Eye tracking is a methodology that has been developing rapidly in the last few years because of technological advances. One of the most recent ones is the predictive eye tracking, which eliminates the use of participants to generate data and creates visual interaction heatmaps based on artificial intelligence. But does predictive eye tracking really work as accurately as real eye tracking? This is the question under discussion in the present article.

What is eye tracking?

Eye tracking is a technology for measuring eye movement and eye positions while the person interacts with a visual stimulus. It has been used in many fields to evaluate people’s attention, such as marketing, UX design, packaging design, product design, etc. This methodology allows researchers to better understand customers’ and users’ unconscious behaviours. Eye tracking data can be used to improve the design or interaction between the screen and people. There are two main ways of conducting eye tracking research: predictive eye tracking and regular eye tracking.

‍Predictive eye tracking

Predictive eye tracking, or AI-generated eye tracking, predicts the outcome of eye tracking research based on a large amount of generated data and AI algorithms. AI algorithm will identify a certain flow of human attention and predict the potential human attention after systemisation of these flows. It predicts where people are most likely to look at while interacting with a visual stimulus.

Here are some features of predictive eye tracking: visual stimuli are easier to recognise; the first 3-5seconds are captured to showcase on heatmaps; human participants are not required.

Regular eye tracking

Regular eye tracking can also be seen as a traditional way of eye tracking. It is based on real interaction between the person and the stimulus, therefore always requires human participants and a platform where they can join in a test to collect data. Researchers are more flexible between choosing qualitative and quantitative methodology since, unlike the fixed testers in predictive eye tracking, in regular eye tracking participants can be selected based on different research questions and intentions behind the test.

Regular eye tracking can be conducted in-lab; however, some eye tracking companies have developed an easy and remote testing solution to complement the laboratory options. For example, Oculid provides its users with a platform where research teams can conduct studies on mobile phones everywhere using the selfie camera as an eye tracker device.

Predictive eye tracking vs.Regular eye tracking

It is undeniable that predictive eye tracking is faster in getting the final result, after all getting an AI to do the work is quicker than getting real participants to interact with an advertisement, packaging, app, or website. In most cases, it is also an affordable option for researchers who only want to get a superficial overview of where most people’s attention is. However, the fast speed and cheap price of predictive eye tracking are at the cost of its accuracy and reliability, since the AI fails to offer more detailed information on the visual interaction and context where the person will interact with the stimulus in real life.

Precision and accuracy

Predictive eye tracking can only offer people a quick overview of visual attention, but lacks the precision offered by a regular eye tracking study. Although it can offer a heatmap to show people’s attention distribution and Area of Interest (AOI) can also be selected on some predictive eye tracking platforms, other metrics of eye tracking like Time to First Fixation(TTFF), Dwell Time and Fixation Duration cannot be accessed only by the way of AI-generated eye tracking. These data are essential to better understand the time that people take to visualise the elements and the cognitive process behind visual attention.

Put a question mark on a universal rule

Predictive eye tracking bases its test on a large amount of AI-generated data and follows mostly the universal rule of eye attraction. But how does it work in real-life cases?

We tried testing landing pages of Zara on a predictive eye tracking platform and a regular eye tracking platform respectively. Although the landing pages look a bit different from each other because the dates of testing were different, the key elements are similar: the big logo of Zara, a large portrait in the middle, filter function and search function. These are potential attractive elements on the Zara landing page.

Looking at these two test results, we found that people’s attentions are very different between the test from a predictive eye tracking platform and a regular eye tracking platform. Which one reflects people’s attention in real life? To answer this question, we need to propose the follow-up questions: why would people focus more on the portrait in the centre according to the test from a predictive eye tracking platform? And why would people focus more on the right top corner based on a regular eye tracking test?

Many people believe that only outstanding features like big images, bright colours and faces attract people, while at the same time they forget that people’s intentions also matter. Zara is an e-commerce platform where people search for items that they want to buy. So, the main intention that customers have to access an e-commerce landing page is not to appreciate it like looking at an artwork, but to find the information they want and the ideal item as quick as possible. That is what predictive eye tracking lacks: context.

The context will influence people’s intentions.

People’s intentions will impact their actions.

The complicated real-life situation tells us that the context matters. The context can be the real-life situation that is set up by a regular eye tracking platform by giving an explanation on the real-life context where testers interact with the app, e-commerce platform, packaging or social media.  In advertising, it is shown that ads in a user-relevant context will receive an increase of the number of attention seconds per 1,000 by 13 per cent.  

Regular eye tracking platforms a real ways done with assumptions and research questions, thus providing testers with a certain context where they perform their actions. An AI cannot predict social and cultural background, the context that the person will be interacting with the visual stimulus and the intention behind the actions. Moreover, predictive eye tracking offers a limited variety of research questions and testing time, therefore lacking context and offering results not as reliable as regular eye tracking can offer.

Predictive eye tracking or regular eye tracking

Now that you know the difference between predictive eye tracking and regular eye tracking, and their pros and cons, it is time to find out where is the best methodology for your research.When it comes to choosing between predictive and regular eye tracking, here are some advice that might be helpful for you:

1. Take your budget and time into consideration.

Money and time are important for any research. Think about your budget limit and how quick you need the result to help you make a wiser decision. Even though traditional eye tracking is usually more expensive than predictive eye tracking, some regular eye tracking platforms can offer more affordable choices since they do not require any additional hardware. Therefore, keep an eye on various possible choices which fit your budget best.  

2. Ask yourself what you want to get from the test and find a research question.

This might be more important than the budget and time. Although money and time matter, spending them on inaccurate results can culminate in spending even more resources in the future. Do you just want to try different colours and sizes? Or do you want to put it in a more concrete context like an e-commerce shopping experience or advertising on social media? Design is more than a simple visual game with pictures and fonts, it is a product based on the assumed interaction with users. The visual aspect is relevant, but the context in which the visual aspect is presented and the person who gets to interact with it also are essential. If you want to get more information and more reliable results, then it is better to consider using regular eye tracking.

Conclusion

Thanks to technology, eye tracking can be offered to companies and institutions for all types of purposes and all types of budgets. Finding the right option ultimately depends on the goals and limitations of each study, with predictive eye tracking coming in hand for quick insights into the average person, while regular eye tracking can offer in-depth information on visual interaction and unconscious behaviour.

If you have more questions regarding eye tracking and how to design an ideal eye tracking research, please contact us today and we are ready to give you advice in more detail.

‍