Six pillars. One reading.
No proprietary black box. No vanity score. The method below is the entire system — what we read, how we weigh it, and what we hand back.
Schema integrity
Organization, Product, FAQ, BreadcrumbList, Article — validated against schema.org and the latest generative-engine guidance.
Entity clarity
Does the site declare what it is, who runs it, and how it relates to other named entities? We score sameAs, knowsAbout, and disambiguation density.
Document structure
Heading hierarchy, semantic landmarks, table-of-contents signals, and the answer-first patterns LLMs prefer when summarising.
Extraction quality
We render the page the way ChatGPT, Gemini, Claude and Perplexity render it — then measure what survived.
Citation surface
Author bylines, source links, original data, quotable lines. Pages with citation surface get quoted; pages without get paraphrased.
Freshness signal
datePublished, dateModified, sitemap lastmod, and the cadence between them. Stale sites lose to fresh ones — even at lower authority.
- Phase 01
Read
We render up to 250 of your most-trafficked pages with a headless engine that mimics OpenAI's, Anthropic's and Perplexity's crawlers.
- Phase 02
Score
Each page passes through the six pillars. Pillar scores roll up into a 0–100 composite, weighted by how each engine actually cites in your category.
- Phase 03
Prescribe
We hand back a ranked list of fixes — the smallest changes with the largest expected lift on AI visibility. Most are one-line edits.
- Phase 04
Re-read
Run the audit again in thirty days. We diff the result and tell you which fixes moved the needle. No fixes, no charge for the re-read.
Read the score on your own site.
One audit is free. No card, no onboarding call, no sales follow-up.
Begin audit