 |
|
Stout begins with the basics, showing the reader what kind of logs Web server software packages maintain, how the logs relate to each other, and what format inconsistencies to look out for when reconciling data from the various files. While he admits that much of the information in this section is elementary, the author correctly points out that reviewing the basic files (and common "consolidated" log formats which bring together the most relevant information from the raw logs) is an essential first step to making sense of the data at hand.
With the basics out of the way, Stout walks the user through a sample analysis, pointing out some of the tricks and traps inherent in analyzing incomplete records. Two potential problem areas are tracks left by users of multi-user machines and those of visitors who connect to an Internet Service Provider which assigns dynamic IP addresses to its SLIP (Serial Link Interface Protocol) and PPP (Point-to- Point Protocol) customers. The former case confounds analysis as multiple users could visit the same pages repetitively but appear as a single user coming back many times, while the latter circumstance makes a single user's visits appear as if several different individuals were viewing the site. There's no way around these difficulties, but if the logs are small enough sorting by various elements of the log file (like the IP address of the machine making the connection or the referring page) can allow webmasters to judge the likelihood of whether a set of visits represents interest from one or more users.
The latter half of the book compares the merits of log file analysis software packages, third-party auditing companies, and ad-tracking services. I can't comment on the accuracy of Stout's assessments of the packages and analysis services directly as I didn't load all of the sample software included on the book's CD, but crosstables in Appendix A listing the features of the packages he reviewed (which included several freeware and shareware programs) seem quite complete. Software and services are upgraded constantly, but Stout has established a site at http://www.websitestats.com to supplement the information in the book and keep readers current on developments in the area.
Web Site Stats is a good book, especially for relatively inexperienced webmasters or individuals who would like an overview of the means available for examining their log files, but Stout does give short shrift to users' privacy concerns regarding "cookies," files some web servers write directly onto users' hard drives. The case Stout cites is one where a Java or JavaScript programmer wrote a program to determine a user's e-mail address, embedded the information in a cookie, and was thereby able to track the user every time he or she returned to the site (not to mention potentially adding the address to lists used for unsolicited commercial e-mail). While Stout is absolutely correct, from a technical standpoint, that the problem is with Java and not cookies, in my opinion he does not sensitize his readers to the growing concern over online privacy.
Despite this drawback, Web Site Stats provides a solid foundation for Webmasters who want to get a clearer picture of who is visiting their sites, whether out of curiosity or to justify advertising dollars.
-- Curtis D. Frye (cfrye@teleport.com)
| Readability |
|
| Originality |
|
| Organization |
|
| Accuracy |
|
| Consistency |
|
| Depth |
|
| Timeliness |
|
| Editing |
|
| Design |
|
| Overall Value |
|
Explanation of ERCB rating scale: No stars = unacceptable, 1 Star = marginal, 2 Stars = average, 3 Stars = above average, 4 Stars = exceptional, 5 Stars = stellar.