Methodology

How the LoudPixel Score works

The LoudPixel Score is a deterministic 0–100 grade of how ready a page is for Google and for AI search engines (ChatGPT, Perplexity, Google AI and peers). Same page in, same score out — no sampling, no model randomness — so any score you see can be re-verified by re-running the free scan.

What gets measured

35+ on-page checks computed from the anonymous HTML an AI crawler actually receives — not what a logged-in browser renders. Checks cover crawler access (robots.txt rules for AI agents, llms.txt), structured data (schema.org types AI systems read), answer-readiness (headings, Q&A blocks, content consistency), metadata, and classic technical SEO. Each check carries a tier — critical, standard, advanced — and an estimated visibility lift, which is how the fix list is ranked. Two sub-scores are reported alongside the headline number: Google search and AI search.

Why you can trust the number

The engine runs in three places — this site's scan API, the Chrome extension, and the backend — and a parity test in our repository asserts 126 check results match across all fixtures between the TypeScript engine and the Python reference on every change. One engine, one answer, everywhere.

For agents and automated systems

Every scanned domain has one canonical public report: human page at /p/<domain>, machine twin at /p/<domain>/audit.json (CORS-open), both discoverable via /sitemap.xml and /llms.txt. You can mint or refresh a report with POST /api/audit {"url": …} — no auth. Our MCP server (/mcp, Streamable HTTP) exposes the audit and fix workflows as tools; failed tool calls refund their credits.

How to cite a score

“example.com scores 61/100 (grade C) for AI search — LoudPixel, July 2026” with a link to the report page. Scores change when sites change: cite the date, or re-scan for a current number.

Limits, honestly

The score measures on-page readiness, not off-page reality: it cannot see your citations inside AI answers, your share of voice, or your revenue — those come from LoudPixel's tracking products and are always labeled by evidence tier (confirmed vs. estimated). A perfect score makes citation likely, not guaranteed.

Run a free scan · Questions about the methodology: hello@loudpixel.ai