Improving Web Usability Through Visualization

Predictive Web usage visualizations can help analysts uncover traffic patterns and usability problems. D espite major efforts in Web design research, many sites remain difficult to use. In a recent study of e-commerce sites, for example, users successfully performed only 56 percent of their intended tasks. 1 Forrester Research has reported that 65 percent of all online shopping trips end in failure, 2 and that 40 percent of all visitors choose not to return to a site because of design problems. 3 Clearly, usability can make or break a Web site. One reason for poor Web site usabili-ty is researchers' lack of understanding about how users surf for information. The field of Web analytics, which is currently a billion-dollar-a-year industry, arose to address such usability issues. Web analytics attempts to increase usability and visitor retention through examination of server log data. A variety of tools extract metrics from these files, such as the number of unique users, page visits, and session lengths. While some of these tools have evolved into products, administrators often must sift through large amounts of simple statistics that provide little insight into user behavior and traffic patterns. Moreover , analysis efforts are complicated by the fact that a site's usage, as well as its content and structure, change greatly over time. In the past decade, some researchers have explored visualization methods to help understand usage data and identify major traffic patterns. In this article, I describe some of the techniques we in the User Interface Research Group at PARC have developed to visualize content changes, linkage structures, site usage, and so on. I explore how these techniques can be used to identify specific usability problems at large Web sites. As part of our work, we have created a predictive visualization model called Information Scent, in which simulated users help uncover patterns and deficiencies in information accessibility. We demonstrate these techniques using a prototype system called ScentViz to illustrate how Web usage analysis can be enhanced using visual-ization methods. Recent advances, from the data mining movement to new statistical techniques in natural language processing and information visualization, have made it easier to manage Web data analysis. Because they help combat complexity in understanding user behavior, these techniques will remain important elements in every Web analyst's toolbox. Nonetheless, visualizing Web usage data to improve usability presents several challenges. Data Mining Effective visualization requires sophisticated data analysis and data …