AdGazer Predicts Human Ad Attention Accurately

Researchers at the University of Maryland and Tilburg University introduce AdGazer, an eye-tracking–trained model that predicts human attention to display ads, reported in the Journal of Marketing. Trained on gaze data from 3,531 ads, it achieves a 0.83 correlation and finds page context explains about 33% of ad attention and 20% of brand gaze; the team released a Gazer 1.0 demo for placement testing.