Transparent by design
Methodology
Corexi's PX Score is not a black box. Every number is grounded in published research, industry standards, and reproducible evaluation criteria. Here is exactly how we measure product experience quality.
How the PX Score is calculated
The PX Score is a weighted composite of 5 UX dimension scores, each derived from AI analysis across 9 evaluation categories.
PX Score =
Visual Quality
0.25
Behavioral Health
0.20
Usability
0.25
Accessibility
0.15
Cognitive Load
0.15
Each dimension is scored 0-100. When multi-viewport analysis is enabled, desktop results carry 60% weight and mobile results carry 40%.
9 AI evaluation categories
Every page scan is evaluated across these 9 categories. Each has specific, measurable criteria rooted in research.
Visual Hierarchy
How effectively the page guides the eye through heading differentiation, visual flow, and attention priority.
CTA Visibility
Whether primary calls-to-action are easy to find with sufficient contrast, size, and placement.
Form Design
Label clarity, error state handling, input field sizing, and placeholder/label correctness.
Spacing & Layout
White space sufficiency, element spacing consistency, grid system usage, and visual density.
Typography
Font size readability (16px minimum body), line length (45-75 chars), hierarchy, and line height.
Color & Contrast
WCAG AA compliance for text/background contrast, palette consistency, and information conveyed by color alone.
Mobile Responsiveness
Touch target sizes (44x44px minimum), absence of horizontal scroll, and mobile text readability.
Cognitive Load
Information density, clarity of next action, navigation simplicity, and decision paralysis prevention.
Accessibility & Compliance
Alt text, form labels, keyboard navigation, ARIA attributes, focus indicators, and touch target sizing.
Neurodiversity scoring
Beyond standard UX metrics, Corexi derives neurodiversity-focused scores from the same analysis data. Five specialized lenses, each with a distinct weighting formula.
ADHD Friendliness
Cognitive Load (40%) + Visual Quality (25%) + Usability (20%) + Behavioral Health (15%)
Dyslexia Readability
Visual Quality (35%) + Usability (30%) + Accessibility (20%) + Cognitive Load (15%)
Autism Predictability
Usability (30%) + Visual Quality (30%) + Cognitive Load (25%) + Behavioral Health (15%)
Sensory Sensitivity
Visual Quality (30%) + Accessibility (30%) + Cognitive Load (25%) + Behavioral Health (15%)
Color Vision
Accessibility (55%) + Visual Quality (45%), with type-specific modifiers for Protanopia, Deuteranopia, Tritanopia, and Achromatopsia
Confidence and data coverage
Not all scans are equal. When behavioral analytics are connected, the PX Score benefits from real user signals. Without them, Corexi relies solely on visual AI analysis.
Visual AI only
Base
~40% coverage
Screenshot-based analysis across 9 categories
+ 1 analytics provider
Moderate
~65% coverage
Behavioral signals validate visual findings
+ 2-3 providers
High
85-95% coverage
Full hybrid reasoning with cross-validated insights
Research and standards
Every scoring criterion maps to published research, international standards, or peer-reviewed studies.
Nielsen Norman Group (1994)
Foundation for visual hierarchy, error prevention, and cognitive load evaluation.
W3C (2023)
Definitive standard for contrast ratios, focus indicators, touch targets, and color usage.
Baymard Institute (2024)
Form design patterns, checkout friction analysis, and error state best practices.
Google (2020)
Correlation between performance metrics and user engagement, bounce rate, and conversion.
British Dyslexia Association (2018)
Font selection, line spacing (1.5x min), letter spacing (0.12em), and text alignment for dyslexic readers.
University of Cambridge (2012)
Research on how reading patterns differ for dyslexic users and impact of typography on comprehension.
Color Universal Design Organization (2015)
Principles for color-blind safe design: avoid sole color reliance, use sufficient luminance contrast.
International Organization for Standardization (2019)
Framework for iterative design evaluation, user research integration, and usability testing methodology.
UX Design Institute (2023)
Research on attention fragmentation, decision paralysis in cluttered interfaces, and progressive disclosure.
Fitts, P.M. (1954)
Foundational model for CTA sizing, touch target dimensions, and interactive element placement.
Miller, G.A. (1956)
Cognitive chunking limits that inform navigation design, menu item counts, and information architecture.
W3C (2023)
Specific contrast ratio thresholds (4.5:1 AA, 7:1 AAA) used in color scoring.
Questions about our methodology? We are transparent by design.