Predict

Built on one of the world's largest single databases of high-quality eye-tracking data collected from real human participants, Predict delivers scientifically grounded attention predictions in seconds rather than the days or weeks required for traditional consumer research.

The platform's machine learning models were trained to predict eye-tracking results with over 95% accuracy by analyzing how visual attention distributes across images and video frames based on factors like contrast, faces, text, motion, and compositional hierarchy.

This accuracy level allows creative teams to use Predict as a genuine optimization tool rather than a rough directional signalmaking real production decisions based on predicted attention data that closely approximates actual viewer behavior.

For image-based advertising assets, Predict generates six distinct behavioral scores covering different dimensions of viewer engagement: where attention focuses first, what captures sustained viewing, how brand elements perform within the overall attention landscape, and other metrics relevant to creative effectiveness assessment.

The Automatic Area of Interest (AOI) Detection feature automatically identifies brands, logos, typefaces, text blocks, and faces within images, then tracks how much attention each element receives in the predicted heat map.

Video analysis extends Predict's capabilities to motion-based advertising, producing both overall and frame-by-frame Brand Attention scores that reveal how viewer attention evolves across the full duration of a video ad.

New capabilities track viewing time and audience reach predictions for image assets and brand performance metrics for video, giving media teams data to forecast how an asset will perform on different platforms before committing to media spend.