Designing with AI Eyes: Artificial Intelligence and Predictive Testing Design

Designing with AI Eyes


Eye tracking is a powerful way to perform empiric, qualitative and quantitative studies in print or digital visual media. Eye tracking is a test methodology where the movement of the examinee’s eyes is recorded while he/she is looking at or interacting with a user interface, newsletter or printed advertisement. As a result, data is gathered that shows where the eye’s focus has been at any point in time, and also shows how long a certain element has been focussed on. This process provides a mapping showcasing how users “see” interfaces. 

Mostly, “heat maps”,  so called Attention Maps, or Gazeplots, are often used to visually indicate the following: 

  • Which elements does a user actually notice? 
  • What catches their attention, and for how long? 
  • How do their eyes browse when viewing an interface? 

All this can provide valuable data in order to optimise usability or conversion of user interfaces; it is also used to test and optimise ad campaigns and even shelves at the POS. This methodology does however come with a lot of challenges, these involve the complexity of the technology, a combination of tracking software and hardware such as tracking glasses. Performing eye tracking studies does not only require special experience, it also comes with a high price tag, which is why the service is often only accessible via special laboratories.



As lab-based eye tracking is not always practical, due to the time and cost factor, there have been many attempts to bring eye tracking based testing to the personal laptop, eliminating the limiting hardware factor. Yet, tools that use the device’s camera as a tracking sensor are not precise enough, and one would still need to acquire test users. Reliable heat maps that show the user's interaction with a user interface can only be found in tools like Hotjar. Yet, this too only shows the user’s mouse movement and clicking behaviour, not their cognitive attention.



By using Machine Learning and AI frameworks, algorithms have been trained using eye tracking results from actual tests in lab environments. The AI was fed with the original designs and the individual results of a large number of eye tracking data. This way, the machine was able to learn how certain layouts, sizes, colours and spacings affect human attention and reaction — it was able to predict the outcome of an eye tracking test surprisingly precisely. 

Simulated eye tracking studies and preference tests are up to 93% accurate in their predictions. 

For us, this is a big game-changer. We are now able to: 

  • Obtain test results for a layout within a minute 
  • Integrate testing even deeper into our design process, performing eye tracking tests within our design tool 
  • Acquire an 80-90% valid test result at any time 
  • Understand how users look at our design and its elements (heat map), and in which order (Gazeplot) 

Even though the assumptions we make during our design work, based on many years of experience in UX and Usability, are often quite close to reality — this embedded test methodology helps us to increase conversion rates and optimise usability even more. 

Using AI testing and predictive design practices on your next project is a critical practice, ensuring that your users have the best experience possible from day one! Got more questions about the benefits of using AI and predictive design, reach out to one of our UX Design experts today! 

Stay in the Loop

We will use the personal data you are sharing with us solely for the purpose of sending you our newsletter. See more here: Data Privacy.


Let us know how we can help you.