How AXIOM scores a sitev1.0
Every number is traceable to specific scanner output. No PRNG, no hidden defaults, no fabricated industry data — just deterministic scoring backed by AI interpretation where it adds value.
Evidence in, score out.
Every dimension score is computed from real, evidence-backed scanner output. Where a dimension cannot be scored due to insufficient data, it is reported as null with a stated reason — never silently filled with a default value.
Where AI helps — and where it doesn't
AI (Anthropic Claude) is used only to narrate evidence-based findings and produce executive summaries on top of deterministic scores. AI does not generate or modify scores.
Weighted scoring across nine evidence-based dimensions
Accessibility
WCAG 2.2 AA compliance, alt text, contrast, keyboard navigation, ARIA semantics.
Usability
Nielsen heuristics, error prevention, recognition over recall, consistency.
Navigation
Information architecture, breadcrumbs, search, sitemap, mega-menu, mobile nav.
Performance
Page weight, lazy loading, critical CSS, asset optimisation.
Content
Reading level, scannability, F-pattern adherence, CTA copy clarity.
Trust & Security
HTTPS, contact clarity, social proof, third-party tracker exposure.
Mobile
Touch targets, viewport meta, text legibility, mobile nav patterns.
Conversion Readiness
CTA hierarchy, above-fold value prop, friction signals, decision support.
Regulatory & Resilience
POPIA/GDPR consent, DSAR mechanisms, PWA, service worker, graceful degradation.