then, implementation could be a factor) it is nearly impossible to compare statistics on an equal level.

To use web statistics effectively, you first need to have a solid depth (history) of data for comparison purposes (be sure to save those log files). Then, look beyond the raw numbers that are reported and begin to find trends in the data. Use multiple web statistics tools...WebTrends, HitBox, Google Analytics (which, FYI, is free)...to see a broader picture of the statistics relating to your site. Finally, keep up-to-date on what those trends are telling you.

While that sounds easy, unfortunately, it is not. And it would take a much longer post to explain all of the details and nuances of web analytics. However, I can recommend several great books on the subject. Including:

- Web Analytics Demystified by Eric Peterson
- Web Analytics For Dummies (basic, but good) by Pedro Sostre and Jennifer LeClaire
- Google Analytics 2.0 by by Jerri L. Ledford and Mary E. Tyler