Marketers Could Use AI to Make Sure You See Their Ads—Here's How
AdGazer is a new AI tool developed by researchers at the University of Maryland and Tilburg University to predict whether users will look at digital ads, aiming to improve ad placement and effectiveness. Unlike traditional approaches that analyze only the ad itself, AdGazer evaluates both the advertisement and the surrounding webpage context. It was trained using eye-tracking data from 3,531 display ads, capturing real users' gaze patterns as they browsed webpages. When tested on unseen ads, the tool achieved an 83% correlation with actual gaze data. The study found that webpage context accounts for at least 33% of ad attention and about 20% of brand-specific attention—significantly challenging the view that creative design is the sole driver. AdGazer uses a multimodal large language model to extract topics from both ads and webpage content, and combines these semantic matches with visual features via an XGBoost model to generate an attention score. The team also released Gazer 1.0, an interface that predicts gaze time and generates heatmaps for uploaded ads. While currently academic, the technology could soon be integrated into production ad-tech solutions.
