Methodology
Site Inspector is a presentation layer over Google's public data. Nothing here is proprietary scoring — every grade traces back to a documented Lighthouse audit or a Chrome UX Report metric. This page is the full audit trail.
Data source
We call the Google PageSpeed Insights API v5 directly from your browser. A single request returns two distinct datasets:
- Lab data (Lighthouse) — one synthetic page load on an emulated mid-tier mobile device with a throttled network, run on Google's servers. Repeatable and diagnostic, but it is not what your visitors feel.
- Field data (Chrome UX Report / CrUX) — anonymised, aggregated metrics from real Chrome users over a rolling 28-day window. Only available once a URL or origin has enough traffic; otherwise it is omitted.
AuditOps is an independent project. We are not affiliated with, endorsed by, or sponsored by Google. "Lighthouse", "PageSpeed Insights" and "Chrome" are trademarks of Google LLC.
The severity model
It mirrors the colour bands on Lighthouse's own gauges, made explicit and consistent across every signal:
| Grade | Score (0–1) | Meaning |
|---|---|---|
| PASS | ≥ 0.90 | Green. No action needed. |
| WARN | 0.50 – 0.89 | Orange / "needs improvement". Schedule a fix. |
| CRITICAL | < 0.50 | Red. Fix first. |
- Binary audits (HTTPS, canonical present, image alt…) report 0 or 1; 1 = PASS, 0 = CRITICAL or WARN by context.
- CrUX field metrics use Google's own buckets: FAST = PASS, AVERAGE = WARN, SLOW = CRITICAL.
- Forced CRITICAL regardless of the numeric score: no HTTPS, and any known-vulnerable JavaScript library.
The roll-up verdict counts CRITICAL / WARN / PASS findings across all sections. The report severity is simply the worst section — one CRITICAL anywhere makes the whole report CRITICAL, by design: you should not ship with a known-broken signal.
What is checked
The four Lighthouse categories (each scored 0–100) plus these individual audits, grouped into the dossier sections:
- Performance & Core Web Vitals (lab): FCP, LCP, Speed Index, Total Blocking Time, CLS, Time to Interactive, server response time (TTFB).
- Asset delivery & caching: render-blocking resources, HTTP/2, text compression, unused JavaScript, unused CSS, next-gen image formats.
- SEO: document title, meta description, crawlability, robots.txt validity, canonical link, structured data.
- Accessibility: colour contrast, image alt text, viewport meta, tap-target sizing.
- Best-practices & security signals: HTTPS, no known-vulnerable libraries, CSP-against-XSS heuristic, console errors.
- Field data: LCP, INP, CLS and FCP percentiles with FAST / AVERAGE / SLOW classification.
Limits & disclaimers
- Not a security audit. Lighthouse checks only a handful of security signals. It does not inspect security response headers (HSTS, X-Frame-Options, X-Content-Type-Options, Referrer-Policy, full CSP), TLS configuration, authentication, or server-side code. A clean score is not a security guarantee.
- Lab ≠ field. A great Lighthouse score with poor field data means real users are slower than the synthetic test; trust the field data for user experience.
- No field data is common. Smaller sites simply lack enough Chrome traffic for CrUX. That is not a failure.
- One run is a snapshot. Lab scores vary run to run. Re-test and look at trends.
- We store nothing. Scans run in your browser against Google's API; results are never persisted on our servers.
About AuditOps Research
AuditOps Research builds free, no-signup web diagnostics. Site Inspector is maintained by the AuditOps team; we keep the grading rules public on this page precisely so the output is auditable. Found a grading edge case? The operator's contact details are on the legal notice.