The Japanese roboticist Masahiro Mori introduced the concept of the “uncanny valley” in which people get turned off by robotic reprodutions of humans that look all too human. You’ve probably squirmed a bit (or been delighted, in a technological sense) seeing CGI at the movies that seemed a little bit too real. It turns out that this “uncanny valley” is the place that filmmakers don’t want to go, because the audience won’t pay enough attention to the story. (It will be interesting to see how this plays out with the movie, Avatar.)

It turns out that in User Experience Design, we have our own uncanny valley. Since we’re usually creating the front end before there’s a back end, we find ourselves in a situation where we want to try it out on users—see how well the design works before it gets coded. This is where we can get into trouble. If we have sketches—a paper prototype—participants in any study can see that it’s not real. But when we create a prototype with the intent of testing usability, we invariably start thinking about the economics of the situation:

  • we’re paying the participants a lot of money,
  • it’s time consuming and expensive to recruit and schedule them
  • it’s a pain to get the development team to attend the sessions, not to mention the time that the team won’t be spending to produce the product

Taking all of this into account, usability studies aren’t cheap. So when we do get the chance to run a usability study, our reaction is to get as much out of the study as we can. After all, we have lots of questions that can be answered by a study, so we want the prototype to be as robust as possible, right? So we make a prototype that looks as realistic as possible, including a new visual design.

Enter the Uncanny Valley. Your prototype is all hooked up—since you’re using HTML and CSS. Naturally you’ve put some effort into visual design, because you don’t want participants (or anyone observing the study) to react by saying that the “new” design looks worse than the existing design.

But there’s a problem. This prototype looks like a finished product. Users in your usability study start clicking on things while they try to complete the tasks you’ve laid out for them and quickly find out that some of the things don’t work. At first they’re a little surprised, because they weren’t quite prepared for the reality that you’d just given them a “prototype” instead of a “product” (or, as Mori would say: a “robot” instead of a “person”).

The surprise quickly changes to disappointment and frustration as your participants start assuming that most of the prototype doesn’t work. Even though that may not be true. So the moderator finds himself explaining more than he’d like, asking the participant to “go ahead and try clicking on it” when the participant hesitates and sighs with his cursor hovering over a button. A skilled moderator can guide participants through the Valley like this, but often after experiencing this frustration, the participant proclaims at the end of the study that the old design was much better than this new one. It seems easy for the development team to conclude that the new design is a failure. So is it possible to avoid the Uncanny Valley altogether?

Yes. Here’s how to avoid the Uncanny Valley when you create a prototype to show in a usability study:

A prototype of a user experience that has been prepared for usability testing. Notice the use of a handwriting font and sloppy visual style for the aspects of the user experience we want users to ignore.

A prototype of a user experience that has been prepared for usability testing. Notice the use of a sketchy  font and sloppy visual design for the aspects of the user experience that we want users to ignore.

Go ahead and create the prototype. But skip the new visual design—save that for later. In fact, it’s better for some parts of your prototype to look decidedly unfinished. Go out of your way to make some parts of the prototype to look sketchy or incomplete. Make the widgets look flat or sketchy – it’s alright, in fact, better for it to look like something you’ve cobbled together. Participants will actually react more positively to it. They’ll forgive the user experience for not reacting to them because their expectations are lower—they know they’re dealing with a “robot”. They may even try fiddling with more of your prototype, especially if you’ve given a more finished visual appearance to the parts that do work.

If you want to test the visual design you’ve been working on, then do it at the end of the study. Show screenshots (or even the same prototype with slick CSS this time) to the participants. When you ask the question, “how do you like the new design compared to the old one?” You’ll find they cut your little robot a little more slack and say things like “I’d love to see it when it’s finished!”