The science
Foresight predicts where people are likely to look and which emotions a creative is likely to evoke, using models that learned from real human eye-movement recordings and emotional responses. No panel sees your creative; every result is a prediction for a typical viewer.
Foresight is a predictive tool. Instead of recruiting a panel and recording their reactions, it uses models that learned from real human data to predict two families of response to your creative:
- Attention: where people are likely to look. The attention model learned from eye-tracking research, in which researchers recorded where many people looked while freely viewing images.
- Emotion: which emotions the content is likely to evoke. The emotion model learned from a large collection of images that people labelled with the emotion each image evoked in them.
A third, smaller model predicts how visually complex a creative feels, which feeds Mental Demand.
Foresight turns these predictions into KPIs and positions them against a library of reference creatives. No person sees your creative during an analysis, and the results describe a typical viewer, not your specific audience.
In this section
- Methodology: how the models learned, how emotions and attention are described, and how predictions become scores.
- Validation: how the models are tested against human data they never saw, and what that does and doesn't prove.
- Limitations and responsible use: what predictions can't tell you, and how to use them well.
For the difference between predicted and measured results in day-to-day use, see Predicted vs measured.