Improving User Performance

by Dr. Bob Bailey

July, 2000

 

A recent task taxonomy of browser use concludes download time and scrolling are two major problems that developers need to address.

Michael Byrne, Bonnie John and Neil Wehrle at Carnegie Mellon University and David Crow of the Trilogy Development Group conducted a study to identify the major activities performed by web users. The researchers attempted to gain a clearer understanding of the tasks users performed most while browsing the web, and to determine the relative time spent carrying-out each of the major activities.

They found that most (58%) of the time, users were involved in “using information.” This included reading, printing and downloading information. Another 12% of the time, users were visually searching for items on pages. Thus, for about 70% of the total time, users were directly involved in interacting with the web pages. In addition, users spent about 13% of the total time providing requested information (e.g., filling out forms), and another 5% responding to requests (e.g., providing a file name for downloading).

Possibly of greater importance, most of the remaining time was spent either waiting or scrolling. Unfortunately, over half of the time involved in moving to other pages was spent waiting for pages to load (using fast T3 network connections). Also, when using the websites, users were forced to do considerable scrolling. When combined with “waiting,” users spent about 1 hour and 27 minutes of the 5 hours either waiting or scrolling.

Benjamin Blinn and David Biers at the University of Dayton conducted a recent study on searching performance when using three different formats. They compared a paper-based document, a computer-based document that required paging, and a computer-based document that required scrolling.

The researchers found that:

(a) people searched fastest using the paper-based document, and

(b) when searching websites with shorter pages, paging enabled users to find information in reliably less time than did scrolling.

Conclusions

What can we learn from these studies? First, to expedite the use of websites, we should design pages that load fast. I continue to be unimpressed with the overuse of meaningless graphics, and totally impressed with well-designed text-oriented pages that load quickly. With today's higher bandwidths, the goal for loading pages should be two seconds or less per page. The old idea that 8 or 10 seconds represents acceptable response times is not true. We have known for years, that several seconds is much too slow for many, if not most, serious computer interactions.

Second, to improve use of websites, we should attempt to reduce the amount of required scrolling.

 

References

Blinn, B. and Biers, D.W., Searching hard-copy (paper) vs. electronic (CRT) documents: Role of experience, amount of text displayed, and the book metaphor, Proceedings of the Human Factors and Ergonomics Society - 1999, 447-451 (1999).

Byren, M.D., John, B.E., Wehrle, N.S. and Crow, D.C., The tangled web we wove: A taskonomy of WWW use, CHI 99 Conference Proceedings, 544-551(1999).