Strategic
Server-Log Analysis
Issue addressed
"With all of this information we have about page-loads and click-streams, there must be something we can use to diagnose design issues that are compromising our site's effectiveness”
The challenge: Leverage existing data to make your site more effective
Site logs and click-streams provide rich archival data of site usage that hold enormous potential for understanding and improving upon the experience of visitors using your site.
Anyone can collect site log files, but it is difficult to use them effectively to evaluate and improve the effectiveness of your site through interactive design.
What is required is a strategy for tying your business goals to their statistical implications for log-file data, and then using this information to evaluate your site's effectiveness and to improve on its design where necessary.
There are indeed e-business intelligence solutions that solve the technical challenge of capturing and aggregating rich and accurate data sets. When used appropropriately, they are powerful in identifying what isn't working in your site, but they cannot answer why those components aren't working.
I provide the why -- and what to do about it.
I work with your technical and marketing teams to set measures of your visitors' online behavior that are meaningful to your business and feasible within your infrastructure. We work with baseline data to develop success criteria for your various business goals. At key milestones, data are gathered to evaluate the effectiveness of your web site on these criteria.
When large gaps are identified between desired and actual effectiveness, I add significant value to the process by conducting the most appropriate type of usability and user-experience evaluation on the identified offending scenarios.
The result is a highly targetted focus on the issues that have been identified as compromising your site's effectiveness.
With this strategic approach to site design and evolution, new or redesigned components can be "soft-launched" as a test-bed and their full effects evaluated in site-logs before the redesign is committed for general release. A typical example is deciding whether or not to include a featured product link at a certain location in the site: does it increase business or just tend to derail visitors from other important activities?
I know how to work with large sets of behavioral-data, I know statistics, I know how to interpret the meaning that can and cannot be derived on the basis of numbers alone, and I know interactive design.
I also have a long history in scientific study of dynamic serial behavior (on the web, click-streams), which is concerned with identifying underlying processes behind statistical patterns in human behavior.
- My work started with my Ph.D. dissertation in the late 1980s, when I gathered and analyzed serial eye-movement data from military pilots. In 1990, I received a major award from the
Human Factors and Ergonomics Society for this work.
- When I was a staff member in IBM's User-Centered Design group, I developed and validated techniques for using trace-logs from instrumented user-interfaces to identify usability issues. I did this work in both usability laboratory testing (see
- In 1999, I presented a paper at the
- As a statistics professor and research psychologist who was also involved in developing data warehousing solutions, I love to make large datasets tell their stories through advanced statistical analysis.
I look forward to applying my knowledge and passion in this area to identifying ways to improve your own site's effectiveness.
Last update: May 07, 2002
(c) 2002 Don Hameluck Usability Consulting Inc.
