Get started

What is Foresight

In short

Foresight uses two prediction models, one for attention and one for emotion, to estimate how a typical viewer will react to your creative. It turns those predictions into 0–100 scores, overlays and AI-written recommendations, so you can compare options and fix problems before launch.

Foresight is a predictive creative-testing tool. You give it an image or a video (an ad, a social post, a pack shot, a key visual) and it tells you how a typical viewer is likely to respond to it:

• Where people are likely to look, and whether their eyes find the elements that matter.
• Which emotions the creative is likely to evoke, how strong the reaction is and whether its tone is positive or negative.

You get these answers without recruiting a panel or running a survey. Foresight's models have already learned from real human data, and they apply what they learned to your creative.

What Foresight predicts

Foresight relies on two prediction models.

The attention model estimates, for every point of your creative, how likely people are to look there during free viewing. It learned from eye-tracking research, where the gaze of many real people was recorded while they looked at images. See The attention model.

The emotion model estimates how strongly your creative evokes each of eight emotions: amusement, anger, awe, contentment, disgust, excitement, fear and sadness. It learned from a large collection of images that people labelled with the emotion each one made them feel. See The emotion model.

Both models describe a typical viewer, not a specific person or audience segment. Predicted vs measured explains what that means in practice.

What an analysis produces

Every analysis gives you three kinds of output.

Scores. A small set of KPIs on a 0–100 scale, led by the E-Score, the headline score. The E-Score combines two building blocks (Emotional Engagement, how strongly the creative moves people, and Attention, how well it captures and organises the eye), so a high E-Score needs strength on both. Supporting scores such as Valence, Clarity, Focus, Spread and Mental Demand help you understand why the headline is high or low. Most scores are positioned against a large set of reference creatives, so 50 roughly means "typical". See the E-Score and the KPIs overview.

Visuals. Overlays on top of your creative that show where attention goes: a heatmap, a fogmap and, for images, a gaze plot and an attention build-up animation. For videos you also get moment-by-moment curves and a scene-by-scene breakdown.

Insights. Foresight Copilot reads all the scores and writes a short summary, a detailed analysis and a prioritised list of concrete recommendations. If you set a campaign objective, the insights are framed around it.

For images analysed on their own, you can also pick a benchmark (a creative type and sector) so every score is shown next to the average of comparable creatives. See Benchmarks.

Typical uses

• Choosing between options. Run two or three versions of a key visual and see which one draws the eye to the product and evokes a stronger reaction.
• Fixing a creative before launch. Spot a logo that nobody is predicted to notice, a layout that scatters attention, or a scene where the emotional reaction drops.
• Checking fit for a placement. See how your creative performs inside an Instagram, Facebook, X or TikTok screen rather than on its own.
• Briefing and reviewing. Share results and notes with your team, compare creatives side by side and export reports.

An example: a food brand tests two pack designs. Both evoke similar emotions, but on one the heatmap shows attention split between three competing claims while the product name gets little attention. The team simplifies that layout before going to print.

How to read it

Think of Foresight as a fast, consistent first read of how people are likely to react. It is strongest when you compare versions of the same idea, or a creative against its benchmark, and look at the reasons behind the scores rather than a single number.

Common pitfalls

• Foresight does not survey or track your audience. Results are predictions for a typical viewer, not measurements of your customers.
• A score is not a forecast of sales or click-through. Use Foresight to compare and improve creatives early, then confirm important decisions with in-market data.
• Images and videos are scored on the same scale but compared with different benchmarks (see how scores are positioned), so don't rank an image against a video by their raw scores.