With user testing, our goal is usually the same: to identify areas on an interface for improvement, by observing users’ natural behavior with that interface. But how can we expect to observe natural behaviour when the user is sitting in a user research laboratory, watched by a researcher who is holding a clipboard? Surely the user is not used to this setup, and hence will not behave naturally. Our test results are spoiled, right? Not quite. 

Granted, the ideal user test study would consist of me, the researcher, being invisible and silently sneaking into people’s homes to peer over their shoulders as they check Facebook, purchase items from eBay, or watch videos on YouTube. Unfortunately this ‘invisibility technology’ is not available yet, although they are working on it. 

In the meantime, there are a number of things you can do to optimise the naturalness of test participants’ behaviour in the lab:

1) Never user test your own designs. Bias is difficult to consciously control. You may be leading the participants through the test without realising it. You want to avoid taking your participants on a hand-held guided tour of the interface. If you do end up testing your own designs, try your best to censor any tendencies to rationalise or defend any design decisions. Resist the urge to help the user, even if they are completely stuck. Accept that the design has failed for this task. Move on.

2) Shut up. When a user is completing a task, stay quiet. Its an observation, not an interview. Questions will interrupt your users thought process and behaviour. Where possible, leave questions until after the test. (This isn’t always avoidable, sometimes its best to ask questions).

3) Put your users at ease by providing an informal and relaxed atmosphere.

4) If resources are available, and it suits the test, allow for a retrospective protocol. A concurrent protocol will give good insight, but when we ask users to think aloud during a test, we run the danger of interrupting their natural behaviour.

5) Where possible, leave the room during the test. If your users are on their own, they will behave more naturally. This may not be suitable for all test designs. The facilitator is very often required to be present.

6) Do what ever is necessary to elicit natural behaviour. If the site you are testing is an e-commerce website, and you want to evaluate the design of a purchase funnel, give the users a credit card to use. Even better, let them use their own, and reimburse them on the day. You will notice a huge difference in the behaviour of somebody who is pretending to make a purchase (i.e., “this is what I would do”), and somebody who is shopping for something they’re actually going to get in the post (i.e., natural behaviour).

The problem is a simple one. Unnatural behaviour from users leads to unreliable data. The resulting scenario is that we are left in a position where we might make poorly informed design decisions. When your designing a user test study, always ask yourself at every decision point, ‘will this lead to any unnatural behavior from the user?’. If the answer is ever ‘yes’, you need to revisit your test design.