Sample AI visibility report
This is a complete AI visibility report for northfield.example, a domain that does not exist. The figures are fixed so the page reads the same to everyone. A real report has the identical structure and arrives as a private link.
- AI crawler access1 of 7 blocked outright25/35
- robots.txt policydisallows an AI crawler5/10
- Renders without JavaScript642 words before JavaScript20/20
- Structured dataOrganization, WebSite6/12
- sitemap.xmlpresent8/8
- llms.txtmissing0/5
- agents.mdmissing0/3
- On-page hygiene2 issues4/7
The maximum is 95. This check reads public signals from outside your site, so the last five points are the crawl history only your own logs can show.
ClaudeBot gets a 403 on northfield.example.
AI assistants cannot read your pages. Every answer ChatGPT, Claude, or Perplexity gives about you is a guess.
One crawler is being turned away at the server, not by policy. ClaudeBot gets a 403 while every other probe gets through, which usually means a firewall rule rather than a decision anyone made about Anthropic.
AI crawler access
Five of seven probes returned 200. ClaudeBot returned 403, which is a hard block at the server or WAF and holds regardless of what robots.txt says. PerplexityBot is allowed through at the server but disallowed by name in robots.txt, so a compliant crawler will leave. Google-Extended is disallowed as well, which opts the site out of Gemini training and grounding without affecting Google Search.
| Crawler | Role | Live fetch | robots.txt |
|---|---|---|---|
| GPTBot | training | 200 | allowed (by name) |
| OAI-SearchBot | indexing | 200 | allowed |
| ClaudeBot | training | 403 blocked | allowed |
| Claude-SearchBot | indexing | 200 | allowed |
| PerplexityBot | indexing | 200 | disallowed (by name) |
| Googlebot (Google-Extended policy) robots token | training | 200 | disallowed (by name) |
| CCBot | training | 200 | allowed |
Present. HTTP 200. Looks hand-maintained.
No Content-Signal line (ai-train / search / ai-input).
A live 403 or 401 is a hard block at the server or WAF. It stops the crawler regardless of what robots.txt says. A robots.txt disallow is advisory. Well-behaved bots honor it, but the page is still reachable.
Machine-readability
The sitemap is present and served correctly. There is no llms.txt and no agents.md, so nothing on the site states in plain terms what it sells or who it serves. Structured data covers Organization and WebSite; Product and FAQPage are absent, and both are the types an assistant leans on when answering a question about what you stock.
Render check
The raw HTML carries 642 words before any JavaScript runs, which is enough for a crawler to quote from. The Next.js fingerprint is present, and it is serving real server HTML rather than an empty shell.
On-page hygiene
The title runs to 74 characters and will truncate in most surfaces. One heading level is skipped, which flattens the document outline a parser reads.
- Title tag74 chars
- Meta keywords tagabsent
- H1 headings1
- Heading hierarchy1 skip(s)
- og:localeen_US
What AI says about you
Assistants describe Northfield Supply as a long-running hardware retailer, and most of what they say is drawn from directory listings and two trade-press mentions rather than from the site itself. Product range, shipping terms and the trade account offer do not appear in any answer.
northfield.example ranks for about 412 keywords, best position #4.
This is the view from outside.
Everything above is what anyone can see by reading your site. It doesn’t show who actually crawled you, how often, or which pages they took. That lives in your logs. Kymo reads them for you and shows the AI crawlers hitting your site in real time, always free, never counted against your limits.
See the live version