The next step is to watch real users using your site. You might be surprised by the way they react to your information and layout. If you want to understand the user experience, there's no substitute for actually watching someone try to achieve something on your site.
It's up to you what you ask your test users to do. You might want to see how they use your current site, or test some suggested improvements to the site, or compare and contrast the two. You could also ask users to come back and re-test the site when you've improved it based on previous test results.
In all cases, the best approach is to set real tasks and observe behaviour. For instance, if you were testing the Student Services Centre website, you might ask a user to find a letter for their bank proving that they are registered as a student at LSE.
You can invite users by email; but plan ahead to give them enough notice. You'll need to invite many more people than you need, as only a small percentage are likely to volunteer. You could ask them to visit you, where you can set things up as you want and where there's plenty of privacy. Sometimes, though, it's easier to go to their desks. Try to offer a small incentive which would appeal to the user group, such as a voucher.
It's useful if you can have two or three staff at the test: one or two observers who write down everything the user does, the links they choose, the decisions they make etc; and one facilitator to prompt the user to think out loud. You need to know what the user is thinking - whether anything has confused them, why they chose a certain link, how confident they are that they'll find what they're looking for, etc. When they find the answer, are they confident that it is the correct/ complete answer?
If possible the facilitator should be unconnected to the particular website, so they can be objective. For observers, try to use a variety of staff from the department or team - it can be a real eye- opener!
You could also record tests. This can help when trying to interpret observers' notes, and means you can show the tests to colleagues. Audio recording is easier but visual would be ideal.
Guidelines and script
You can base the tasks on common queries your team receives; core areas of your service that you know are important; and areas of your site that you suspect are not being found, or are problematic for some reason. Try to provide realistic tasks - something the user would actually do. Don't use jargon, or words you know are used in your site - you need to avoid pushing the user in any particular direction on the site, even subconsciously. Add a little context to the task by creating short scenarios, but keep them simple.
We suggest setting ten such tasks for each user. You may not have time to cover them all, so prioritise them - the time taken over the test can vary depending on how chatty the user is and how easy they find the tasks. Tests usually take about 45 minutes, including greetings, any documentation, instructions, and questions.
Three to five users are usually enough. They should find the major problem areas in your site, allowing you to fix those before doing more testing with other users. This way, you won't have to listen to lots of users telling you about the problems you've already uncovered. Instead you can discover other issues that might have been blocked or hidden by the larger problems, and keep improving. It's up to you how many iterations you want to work on.
It's important to emphasise to the user that you are not testing them, but the site. The user can't do anything wrong, or fail the test. If they can't find something, it's the fault of the site, and they've helped you out by highlighting the problem.
You should expect to be surprised by the test results. Areas you think are easy to find might not be obvious to the users; phrases that mean something to you might mean something else to the user. Not everyone uses the web in the same way. Some like to use the search facility, some prefer to browse; some read all the instructions and text, some skip straight to a useful-looking link.
Once you have your test questions or tasks set up, you can re-use them at will. For instance, you could ask users to try them out on other universities' websites and see how easy it is to achieve the same tasks there. Or you could give them to new members of staff who aren't yet familiar with your website, and see what they make of it.
Next section: What to do with your results
^ Back to top