As described in the previous section, usability testing is a key part of the design process, but testing design prototypes is only one part of the picture. Usability assessment should begin in the early stages of product development, where you can use it to gather data about how users do their work. You then roll your findings back into the design process. As the design progresses, usability assessment continues to provide valuable input for analyzing initial design concepts and, in the later stages of product development, can be used to test specific product tasks. Apply usability assessment early and often.
Consider the user's entire experience as part of a product's usability. The usability assessment should include all of a product's components. A software interface is more than just what shows up on the screen or in the documentation.
Usability testing involves a wide range of techniques and investment of resources, including trained specialists working in sound-proofed labs with one-way mirrors and sophisticated recording equipment. However, even the simplest investment of an office or conference room, tape recorder, stopwatch, and notepad can produce benefits. Similarly, all tests need not involve great numbers of subjects. More typically, quick, iterative tests with a small, well-targeted sample, 6-10 participants, can identify 80 to 90 percent of most design problems.
Like the design process itself, usability testing begins with defining the target audience and test goals. When designing a test, focus on tasks — not features. Even if your goal is testing specific features, remember that your customers will use them within the context of particular tasks. It is also a good idea to run a pilot test to work out the bugs of the tasks to be tested and make certain the task scenarios, prototype, and equipment work smoothly.
When conducting the usability test, provide an environment comparable to the target setting; usually a quiet location, free from distractions, is best. Make participants feel comfortable. Unless you have participated yourself, you may be surprised by the pressure many test participants feel. You can alleviate some pressure by explaining the testing process and equipment to the participants, and stating your objective in testing the software and not them; if they become confused or frustrated, it is not a reflection upon them.
Allow the user reasonable time to try and work through any difficult situations. Although it is generally best to not interrupt participants during a test, they may get stuck or end up in situations that require intervention. This need not necessarily disqualify the test data, as long as the test coordinator carefully guides or hints around a problem. Give general hints before moving to specific advice. For more difficult situations, you may need to stop the test and make adjustments. Keep in mind that less intervention usually yields better results. Always record the techniques and search patterns that users employ when attempting to work through a difficulty, and the number and type of hints you have to provide.
Ask subjects to think aloud as they work, so you can hear what assumptions and inferences they are making. As the participants work, record the time they take to perform a task as well as any problems they encounter. You may also want to follow up the session with a questionnaire that asks the participants to evaluate the product or tasks they performed.
Record the test results using a portable tape recorder, or better, a video camera. Since even the best observer can miss details, reviewing the data later will prove invaluable. Recorded data also allows more direct comparisons between multiple participants. It is usually risky to base conclusions on observing a single subject. Recorded data also allows all the design team to review and evaluate the results.
Whenever possible, involve all members of the design team in observing the test and reviewing the results. This ensures a common reference point and better design solutions as team members apply their own insights to what they observe. If direct observation is not possible, make the recorded results available to the entire team.
There are many techniques you can use to gather usability information. In addition to those already mentioned, "focus groups" are helpful for generating initial ideas or trying out ideas. A focus group requires a moderator who directs the discussion about aspects of a task or design, but allows participants to freely express their opinions. You can also conduct demonstrations, or "walkthroughs," in which you take the user through a set of sample scenarios and ask about their impressions along the way. In a so-called "Wizard of Oz" technique, a testing specialist simulates the interaction of an interface. Although these latter techniques can be valuable, they often require a trained, experienced test coordinator.