The value of eye tracking in user research

I usually use a very simple example to explain the value of eye tracking in user experience research compared to traditional methods like observation. In this example I use a user test on a door!

door1Let’s say I am making a new door and I want to test the usability of that door. To open the door you need to pull the handle. So I want to do a small user study to see whether or not people understand how to open the door. I invite three typical users to take part in my test and I ask them to use the door while I’m observing their behavior. The first user pulls the door and manages to get to the other side without any problem.



door2The next user tries to push to get through the door so it does not open. And the third one pulls and gets through. We have in other words found a usability problem; one out of three did not understand that the door needs to be pulled to open.





door3So we think we can easily solve the problem by adding a PULL sign on the door. But I invite one more user to test if the sign actually had a positive impact on the behavior. But even though the sign is on the door this user still tries to push to get through. We had found the problem and we thought we had solved it by putting the sign on the door, but still we notice that something is wrong. Why does the user not pull the door even though there is a clear instruction?



door4This is where eye tracking comes in to give more information on the actual behavior which is hard to notice by just using observations. So we look at the eye tracking data from the test and notice that the natural behavior when opening a door is to focus on the handle which can be seen in the heat map to the right. Thus the pull sign we put on the door is missed as it is placed in the wrong location were users don’t look.



door5So to really solve the problem and make the door more user friendly we simply need to move the pull sign to an area where people are looking while solving a task like this.






This very basic example shows clearly the value of eye tracking. Eye tracking can reveal the actual behavior which is hard to detect by just observing, eye tracking can also create visual illustrations of the behavior, like a heat map, which describe the users behavior instantly, and last but not least eye tracking data can give a valuable insight into how to solve a problem based on the actual behavior of users. Like in this case by placing the sign where people are actually looking while solving the task.


3 comments to The value of eye tracking in user research

  • Tommy, it’s a good simple example. I assume you followed the dialogue with Jared Spool last week. I think Jared would still object that the same result could be obtained by asking the participants whether they saw the sign. It seems that the main objection about the value of ET is that it does not provide incremental benefit over traditional talk aloud. How would you address that problem in your example?

  • admin

    A lot of our behavior is unconscious, maybe in my simple door example some participants would say they looked at the handle, others not. But in a more complex environment where the goal is to solve a task on a web site, we focus on solving the task, not thinking about HOW we solve it or WHERE we look. If we want to make users think about how they solved it they will have to think aloud or do a retrospective interview. This will however not necessary make them remember where they looked. The two other benefits I point out is visualizing the behavior, this is not possible in an interview, telling the users’ story in an appealing way is also very important as a usability problem that is not solved is still a problem. And the third benefit is getting insight into how to solve the problem, in this example I can very easily spot how the problem can be solved by just looking at the heat map i.e. I need to move the sign down towards the handle. I’ll post some real examples later on this week…

  • Hi Tommy, Don Norman used pretty much the same example you have used here in his excellent book The Design of Everyday Things, without the benefit of eyetracking. As Nick notes you need to do better than this.

    The case of eyetracking is SIMPLER than the point you make: Talk Out Loud Protocol causes the user to behave differently to the way they normally do. Post Experience Eyetracked Protocol (PEEP) addresses this problem. The facts of the case are well documented in a paper I co-authored: