robots.txt10 crawlers checked
GPTBotAllow
OAI-SearchBotAllow
ClaudeBotAllow
PerplexityBotAllow
Google-ExtendedAllow
CCBotBlocked
Discoverability
Whether ten named AI crawlers can reach you at all — and whether they're served the same page you are.
Can AI find, read and quote your website?
Buyers ask ChatGPT, Claude and Perplexity before they ever open Google. This audit runs 79 deterministic checks against your site and shows exactly what those assistants see — with a scored fix plan and the snippets to implement it. No account, no sales call, results in seconds.
Ranking on Google no longer means being in the answer.
An assistant doesn't browse. It fetches a few URLs, parses whatever HTML comes back, and decides in milliseconds whether there's a quotable answer on your page. Three things reliably mean there isn't:
- Your content renders client-side. The crawler is handed an empty page.
- Your identity exists only in a logo. There's nothing machine-readable to name.
- Your bot protection serves crawlers a challenge page. They never reach you at all.
None of it shows up in a rankings report. It shows up as an absence — you're simply never named.
Six areas. 79 signals. Every one with its evidence.
Signals that can't apply to your site are marked not-applicable with a reason and excluded from the grade, so the score reflects your site rather than a generic checklist.
<head>
title58/70
description142
canonicalself
og:imagemissing
Page foundations
The head tags and heading structure every engine parses before it decides what your page is about.
JSON-LDvalid
{
"@type": "Organization"
"name": "Northwind"
"url": "https://…"
"logo": "/logo.png"
"sameAs": [ … ]
}
Structured data
The machine-readable version of your site. Without it, an assistant is guessing who you are.
Your page
How long does installation take?
A standard cell is installed in 3 days.
AI answer
“A standard cell is installed in 3 days.”
yoursite.com
Answer extractability
Whether there's a passage an assistant can lift as a clean, attributable quote — and cite you for.
Entityresolved
Northwind Goodsnorthwind.example
✓sameAs
LinkedInGitHubCrunchbase
Entity clarity
Whether the site resolves to one real, nameable organization an engine is willing to name.
68%primary content
Content quality
Whether the substance is real content or navigation chrome wearing a page's clothes.
How it works.
- Step 01
Four requests, run together.
Your homepage, robots.txt, sitemap.xml and llms.txt, fetched in parallel from a server — the same way a crawler arrives. Then a budgeted second wave: the declared sitemap, your About/Contact/FAQ pages, and one re-fetch identifying as GPTBot.
- Step 02
79 deterministic signals, scored on the ones that apply.
No model is asked for an opinion. Every result is a measurement with cited evidence. Signals that can't apply to your site — Product schema on a site with no products — become N/A with a reason and leave the denominator instead of quietly failing you.
- Step 03
A fix plan, not a list of complaints.
Related failures are grouped into root issues with combined point values, an estimated score after the fix, and copy-paste JSON-LD, robots.txt and llms.txt snippets. The full report — every finding, every fix — goes to your inbox.
What it deliberately does not do: run a headless browser, call Lighthouse, ask a language model for an opinion, or crawl your whole site. It reads your homepage and up to three high-value pages — the same shallow pass an answer engine makes.
Questions about the audit.
What it measures, how the score works, and where its limits are.
Want the fixes shipped?
The audit tells you what's broken. If you'd rather have it fixed than filed, tell us what you're building — we reply within two business days.
Start a conversationWhat is AEO, and how is it different from SEO?
AEO is answer engine optimization: making your site legible to the assistants people now ask before they search. SEO optimizes for a ranked list of blue links; AEO optimizes for being the source an assistant quotes inside its answer. The overlap is real but partial — a site can rank well on Google and still be invisible to ChatGPT, usually because its content arrives only after JavaScript runs, because it has no structured identity, or because its bot protection serves crawlers a challenge page.
What exactly does this audit check?
79 deterministic signals across six areas — discoverability, page foundations, structured data, answer extractability, entity clarity and content quality — plus 10 per-crawler diagnostics covering OAI-SearchBot, GPTBot, ClaudeBot, PerplexityBot, Google-Extended, CCBot, Meta-ExternalAgent, Applebot-Extended, Amazonbot and Bingbot. Every signal reports the evidence it measured, so you can verify any conclusion yourself.
How is the score calculated?
100 points are apportioned across four buyer questions by explicit importance weights: can AI find you (25), read you (20), understand who you are (30), and quote you (25). Within each group the points are split across the signals that actually apply to your site. Score plus warnings plus fixes always sum to exactly 100, so every recommendation carries an honest point value rather than a vague severity label.
Does it use AI to grade my site?
No. There is no language model anywhere in the audit, and no headless browser or Lighthouse call either. Every check is a deterministic measurement over the HTML, headers and text files your server returns. Two runs against an unchanged site produce the same score, which is the point — a grader that hallucinates is worse than no grader.
Why do you ask for a work email?
Two reasons, both honest. The audit makes outbound HTTP requests on your behalf, and the email gate is the main thing stopping it from being used as a free scanning service. And it is lead capture: this tool is free because some of the people who run it will eventually want the fixes shipped rather than listed. We send the report you asked for. We do not sell the list.
My site is behind Cloudflare. Will the audit work?
It will run, and it will tell you if it was blocked. Bot protection often serves datacenter IPs a challenge page instead of your site — and AI crawlers come from datacenters too. If we get a challenge page, that is itself the finding, reported with lower audit confidence rather than dressed up as a score.
How accurate is this compared with a manual audit?
It reads what a crawler reads on your homepage plus up to three high-value pages, in about a second. That catches the structural problems that keep sites out of AI answers, which is most of them. It does not read your whole site, judge whether your content is any good, or know your market. Treat the score as a triage instrument, not a verdict on your business.
What happens to my results?
The full report goes to the address you entered. We keep a record of the audit — the domain, the score and your email — because that is what makes the tool worth running for us. Results for a given domain are cached for an hour, so a repeat run returns the same report instantly. See our Privacy Policy for the details.
A score is a starting point. Shipping the fixes is the work.
We build and operate the software and brands behind this studio, and we run the same audit on our own properties. If you want a senior team to take the fix plan and execute it, that's the conversation.