Realeyes Unveils Revolutionary AI-Powered Heatmaps Technology

Heatmaps have long been an integral part of the advertising industry, especially attention measurement. From website measurement to tracking ads, marketers have long recognized the importance of tracking gaze to ascertain what part of a creative captured attention. 

Key Takeaways 

• Heatmaps model uses neural network trained on 2.5 billion data points 
• 78% accuracy in predicting which ads will drive more attention
• Predicts reactions as well as gaze, unlike calibration-free eye tracking models 
• No audience sample, calibration, or equipment setup needed 

Heatmaps have long required eye tracking, where web cameras track the gaze of individuals as they view creative. While there have been advances in eye tracking, such as calibration-free models; they still fundamentally only account for gaze, which an essential part of the attention equation, but not the whole story. They also suffer from the inability to scale to meet the increasing proliferation of ad creatives, driven by personalization and social media. 

Realeyes is taking a fundamentally different approach. Our years of studying human attention to content has produced 2.5 billion datapoints of involving people watching video and image ads. Using a neural network trained on these millions of Human Tests, our advanced heatmap model can predict where the audience is likely to place that attention on any given advertisement without requiring any eye tracking. The attention heatmap video is created by overlaying a visual representation of where viewers focus their attention as they watch an advertisement.  

We’re pleased to announce this technology is now part of our Synthetic Testing API. 

Our large data set means our ability to regress or predict attention scores is more valuable and closer tied to the use case of attention than predicting gaze tracking.  

The technology adds an additional layer of insight about viewer attention, from the same participant videos that we use to understand their emotional responses. By doing so we can understand not just how they reacted, but specifically the elements of the content or experience that drove that reaction. 

For each ad, the system records how viewers reacted (e.g., facial expressions) in sync with every frame. The AI learns patterns in visual elements (like colors, faces, or text) to understand what captures attention and drives emotional responses. The resulting heatmap shows which parts of the ad attract the most focus, helping advertisers optimize their content. 

We’ve already demonstrated 78% accuracy in a pilot with Mars where our heatmap and a human test identified the same ad producing the highest level of attention. 

Realeyes’ heatmap enables marketers to get additional insights into human behavior. Since attention is a pre-requisite for effectiveness, and reactions are critical to shifting perceptions and creating enduring memories, heatmaps will continue to be an important asset for marketers looking to capture attention.

We're excited to bring this new feature to customers within Synthetic Testing.