Do-it-Yourself Usability Testing

- Gerry Gaffney

Many websites (and software packages in general) suffer or even fail as a result of problems that could have been identified by usability testing. Although most large organisations now conduct usability testing as a matter of course, the practice is not yet universal.

There are three primary barriers to usability testing: the perception that usability testing is highly specialised and expensive; the fear that testing may result in delays to release schedules; the belief that usability testing is unnecessary because the system or website is already "intuitive".

Although formal usability testing often takes place in well-equipped (and hence expensive) laboratories, testing can be conducted relatively cheaply, and to good effect, in a more informal setting.

Delays to schedules are indeed likely if usability testing is conducted only at the end of the development cycle. To reduce the risk of such delays, testing should take place early and often during development, beginning at the concept phase. Even if this has not occurred, it is better to discover usability issues before release and have a strategy to deal with them, than to have them discovered by the marketplace.

The belief that sites or applications do not require testing because they are "intuitive" is generally mistaken. The fact that developers and technical personnel can use a site easily is not an indication that "real people" will able to do so.

Conducting do-it-yourself usability testing is relatively straight-forward, provided you have a clear plan and are aware of potential pitfalls.

No specialised equipment is required, although video-taping can provide powerful evidence to support any recommended changes.

The secret of usability testing is to have representative users attempt typical tasks on a typical set of hardware, and to log in detail their activities and comments.

If people will access the site using dial-up modems, you should test using dial-up modems. Far too many websites contain massive graphics that look pretty during boardroom presentations but take minutes to download in the real world (by which time the intended users may well have clicked "Stop" and gone elsewhere).

If the intended users are bus drivers or mechanical engineers, testing should be done with bus drivers or mechanical engineers. If the audience is "the population at large", testing should include a reasonable cross-section, with consideration given to age, education, language and computer knowledge. However, it is not necessary to test with large numbers of people.

Using six participants will generally identify the majority of significant issues. Indeed, after the first three or four, the same issues generally start to emerge. Even a single test can reveal important issues. If you do not have direct access to representative users, there are several market research companies that can organise the appropriate people for a reasonable price.

Usability testing is a much more accurate tool than obtaining subjective feedback from customers. It is quite common for people to rate a site highly based on visual appeal, even though usability testing may reveal that users are unable to complete the tasks the site has been designed to support.

Usability testing can be quite stressful for participants, who often feel that their own performance is under scrutiny. It is important to put them at their ease. It is also important that they feel they can speak freely. Introducing an observer as "the person who developed this site" is unlikely to make a participant feel free to complain about it.

Testing can also be stressful for developers. It is difficult to spend months on an interface and then watch people conspicuously fail to use it "correctly" during testing. Conducting usability testing requires developing an outlook that sees testing as part of the design process, rather than as a critical activity with potentially dire consequences. After all, it is widely accepted that all software has bugs that need to be addressed. It is unreasonable to expect user interfaces to emerge magically bug-free from the development process. When seen in this light, testing is extremely effective at helping developers to think about users and their needs.

It can also persuade organisations that adopting a customer-centred design process has the potential to provide major benefits.

It is important that organisations are committed to acting on the results of testing. If the commitment does not exist, testing is a pointless waste of effort.

Because usability testing is so easy and inexpensive to conduct, and provides such benefit, it is simply irresponsible to neglect it. Every product undergoes usability testing - either before release or in the real world.

Pre-release testing is by far the better option.

Note: This article first appeared in The Age and Sydney Morning Herald newspapers, under the title 'Proof of the Site is in the Using'.