It's Not Web Analytics. It's People Analytics

Web site analysis in the past yielded very superficial data, or required highly complex customized databases. The cause of the problem in both cases is simply the extremely large number of variants involved in the data. Put simply, web analytics is actually people analytics; you're looking at the behavior of people at your site. We only need look around the world to see that people behave in very complex and contradictory ways. Analysis of data with such wide ranging variables is notoriously complex.

Traditional web analytics solved the problem of complexity by eliminating it, reducing the countless ways visitors move around websites into a basic list of top pages, top referrers etc. This basic list of stats provided marketers with some idea of events taking place on the site, but without any way to correlate the data derived from the referrer to, for example, a breakdown the pages seen by those referred visitors. Each set of stats lived in a separate world.

The second solution to the complexity was to develop expensive, highly customized systems that could establish the above correlations but only by loading databases full of very customized information. Since even a large and powerful computer cannot handle everything from the logfile simultaneously, the data must be optimized when loaded initially. This requires extensive custom development and programming, resulting in expensive software that needs constant tweaking and updating. Having done this the data is then loaded in such a way that it cannot be correlated in a different way. The optimization process is one-way and cannot be changed if the different data needs to be extracted.

ClickTracks combines the best of both approaches using a unique system. The default reports show behavior for the average visitor but can be modified on-the-fly to show comparative analysis of different groups or types. Thus a simple report of search engine keywords is available within seconds, and a more complex report of the keywords used by visitors that later see a coupon page can be generated in a few minutes.

Divide and conquer

The system used by ClickTracks to divide visitors is named Visitor Labeling. Internally ClickTracks will attach a colored label to a visitor if they meet a certain criterion, such as they were referred from a particular site. The label sticks with the visitor for the entire time they are part of the site analysis, so it's easy to label visitors and see if they later reach a certain page (or more importantly if one group is more likely than another to reach a certain page)

Labeling is designed to facilitate side-by-side comparisons of different groups. You'll see some reports like the site overview are formatted to make such comparisons very easy, with a column for each labeled group. Labels can in theory be used to examine a single individual visitor, though that's not what they're intended for. Examining individual visitors rarely produces statistically sound data.

Labels can also be combined together in AND and OR relationships, so intersections of different groups becomes possible.

Remember: comparisons produce the best data

Knowing the most popular page on the website seems very useful the first few times you use ClickTracks. You'll quickly discover however that this metric is rarely significant on its own. Far more useful is to examine the most popular page for two groups of visitors side-by-side. Choose these groups wisely and you'll be amazed at the depth and quality of decisions you can make.