Conference Report: Usability and Web Site Success - Patterns That Indicate User Failure
(Page 3 of 4 )
According to Jared Spool, "The biggest problem with the design process is that not only do we not learn from our mistakes, we don't have a mechanism for finding out what our mistakes are. Amazon, Ebay and Lands' End all have a great feedback loop built in to their system. But you still have to know what to look for."
Lands' End, for instance, relies heavily on filtered feedback from their customer service reps to decide on ongoing improvements.
To help developers know what to look for, UIE has identified some common patterns that indicate user frustration or failure, including:
- Pogo-sticking
- Back button usage (doesn't show up in server logs!)
- Search behavior (requires that you look at the page from which the search was initiated and the search terms used)
Metrics & Analytics - How Do We Track User Success Or Failure? Nearly all of the Web log analysis programs on the market today focus on hits, page views and user sessions. These programs do not analyze the actual content of the pages being viewed. UIE said that site developers and managers simply don't have the measurement tools right now to tell them if users are actually succeeding or failing on their Web site.
Ed Chi, Ph.D. of Xerox PARC, made a presentation on his progress developing software tools which can:
- analyze how users find their way through a Web site
- predict their most probable paths and likelihood of success based on given objectives
- analyze logs along with page content to essentially reverse engineer the objectives of users, and then to gain some idea of their success or failure.
The webmasters in the crowd were drooling over the thought of having such powerful analytics tools. Dr. Chi was hopeful that these tools will see daylight in commercial form sometime in the not-too-distant future.
[Read
Dr Chi's papers on information scent, intelligent analysis of Web traffic and categorizing user sessions.]
Design Guidelines - Which Ones Should You Follow? Jared said that the first thing reporters always ask him is "What are the five things Web designers should not do?"
Jared's typical response is "Well, uh ... cocaine."
Seriously, rules are comfortable. They offer us some assurance that our work will succeed.
Of course, pundits have provided countless guidelines for how sites should be designed. But are these guidelines based in opinion or research? Will Schroeder of UIE explained that we often accept the guidelines on faith because verifying them takes time and money.
Fortunately, UIE has been testing the more common guidelines against their research data. Schroeder emphasized that their data set is not large enough to prove a guideline, but it is large enough to disprove one. They have found that some guidelines help, some make no difference, and some can actually hurt.
Flash Usability Guidelines UIE hosted a panel discussion on Flash usability guidelines, with expert Flash developers Chris MacGregor and Josh On, Eric Pressman, Flash usability engineer for Macromedia, and Christine Perfetti, author of UIE's study on Flash usability. They discussed the need for Flash usability guidelines, how and whether they should be developed, and how to determine if the guidelines are successful.
Flash has developed something of a bad reputation for usability.
"Blaming the tool [Flash] is sort of like blaming VCRs for Jim Carrey movies," Spool quipped during the forum.
Chris MacGregor explained the early days of Flash development by saying, "Web designers' sites needed a big sledgehammer to show clients how cool they were. Flash was just the biggest sledgehammer available."
Eric Pressman said that, "Macromedia's position has not been so much to tell developers how they should do their work, but to give them good examples of things they can do, or the way things should be done."
We did see some exciting examples of useful Flash sites: one for
information visualization, and one for
hotel reservations.
Next: Conclusion >>
More HTML Articles
More By David Poteet
/ 3 
