Features
NEXORA runs a hardened, dependency-free engine against the live site — real responses, real headers, real robots.txt and sitemap. Here is exactly what each layer measures.
Scores are weighted into a single NEXORA Score: performance and security at 0.20 each, SEO at 0.20, accessibility and UX at 0.15, content at 0.10. The weights are published — not a secret algorithm.
9 automated checks · weight 0.20
Document size
Total HTML bytes on the wire vs. a 3.5 MB ceiling.
Compression
Presence of gzip/br Content-Encoding serving.
Cache headers
Cache-Control / Expires on the document.
Render-blocking CSS
Count of stylesheets blocking first paint.
JavaScript weight
Number and size of inline + external scripts.
Image dimensions
Width/height or aspect-ratio on every image, to avoid layout shift.
Lazy-loading
loading="lazy" usage and responsive srcset coverage.
DOM weight
Approximate node count — signal for client-side cost.
TTFB
Server response time measured by the NEXORA probe.
9 automated checks · weight 0.20
Meta title & description
Presence, length and content quality.
Single H1 + hierarchy
One H1 and a sane heading progression.
Canonical tag
Explicit canonical URL.
robots meta & robots.txt
Indexing policy and crawler instructions file.
XML sitemap
Sitemap presence and reachability.
Structured data
JSON-LD blocks found in the page.
Internal linking
Internal link inventory and sampled broken links.
Open Graph & Twitter
Social preview tags completeness.
Declared language
html lang attribute.
6 automated checks · weight 0.15
Image alt text
Alt coverage across all <img> elements.
Form field labels
Associated labels / placeholder-only or unlabeled controls.
Semantic landmarks
main / nav / footer landmark presence.
Duplicate IDs
DOM identifiers that break label and aria wiring.
Link names
Links with empty or placeholder text.
Form autocomplete
Autocomplete attribute coverage on controls.
8 automated checks · weight 0.20
HTTPS + HSTS
Encrypted transport and HTTP-Strict-Transport-Security.
CSP header
Content-Security-Policy presence.
nosniff & framing
X-Content-Type-Options and frame-ancestor protection.
Mixed content
HTTP sub-resources on an HTTPS page.
Cookie flags
Secure / HttpOnly coverage on Set-Cookie headers.
Exposed secrets
Common API-key patterns scanned in page source.
Sensitive comments
Credentials or private data in HTML comments.
Server banner
Info-leaking Server header detection.
8 automated checks · weight 0.15
Mobile viewport
viewport meta for mobile-first rendering.
Navigation & anchors
Working nav with internal destinations.
CTA clarity
Action-directed primary calls-to-action.
Responsive signals
Viewport + responsive images + media queries.
Trust signals
Testimonials, guarantees, ratings, logos near conversion points.
Footer & contact
Footer landmark, site identity and openings.
Social presence
Linked social profiles.
Favicon & theme
favicon, apple-icon, theme-color metadata.
5 automated checks · weight 0.10
Content depth
Word count threshold on the page body.
Text-to-code ratio
Share of human text vs. markup.
Link wording
Descriptive link text.
Media richness
Images and video inventory.
Readability
Words-per-sentence estimate from real body text.
NEXORA never invents numbers. Chevron honesty is a core product value: what we can’t measure, we say so.
Real Core Web Vitals (LCP, CLS, INP), color contrast and keyboard interaction need a rendered browser. Those checks return unavailable on every NEXORA report — plainly — instead of a made-up score.
The engine fetches the actual document, reads response headers, parses robots.txt and the sitemap, and samples internal links for 4xx/5xx. Nothing is seeded from page claims or heuristics pretending to be measurements.
Auditing fetches validate every URL and DNS result against private, loopback, link-local and cloud-metadata ranges — including re-checking each redirect hop — so NEXORA can’t be pointed at your internal network.
Every saveable audit is summarized into a client-ready report: executive summary, biggest problems, biggest opportunities and a phased 90-day roadmap — written by the deterministic engine, or by any OpenAI-compatible model you connect.
Pit any two websites across all six categories, or attach up to five competitors to a single report. Deltas, category leaders and borrowed opportunities, all from the same real checks.
Agencies can brand reports with their own logo, color, contact line and footer note — then export a real PDF through the browser’s print pipeline. No vaporware export, no fake watermarks.