Arve Kvaloy — Tue, 20 Nov 2007 22:37:36 GMT

This article would have been groundbreaking six to seven years ago, when advanced users of analytics switched from analyzing serverlogs to pagetaging solutions. The screenshot from good ol’ Awstats illustrates the historic value of the article.

All modern analytics solutions rely now on pagetaging: embedding a small javascript file on each page:

- Pagetagging remedies caching: I can save a pagetaged file localy on my computer, the pagetag script would still be set off when i open the page
- Robots and spiders? They dont set off pagetaging scripts so they want get counted
- Pagetags remedy proxies by placing a cookie which helps identify the user (and simple beacons for those that have javascripts turned off)

Even if the world of analytics is vastly improved from the above 2000/2001 scenario re-enacted, no serious analyst would try to suggest that you could establish real persons behind the term “Unique visitor” unless the user identified him-/herself by logging on to the site. But what analytics gives you are trends and patterns that can help you test and optimize your site.

And to some ecxtent you can infer visitor intent for example by:
- search phrases used to find your site (why do they visit the site?)
- search phrases used use in internal search (could show a lack in the navigation structure)
- combining clickstream with online surveys

A serious quantitative analysis should actually be mandatory before doing an expert review (at least on medium to large sites) so as to give the reviewer some possible problem areas to focus on instead of relying solely on iterating some heuristics, it could also give a more specific focus for user testing (analytics could point out the where and what, the user test could explain the why).

Boxes and Arrows: Comments by Arve Kvaloy