Predicted vs measured
Foresight's models learned from real human data, but no person sees your creative during an analysis. Heatmaps, gaze plots, emotion values and element shares are predictions for a typical viewer. Use them to compare and improve creatives, and confirm big decisions with real-world data.
Foresight predicts how people are likely to react to your creative. It does not measure the reactions of real people looking at it. This distinction shapes how you should read every result.
What it is
Traditional creative testing measures real people: an eye-tracking session records where participants look, a survey asks them how they feel. Foresight takes a different route. Its models were trained on exactly this kind of human data (eye-movement recordings and people's reports of the emotions images made them feel) and learned the patterns behind it. When you analyse a creative, Foresight applies those patterns to predict the response of a typical viewer.
No panel or participant is shown your creative during an analysis. That is why results arrive quickly. Re-analysing the same creative can give slightly different values, because parts of the analysis are regenerated each time.
What is predicted, output by output
| What you see | What it really is |
|---|---|
| Heatmap and fogmap | A predicted map of how likely each area is to be looked at, scaled to your creative's own strongest point. |
| Gaze plot (images) | A simulation built on that map: numbered stops at the main hot spots of predicted attention. It is not a recorded eye path. |
| Attention build-up (images) | A simulated replay of how attention is likely to spread over the first seconds of viewing. |
| Areas of interest (images) | The share of predicted attention that falls inside an area you drew, plus an estimated viewing time that moves with the share. |
| Eight emotion values | The model's predicted strength for each emotion being evoked, not the share of people who feel it. |
| KPIs (E-Score, Attention, Emotional Engagement…) | Scores computed from these predictions and positioned against reference creatives that were analysed with the same models. |
Because the reference creatives are also scored with Foresight's predictions, every benchmark comparison is like for like: predicted against predicted.
Elements and the gaze plot
All element figures (a face, a logo, a block of text) come from the same predicted attention map. That has two practical consequences:
- Within one creative, shares compare directly. If the face takes a larger share of predicted attention than the logo, that comparison is meaningful.
- Across element types, gaps to the reference don't rank. Areas of type Face, Logo or Text are each compared with typical values for their own element type and size. Each type has its own yardstick, so don't conclude that a face "beats its reference by more" than a logo does. See Element benchmarks.
The gaze plot has no benchmark of its own. Its numbered stops mark the creative's attention hot spots; treat their order as indicative only. See Gaze plot.