Understanding AI Creative Insights

What is AI Creative Insights

AI Creative Insights is a feature in Decode that evaluates creatives using AI without needing any research participants. You upload an image or video of your creative and Decode analyses it using facial coding and eye tracking models to produce attention, clarity, focus and engagement scores instantly.

The analysis is based on how a viewer's attention and emotions would respond to the creative, predicted by AI models trained on data from thousands of real participants across a wide range of creative categories. Results are available as soon as the file is processed.

AI Creative Insights is designed to be used early in the creative process, before any live research is commissioned. It gives you a fast read on how a creative is likely to perform so you can shortlist the strongest options and take only those into live research with real participants.

What it measures

AI Creative Insights produces five scores for each creative. Each score is shown alongside a Benchmark Score, which is the average for creatives in the same category and sub category. The benchmark lets you read each score in context rather than in isolation.

1. Decode Score

A composite score from 0 to 100 that combines facial coding and eye tracking data. It is benchmarked against 8,000 or more creatives in the same category. Use this as the overall performance indicator before going into the individual metrics.

2. Focus

Measures whether the creative directs viewer attention to the key elements the offer, price point, CTA or logo. Focus is the most actionable metric for conversion intent. A low Focus score means viewers are looking at the wrong parts of the creative rather than the elements that drive action. Use this when evaluating visual hierarchy and whether the intended elements are being noticed first.

3. Clarity

Measures how clearly and legibly the core message communicates at a glance. Clarity is influenced by typography, contrast and information density. A low Clarity score means the message is difficult to read or understand quickly. Use this when evaluating whether the message is easy to grasp, particularly for creatives with a lot of text or competing visual elements.

4. Attention

Measures how strongly the creative attracts the eye on first contact. A high Attention score means the creative is effective at stopping a viewer. Use this when comparing creatives for their ability to capture initial interest.

5. 1.Engagement

Measures whether viewers remain engaged with the creative past their initial attraction to it. A high Engagement score indicates the creative holds interest beyond the first glance. Use this alongside Attention to understand whether a creative not only stops viewers but also keeps them.

Visual outputs

In addition to the five scores, AI Creative Insights provides three visual views of the creative and a set of qualitative outputs in the right panel.

1. Original

The creative as uploaded, with no overlay. Use this as the baseline view before switching to Heatmap or Fogmap.

2. Heatmap

An overlay showing where viewer attention is concentrated across the creative. Areas in red and yellow indicate high attention. Areas in green indicate lower attention. Use this to identify which parts of the creative are drawing the most focus and whether those are the intended elements.

3. Fogmap

An overlay showing the areas of the creative that viewers are least likely to notice. Areas covered by the fog are predicted to receive little or no attention. Use this to identify elements that may be overlooked, such as a CTA or logo placed in a low attention zone.

4. Area of Interest

Area of Interest lets you define specific regions of the creative and measure how much attention each region receives. Each region you mark shows two values. Attention is the percentage of total viewer attention captured by that region. Time to Discover is how many seconds it takes for a viewer's attention to first reach that region. A lower Time to Discover means the element is noticed sooner.

Use Area of Interest when you want to measure the performance of a specific element for example, how quickly the discount offer is noticed, or how much attention the brand logo is receiving relative to the rest of the creative.

5. AI Summary

A written description of the creative's overall tone, mood and visual properties, generated by AI. It identifies the dominant visual cues and describes how they contribute to the creative's impact. Below the summary, Visual Cues lists specific elements in the creative such as colour, faces, text placement or imagery and describes how each contributes to the overall impression.

6. Emotional Insights

A breakdown of the predicted emotional responses the creative is likely to trigger in viewers. Each emotion is shown as a score. The dominant emotion reflects the strongest predicted emotional response overall. Use this to understand the emotional tone the creative is communicating and whether it matches the intended audience response.

When to use it

AI Creative Insights is most useful in the following situations.

Shortlisting creatives: When you have multiple versions of a creative and need to narrow them down before investing in a full research study. Upload all versions and compare their scores to identify the strongest candidates.
Checking visual hierarchy: When you want to confirm that viewer attention is going to the right elements the offer, CTA, logo or key message before the creative goes live.
Evaluating message clarity: When you want to check whether the core message is easy to grasp quickly, without running a survey or interview.
Pre-launch quality check: When a creative is near final and you want a quick read on its predicted performance before it is published or submitted to a media channel.
Comparing against benchmarks: When you want to understand how a creative performs relative to others in the same category. The Benchmark Score shown alongside each metric is the average for creatives in that category and sub category.

How it fits into a creative workflow

AI Creative Insights is the first step in a creative review pipeline. Upload your creatives and review the scores to identify which perform strongest. Use the Heatmap and Fogmap to understand where attention is going. Use the Benchmark Score to compare each creative against the category average. Use the Prediction tab to get AI-generated analysis and design recommendations.