The Science Behind Our Predictions

Built on the world's largest dataset of real eye-tracking data with industry-leading accuracy

Built on the world's largest dataset of real eye-tracking data with industry-leading accuracy

✓ 1.7M+ Tests ✓ 34M+ Fixations ✓ 4.5% Error Rate

Dataset Foundation

Our model's accuracy stems from training on the most comprehensive real-world webcam eye-tracking dataset ever assembled from people viewing content on desktop and laptop screens

Our model's accuracy stems from training on the most comprehensive real-world webcam eye-tracking dataset ever assembled from people viewing content on desktop and laptop screens

1.7M+

Eye-Tracking Tests

Individual testing sessions from real webcam eye-tracking studies ensuring robust attention pattern recognition across diverse viewing scenarios

Individual testing sessions from real webcam eye-tracking studies ensuring robust attention pattern recognition across diverse viewing scenarios

12,000+

Unique Items

Distinct designs analyzed including advertisements, websites, and packaging viewed on desktop and laptop screens in real-world studies

Distinct designs analyzed including advertisements, websites, and packaging viewed on desktop and laptop screens in real-world studies

34M+

Fixation Points

Over 33.9 million individual eye fixations captured via webcam eye-tracking on desktop and laptop devices, reflecting authentic visual behavior

Over 33.9 million individual eye fixations captured via webcam eye-tracking on desktop and laptop devices, reflecting authentic visual behavior

Exceptional Predictive Accuracy

We measure performance using Mean Absolute Error (MAE) against real webcam eye-tracking data from desktop and laptop viewing sessions - the lower the score, the closer our predictions match actual human gaze patterns.

Latest Model Performance (July 2025)

3.85%
Training Error
4.52%
Validation Error

Overall Mean Absolute Error across all attention components

Component-Specific Accuracy (Validation MAE)

0.043
Hold Component
Sustained attention
0.052
Speed Component
First fixation speed
0.040
Reach Component
Participant coverage

What Does <5% Error Mean for Your Business?

In simple terms: When our AI predicts how much attention a specific area of your design will get, it's less than 5% wrong.

🎯
Highly Accurate Predictions

If we predict 30% of people will notice your logo, the real number will be between 28.5% and 31.5%

💰
Confident Design Decisions

Make expensive design changes with confidence - our predictions are reliable enough for business decisions

Skip Costly Testing

Get research-grade accuracy without running expensive eye-tracking studies for every design iteration

Understanding the Accuracy Metrics

Hold Component (0.043 MAE): Predicts how long people fixate on areas with outstanding accuracy - best-in-class performance for sustained attention prediction.

Speed Component (0.052 MAE): Exceptional accuracy in predicting time-to-first-fixation, determining which areas capture immediate attention.

Reach Component (0.040 MAE): Industry-leading performance in forecasting what proportion of users will notice specific areas across different viewing durations.

Deep Learning Architecture

Input Image
Feature Extraction
Multi-scale Processing
Attention Prediction
VAI/Hold/Speed/Reach

Technical Implementation

Deep Neural Networks

State-of-the-art convolutional neural networks trained specifically on webcam eye-tracking data from real desktop and laptop viewing sessions

Multi-Scale Analysis

Processes images at multiple resolutions to capture both fine-grained details and overall composition patterns that influence visual attention

Real-World Training

Trained exclusively on actual human viewing data from webcam eye-tracking studies, not synthetic or laboratory-controlled environments

Experience the Accuracy Yourself

See how our scientifically-validated predictions can enhance your design process