Methodology pillar

Evidence-First UX Audit

An evidence-first UX audit anchors every finding to observable interface data before drawing conclusions. It is a discipline of constraint: interpretation follows evidence, not the other way around.

Principle

What evidence-first means

• Every finding is anchored to a specific, observable signal captured from the interface — not inferred from category or product type
• Interpretation follows capture: the rendered state of the interface is examined before any conclusion is drawn
• Uncertainty is surfaced explicitly when evidence is incomplete — findings are not generated to fill a quota
• Generic advice without a traceable interface signal is rejected at the system level
• Human judgment is required at the output stage — evidence-first produces structured starting points, not final decisions

Evidence package

What UXMachine captures

• Rendered screenshot

  The visual state of the page as a browser renders it at capture time — layout, colors, contrast, visual hierarchy, and spatial relationships between elements.

• Structural signals

  Heading hierarchy, form count, interactive element count, and accessibility attributes extracted from the DOM at the time of capture.

• Visible text above the fold

  Headlines, primary CTA text, and key interface labels visible in the captured viewport — plus page metadata where available.

• CTA affordance

  Whether primary actions are visually distinct, positioned prominently, and communicate clearly what happens when activated.

• Visual hierarchy signals

  The observable priority structure of the page — whether size, contrast, and spacing guide attention toward a primary path or create competing signals.

• Contrast and accessibility baseline

  Observable contrast ratios and heading structure that contribute to both accessibility compliance signals and general readability.

Output

Output organized by actionability

• 01 Recommended actions

  Specific, evidence-anchored findings that product teams can review and prioritize. Volume is limited — three to five per audit. Each is traceable to a captured signal.

• 02 Verify before changing

  Observations where evidence is present but incomplete. Each includes a stated verification path so teams know what to check before acting.

• 03 Accessibility and hygiene

  Valid technical signals — contrast, heading structure, form attributes — tracked separately from strategic decisions so they do not inflate the primary finding count.