Predictive Eye-tracking

Brainsight has a 94% accuracy rate for instant attention and intuitive viewing behaviour, making it one of the most reliable AI-based saliency prediction tools, reaching a so called 'Golden Standard'.

Below is a comparison video of live eye-tracking in a neuroscience research lab compared to predictive eye-tracking with Brainsight. As you see, the instant attention and initial, instictintive viewing behaviour is extremely close to the live eye-tracking data.

‍

The explanation for this high accuracy rate is in the fact that the brain processes information in three steps:

1. Instinctive processes (reptilian brain; System 1)
2. Emotional processes (limbic system)
3. Reflectieve, rational processes (neocortex; System 2)

The first phase is what Brainsight, or pedictive AI in general, is about. Brainsight's models for this phase are trained with live eye-tracking datasets from global scientifically validated studies and proprietary data- and neuroscience research / expertise.

So, what can or can't predictive eye-tracking do, and when is live eye-tracking relevant? 

Instant Attention & Intuitive Viewing Behaviour versus Engagement

Brainsight predicts instinctive viewing behaviour (Phase 1), typically within the first 3 seconds (up to 5 max). This includes not only attention and intuitive eye movements, but also cognitive load — how easy or hard it is to visually process a scene. High visual clutter increases mental effort, reducing clarity and ultimately brand impact. In Brainsight, this is reflected in the Clarity Score.

After this phase, intuitive System 1 processes transition into Phase 2: still subconscious, but influenced by emotional responses that go beyond generalised predictions. In Phase 3, reflective processes (System 2) take over, including beliefs, reasoning and personal context. Engagement begins in Phase 2 and deepens in Phase 3.

If you want to understand what drives attention beyond the 3-second window — such as dwell-time or emotionally-triggered gaze shifts — you’ll need live eye-tracking. If your goal is to assess emotional connection or brand affinity, live neuromarketing or qualitative research is required.

For more rationalised feedback (e.g. preferences, opinions or intent), standard research tools like surveys, click data or panels are sufficient — though keep in mind these are often subject to bias.

Bottom line

- Use Brainsight/predictive eye-tracking to predict attention, visual clarity, and cognitive load — instantly and at scale
- Use live eye-tracking to capture (measure) actual visual engagement beyond the initial attention window
- Leverage live neuromarketing or behavioral research to assess (measure) emotional or brand-level impact

📺 Example: Pretesting a TV Commercial

When pretesting a TVC, Brainsight enables you to predict how your creative visually performs in the first few seconds — before emotions or context come into play. You can assess:

• Whether your brand assets stand out early enough
• How visually cluttered or clear your scenes are
• The predicted Clarity Score, which reflects how easily the brain can process the visual
• The expected cognitive load — high clutter leads to higher mental effort, reducing message retention and brand impact

These factors are all predictable, based on universal attention mechanisms and visual processing behaviour. They provide actionable input to optimise your content before ad spend, improving saliency, comprehension, and branding effectiveness.

Live Research:

• To understand how viewers engage with your content beyond the first few seconds — e.g. dwell time, secondary attention, re-engagement — you’ll need live eye-tracking to observe real behaviour in context.
• And to measure emotional impact, such as whether viewers like your commercial or feel connected to your brand, you need live (neuro)research. Emotions, memory, and brand associations are subjective and cannot be predicted — they must be empirically measured.

‍