Methodology

How we grade a website

Every number on a CloudSpeed report comes from a named, checkable source. Here's exactly what we measure, where the data comes from, and — just as important — what it doesn't tell you.

Performance & Global Grade

Performance, Accessibility, Best Practices, and SEO scores come directly from Google's PageSpeed Insights API (Lighthouse v10+), run separately for Mobile and Desktop. Field data (how your site performs for real visitors) comes from Google's Chrome UX Report (CrUX), aggregated over the trailing 28 days.

Global Grade is our own composite, not a Google metric: 60% Performance score + 40% Structure score (the average of Best Practices, Accessibility, and SEO), mapped to a letter grade. Core Web Vitals pass/failfollows Google's official thresholds for LCP, CLS, and INP.

AI & Search Readiness

This section is our own analysis, built from two pillars:

Technical Discoverability

robots.txt rules for named AI crawlers (OAI-SearchBot, ChatGPT-User, GPTBot, PerplexityBot, ClaudeBot, Google-Extended), XML sitemap presence, llms.txt, meta description, and noindex directives — checked live against the page's public HTML at scan time.

Machine Interpretability

Structured data (JSON-LD, including nested @graph blocks), business identity signals, page title/heading/meta clarity, trust signals (contact info, about page, policies, proof of work), question-answering content, plus accessibility and layout-stability checks from Lighthouse.

These checks read the same publicly available HTML, robots.txt, and sitemap that any crawler would see — nothing requires access to your site's backend, analytics, or Search Console.

What a good score means — and doesn't

A passing check means a real, verifiable barrier has been removed — a crawler that was blocked can now reach your pages; a business with no machine-readable identity now has one.

It does not mean you'll rank higher, get cited by ChatGPT or any AI assistant, or convert more visitors. Those outcomes depend on content quality, relevance, competition, and factors no automated tool can measure from the outside.

We can't confirm whether Google has indexed a page, whether any AI system has actually crawled or cited it, or verify that claims made on the page (pricing, credentials, reviews) are accurate.

Limitations

We analyze the single URL you submit, not your entire site.
Sites with bot protection, CAPTCHAs, or aggressive firewalls can block the automated test entirely, which shows as an analysis failure rather than a score.
Lab data (Lighthouse) is a simulated, single-run snapshot; field data (CrUX) needs enough real-user traffic to exist at all — very low-traffic sites may show no field data.
robots.txt and sitemap checks reflect the rules in place at the moment you run the check; they can change at any time.

Questions

If a result looks wrong or you want to understand a specific check in more depth, we're happy to walk through it — get in touch.