A score is only useful if you can trust it. So here is exactly how Fanalis builds one — every area, what it looks at, and how much it counts. No hidden math.
The weights below are the defaults — they shift to fit the kind of site being checked.
We run Google Lighthouse three times and take the middle result, because a single run can swing by 5–10 points. We score the Core Web Vitals — how fast the main content appears, how stable the layout is, how quickly the page responds to a tap — against Google's published 2026 thresholds.
We run axe-core, the same WCAG 2.2 engine inside Chrome's own tools, then add our own checks for skip-links, landmarks and language. Each violation lowers the score on a curve that saturates — a site with real problems lands in the 50s, not at a misleading zero.
Title and description length, a single clear heading, clean heading order, canonical URL, language tag, Open Graph and Twitter cards, structured data, sitemap, robots.txt, image alt-text, HTTPS, mobile viewport and internal linking — the signals search engines actually use.
The newest area, and one most tools ignore. We check for llms.txt, whether your robots.txt welcomes the AI crawlers (GPTBot, ClaudeBot, PerplexityBot, Google-Extended and more), structured data, semantic HTML, and whether your content is readable without JavaScript — because most AI crawlers don't run it.
Computer-vision measures of colour discipline, harmony, whitespace and visual clutter, blended with an AI judge of typography, hierarchy and rhythm. A trained model — not a checklist — because whether a design looks trustworthy is a judgement, not a rule.
Is there a clear call to action? Does the headline say what you offer? Is there supporting copy, social proof, a contact path, transparent pricing? We score whether a page is set up to turn a visit into a customer — readiness, not a promised conversion rate.
Fully deterministic, modelled on Mozilla's HTTP Observatory: HTTPS, Content-Security-Policy, HSTS, clickjacking and content-type protection, Subresource Integrity on third-party scripts, cookie flags, and no leaking of software versions.
A search engine, an online shop and a documentation site are not the same thing — and judging them by the same rubric is how other tools score google.com a failure. Fanalis detects the genre of your site and adjusts the weighting to fit.
A pillar score is the weighted average of its checks. Each check returns a status, a 0–100 score, and — when something's wrong — the exact issue and a specific fix. Nothing is a black box.
When an area scores 95+, the report tells you plainly: you're in the top 5%, this doesn't need your money. Built in — we can't override it to sell you something.
Every report carries a SHA-256 derived from its findings. Run the same site twice with no changes and the signature matches. If it ever drifts, that's a bug we want to catch.
We don't predict your conversion rate. We don't measure real visitor data. We check one page per run. We'd rather under-claim than mislead you.
