The Neurons Trust Prediction is only available on API
What is the Trust score?
The Trust score measures how much a creative feels credible, sincere, and emotionally aligned to the viewer. It reflects subconscious, intuitive perceptions of honesty and authenticity, based on quick reactions to words like Trustworthy, Honest and Sincere.
This metric provides a new way to assess brand credibility, moral alignment, and emotional resonance in advertising—essential signals for building brand equity and long-term loyalty.
How is the Trust score calculated?
The Trust metric was developed using our Fast Response test (FRT) methodology. Participants view content for a few seconds and then rate it using association words like "Trustworthy," "Honest," and "Sincere" by pressing "yes" or "no." Their response time is also recorded, giving weight to how strongly they feel. By normalizing response times and responses across multiple participants, we ensure the FRT score is both comparable and reliable.
The Trust score combines three key inputs:
- The yes/no answer
- The reaction time, normalized per individual (to account for faster/slower responders)
- The level of agreement across participants
Faster, confident "yes" responses increase the score, while uncertain or slow responses reduce it. This enables the model to capture both what participants feel and how strongly they feel it.
For a more detailed explanation of the FRT methodology, see this article.
How the Trust model is built
The Trust prediction model is trained using data from close to 6000 participants. It uses an EfficientNetB2V2 convolutional neural network, pre-trained and fine-tuned to predict Trust based on visual features.
- Output: Continuous Trust score per image (or video frame).
- Explainability: Grad-CAM heatmaps show what parts of the image influence Trust predictions—often human faces, eye contact, and emotionally expressive areas.
For videos, predictions are made frame-by-frame using the image model, producing a timeline of Trust scores over the asset’s duration.
What affects the Trust score?
Based on internal regression analyses and explainability checks, the following factors influence Trust:
- Early visual focus: If the viewer’s attention is captured early, Trust increases.
- Emotive elements: Human faces, eye contact, and sincere expressions help build Trust.
- Cognitive overload: Complex visuals, unclear messaging, or excessive detail may reduce Trust.
Narrative reward: In videos, if focus is gained early but not rewarded with clarity or story, Trust may drop.
Interpreting Trust scores
Trust Scores range from 0 to 100:
- Low scores (0-40): The asset does not invoke trust or authenticity and might not connect with viewers.
- Medium scores (41-60): The asset has moderate trust building and some viewers may find it authentic and trustworthy.
- High scores (61-100): The asset invokes high trust and is found authentic by the majority of general population viewers.
Trust is weakly correlated with traditional KPIs, but that’s a strength: it reveals emotional and cognitive patterns that conventional metrics overlook.
In video ads, Trust correlates positively with VTR (+0.55) and CTR (+0.45), suggesting that emotionally credible ads perform better overall.
Use Trust as an indicator of perceived authenticity and brand alignment, especially useful in upper- and mid-funnel strategies.