A way to the perfect design

Analyzing involuntary eye movements to create a customized design

Human eyes reveal measurably different reactions, depending on how much the individuals like what they see. A research study shows it’s possible to interpret subjects’ reactions in order to offer them perfectly customized products in a matter of seconds.

When the moment finally arrives to unpack the Christmas presents and long-held wishes have become reality, you can literally see the joy in the recipient’s eyes. They seem to light up.

In fact, our eyes are very good at revealing what we like, and not just when we receive an absolutely wonderful gift. Even when it comes to more mundane matters, we tend to spend longer looking at things we find nice than at other items. When that happens, our pupils dilate a little. Researchers at the Siemens Research Unit Technology are currently investigating how these involuntary eye reactions can be utilized to provide users with perfectly customized products.

Study looks for the T-shirt with the most appeal

“In our first study, we developed a procedure to enable users to design the T-shirt that’s perfect for them,” explains Theo Papadopoulos from Siemens Technology. “Users can choose between three different materials, eight different colors, four different designs for the printed logo, and six different color options for the logo. That means there’s a total of 576 (3∙8∙4∙6) different designs to choose from. Using our approach, the process takes less than a minute.”


Papadopoulos demonstrates how his trial is structured: A standard laptop and an eye tracker – a small commercial device about as long as a ballpoint pen and twice as thick, which records eye movements but doesn’t store facial data. If you want to pick a T-shirt, you first see an image showing all the available materials. “The eye tracker records how the viewer responds to the different patterns. For example, it records how long you look at an image, and how often. We evaluate this data, and in a few seconds we know, with the greatest probability, which of the options the viewer will like most.” The color and other design features of the T-shirt are then determined in the same way.


83 percent very satisfied

Among the participants in the T-shirt study, this eye-tracking procedure was remarkably successful: 83 percent were very satisfied with their new T-shirt, another ten percent were somewhat satisfied, and only seven percent were not satisfied. “For a procedure in commercial use, the customers who are not satisfied or only somewhat satisfied must, of course, have the opportunity to actively improve the selection that was made based on their eye movements,” says Dirk Hartmann, also from Siemens Technology. “But for our study, an 83 percent hit rate is a great result that shows it’s worthwhile taking the project further."

Potential for every customized product

This procedure offers lots of opportunities, even if Siemens isn’t likely to be selling T-shirts any time in the future. “We see potential wherever customers get to make an individual choice and there’s the risk that giving them too many different options and parameters will make them feel overloaded rather than enriched,” says Papadopoulos. “We also see opportunities in assisting the process of automated design engineering in this way. Automatically generated designs can sometimes look weird – even if, in mathematical terms, they represent an ideal form that’s particularly stable, for example. Using our generative design method, we can automatically create designs that people will find appealing. 

In our next study in this area, we want to invite employees from Technology to help shape the interior architectural design of our new research location, which is being built in Garching, near the Technical University of Munich. The participants will use eye tracking to create their ideal working environment.” That will make working there almost as exciting as Christmas.

Aenne Barnard, December 2020

Subscribe to our Newsletter

Stay up to date at all times: everything you need to know about electrification, automation, and digitalization.