Define the goals of the usability test.
- What are you trying to confirm or find out?
- Focus on a specific area of interest for each test. It's better to do multiple tests and get good results than to do one test that covers too broad a range of topics.
Design the test to meet the goals.
- Can be online or paper prototype.
- Ask open-ended questions that do not influence the user's actions.
- Create scenarios using a test script.
- Ask only questions that address the goals. Don't be tempted to insert other questions that provide useful information, but aren't on topic.
- Try not to build questions that depend on the prior questions. This makes it hard to track and analyze if the participant can't find the answer, or is someplace that's totally wrong (to say nothing of embarrassing the participant.)
- Try not to give too much information in framing the scenario.
- Put general questions at the end of the test.
- Design a test that will take about an hour to complete ― no more than 20 scenarios.
Decide on format of the test: group or individual.
Individual testing is sometimes easier to do because it requires fewer concurrent resources.
- Requires only one PC (if online) and one person to administer and take notes.
- Takes more elapsed time than a group test.
Group testing can be completed more quickly but requires multiple resources.
- Requires a room with multiple PCs (if online) spread out so that testers do not interfere with each other.
- Requires one person to administer the test.
- Requires multiple notetakers, one for each tester.
Select participants for the usability test.
- Plan on 5-7 participants for each test. Invite 10-12, assuming a 50% non-acceptance.
- Invite people who are not expert on websites, such as webmasters, or those who are expert on your site already. (We've found they have preconceived ideas.)
Conducting the test
- Introduce yourself.
- Make participants feel comfortable.
- Be clear on the objectives of the test, how the test will be conducted and what participants need to do.
- Have another person present to record comments.
- See handout for tips on talking with participants.
- Be careful what you say; don't influence either during test or during discussion.
- Discussions should occur only after the test (or after each question) depending on test.
- Have a concrete method of capturing answers.
- Have a few open questions at the end for discussion.
- Thank everyone for participating.
- Consolidate results without analyzing.
- Analyze compilation.
- Draw conclusions. It's possible that you might be able to infer several different conclusions, which may signal the need for additional usability testing.
Society for Technical Communication (STC) Usability Special Interest Group, http://www.stcsig.org/usability/