 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Site Log Analysis Case
Study
Introduction
Published Spring 2001
Tools Used
- WebTrends: report generation
- Excel spreadsheet: analysis tool
- Microsoft Word: recoring insights &random notes
- PowerPoint: findings presentation to client
Background
A large consumer packaged goods company retained my employer to provide ongoing maintenance and support for a small marketing-oriented site for a consumables product. One day in April, one of the Client Services staff assigned to the client asked me to take a look at the server logs. She wanted me to see if I could make recommendations for the upcoming major site overhaul.
I should add that our company inherited the client's site from a competitor who had originally built it. So we had no responsibility for the (in my opinion) poor information architecture and poor choices in the technical implementation of the site!
Approach
I
spent about a week's worth of time altogether on the project. The first
part was spent familiarizing myself with the site's content, cicking around
the site to see both what was there but also how well it worked.
The site used complex frames to maintain the desired look and feel. So analysis of the server logs also required familiarity with the way the site was implemented: framesets, file names, cookies, and the homebaked CGI system used for serving content.
The client's hosting company offered monthly web site usage reports using a server-based WebTrends solution with default settings. I needed at least a quarterly report to make out some short-term trends. Fortunately, a co-worker in the New York office offered to help out by putting together a spreadsheet with site usage stats from the entire first quarter. It featured monthly figures, quarterly sums, and averages for several areas. It was tedious work to put the spreadsheet together, but it helped immensely!
Under
the Microscope
It
was fascinating to look through the stats, trying to figure out the connection
between content (its inherent nature as well as how it was chunked), navigation,
technology, user interface, and user characteristics. In addition, it
was important to know the offline context of the client's marketing and
communications activities. They are all deeply entwined. As a result,
the mystery is to attempt to figure out which are the stronger factors
causing undesirable user behaviors on the site.
Some general metrics to look at, generalizable for all sites, include the following:
- Unique user sessions
- Page views
- Average session length
- Sources of referring URLs; that is, where is the user before he or she clicked on a link and ended up at your site?
- Rates of return visitors
- Keywords used on search engines by which users found the site
- And though they won't be recorded in your server logs, look at the site registration rates, if applicable
Other Links in this Case Study
- Case Study Introduction (this page)
- Lessons Learned
- Tips for Customizing Server Log Report Tools