Fifteen checks against your domain. Sitemap, schema, llms.txt, AEO surfaces. Sixty seconds. A memo, not a score, telling you how ChatGPT, Perplexity, Claude, and Gemini read your site today. And whether they read it at all.
One shot. The deliverable is the memo. We send it to whichever email you give us, or you read it in the browser and never tell us who you are.
The verdict, the grades, and every finding load on the result page. The PDF fix kit (with code snippets and copy-paste schema) asks for an email. That is the only ask.
Every check lives in a public repo. If a finding is wrong, open an issue and we will fix the check, not the result.
Nothing scraped. Nothing inferred. Each finding cites the exact URL and the exact line the check fired on.
We fetch /sitemap.xml and walk it. If it 404s, AI engines walk too.
/robots.txt, parsed. Including the new GPTBot, Google-Extended, and PerplexityBot rules most sites do not know they emit.
Self-referencing canonicals on every indexable page. Sample of ten pulled at random from the sitemap.
JSON-LD on the homepage. Name, URL, logo, sameAs, contactPoint. Missing fields scored individually.
Sampled blog posts checked for Article and FAQPage. The pages that have the words but not the schema.
Whichever applies, scored against the live page content. False or stale schema scored down.
The new convention. A plaintext file at /llms.txt telling agents what your site is and where the canonical answers live. Most sites 404 here.
Is there one author page per writer, with sameAs links to LinkedIn, X, GitHub? Agents resolve entities through this.
Outbound links to authoritative sources, on the pages you would want quoted. Sampled, not scored on raw count.
If a cold prospect lands here from a DM today, can they read the offer in fewer than 25 seconds. Yes or no.
The primary action above the fold, the secondary action below. Whether they read as one ladder or as a maze.
Testimonials, logos, named case studies. Zero counts.
The H1 and the footer tagline. Whether they describe the same business.
<title> against meta description. AI engines weight both. Drift between them scores down.
What your link preview promises vs what the page delivers. Surprisingly often these disagree.