Introduction

Why is computer system usability becoming so popular?

Some of us were interested in usability issues (also known as "human factors," "software ergonomics," etc.) when all we had was keypunched input and paper printouts for output. Few people were interested in the interfaces. Next we dealt with character-based interfaces (green screens) on moderate-sized monitors, and eventually colored interfaces on even larger monitors.

Most of the original users dealt primarily with accessing databases, while later users moved to interacting with a multitude of different applications. In most cases, the number of users was relatively small (from ten to a few hundred), and they tended to use the systems enough to become quite experienced. When users did have serious concerns about a difficult interface, the major complaints were responded to by making changes to the next software release. Having a small number of experienced users meant that a poorly designed user interface could (and often did) go unnoticed.

Times have changed. With most new systems being Web sites, hiding poor interfaces is no longer possible. For example, Randy Souza reported a revealing study in the Forrester Report (October, 2000). He conducted a study of shopping experiences using 30 well-known online retail sites. He found that not one Web site delivered both acceptable content and ease-of-use. Among other findings, he reported that only:

63% provided navigation that was available and consistent at all times,
53% had a useful search capability, and
53% had a checkout process that was fast and distraction-free.
Many popular computer systems (Web sites) are now being used by millions of people per day. Consider the following number of unique visitors reported for one month from Media Metrix:

Assume that we do usability testing and find and fix all problems that created usability difficulties for 99% of users. The remaining problems, which create difficulties for about 1% of users, is still a large (very noticeable) number:

Will effective usability tests help to improve the usability? Unfortunately, many user interface professionals still use very few test subjects when conducting performance tests. Nielsen in his March 19, 2000 Alertbox suggested that usability testers only need to test with 15 users to discover all the usability problems in the design. I suggest that to be 90% confident of finding usability problems that will affect 99% of users requires more like 112 representative test participants.

By using "Nielsen's 15 test subjects," usability professionals will detect problems that only will affect about 85% of users. This means that 15% of users still could have problems. To be more concrete, Amazon, which is used by 15,000,000 people each month, could have over 2 million people each month having difficulty using their site.

The large number of users makes poor Web site design more apparent than ever before.