Paste your website. In about ten seconds you get a 0 to 100 score on whether ChatGPT, Claude, Perplexity and Google AI Mode can fetch your site, read it, and cite it. Plus the exact fixes, in the order that recovers the most points.
Ordered by points you get back. Do the top three, re-run, repeat. That's the whole game.
| Check | Points | Result |
|---|
Download the PDF, drop it into ChatGPT, Claude or your website chatbot, and paste this:
Here is my AEO audit. Write the JSON-LD block, the meta description, the llms.txt file and rewrite my homepage opening paragraph so every item in "Fix these first" is done. Give me the exact code to paste into my site.
Paste what it gives you. Publish. Run the check again. Watch the number move.
When someone asks an AI "who's the best agent for waterfront homes in Rochester," the engine hands back one to three names. It picks from the pages it can actually load and read. If your site is a pretty JavaScript shell with no words on it, you're not losing the lead. You never existed.
We fetch your page the way OAI-SearchBot does. If your host or firewall throws a challenge at bots, you're invisible to ChatGPT, no matter how nice the site looks to humans.
No headless browser, on purpose. Most AI crawlers don't run JavaScript. We score what arrives in the raw HTML, because that's all the engine gets.
Structured data, an actual bio in actual paragraphs, phone, market, service area, freshness. The stuff that lets a machine say "this is JMan, he sells waterfront in Rochester."
Fetchability weighs the most because it's binary: a blocked page gets cited exactly as often as a blank one. Content heuristics weigh the least because they're proxies, not measurements.
Bot fetch, robots.txt for the nine crawlers that produce citations, server-rendered text, HTTPS, page size.
JSON-LD entity, extractable body copy, llms.txt, site depth, author and sameAs provenance, dateModified.
Title, meta description, sitemap, canonical, Open Graph, Twitter card, language. Google AI Mode is still search underneath.
Landmarks, one H1, a heading outline with no skipped levels, alt text. The skeleton a parser uses to find the answer.
Answer-first opening paragraph, real numbers, question headings, quotations. The two effects that replicated in the Princeton GEO research.
Phone, address, hours, service area. Scored only for businesses and people, so a blog post isn't penalized.
A is 90 and up. B 80. C 70. D 60. Below that, F. Score = round(100 × earned ÷ possible).
JMan trains brokerages and agents to actually implement this stuff. One expert, one tool, immediate implementation. Bring your report to a call and leave with a plan.
Book a free call See Big Brain Chatbots