SEO & AI Auditor
Can AI read your site and recommend you?
Paste your website address and in less than a minute you will know whether Google, ChatGPT and AI agents can read it, understand it and cite you. What’s included
Four scores —five if you have premises—, the five fixes that recover the most points and, if you want, the same measurement on your competitors. Free, no sign-up, and nothing of yours is kept.
What it does, and what it cannot do
The auditor downloads the URL you paste —your home page or any specific page— exactly as Google would —no browser, none of your scripts executed— and scores it across 33 checks. Four scores come out of that. If your site declares premises, it also measures 8 Local checks, which are shown but do not count towards the score yet. And if you analyse a single site, it also downloads a sample of up to 50 pages from your sitemap, one at a time and with a pause, and checks the headers of another 100: that is where your site score comes from, not just that URL's.
What matters is what it does not measure. Checking where you actually rank, whether Perplexity really cites you, or who mentions you outside your domain takes live searches, and none of that happens here. Those checks are not invented: they stay out of the score —they neither add nor subtract— and they are listed one by one under your scores. Each score also carries a bar telling you what share of it was actually measured.
A free tool that tells you it has measured 100 % of your GEO is lying to you. This one gives you the exact number.
What people ask before using it
What exactly does the SEO & AI Auditor measure?
It downloads the page you give it —your home page or any specific URL on your site—, the way a search engine would, and scores it across 33 checks grouped into four scores: SEO (whether Google and Bing can reach it, read it and understand it), search answers or AEO (whether they use you to answer a question and not just to list you), on-site GEO (whether ChatGPT, Perplexity, Claude or Gemini can read you and cite you) and Agent-Ready (whether an AI agent arriving on a customer's behalf can do anything useful with your site). If your site declares physical premises, it also measures 8 Local checks —address, phone, opening hours and coordinates— that are shown but do not count towards the score yet. And if you analyse a single site, it also downloads a sample of up to 50 pages from your sitemap to give you a score for your site, not just that URL.
Can I analyse a specific page instead of the home page?
Yes. Paste the exact URL —an article, a product page, a landing page— and the four scores and the 33 checks come from that URL's HTML, not from your home page. Three things are still domain-wide, because they cannot be anything else: robots.txt, sitemap health, and the «your site» score, which comes from a sample of up to 50 pages. And real-world speed is requested for that specific URL first; if Google does not have enough visits for that page, the domain average is used and that signal counts slightly less.
Do I need to sign up or pay?
No. It is free, with no account and no email. There is only a daily cap on analyses so the tool stays sustainable, and a per-connection limit so nobody can take the whole thing.
What data is kept about my site?
The analysis and its result are deleted after 24 hours. Your IP address is not stored: the rate control uses a hashed fingerprint that is never saved anywhere. Your site's content is not sent to any AI model.
Why do some checks say they could not be measured?
Because checking them means searching Google, asking an AI assistant live or judging the content page by page, and none of that happens here. Those checks stay out of the score —they neither add nor subtract: the score comes only from what was actually verified— and are listed one by one so you know exactly what was measured and what was not. Each score has a bar telling you what share of it was actually measured.
Why do the comparison table scores differ from my individual score?
Because in the comparison every site plays the same match. If a check cannot be measured on one of the compared sites, it is set to neutral on all of them: otherwise it would look as if one site scored worse when what actually happened is that the other could not be measured. Your individual score, at the top, is the real one.
What is Agent-Ready?
It is my own measure, not an industry standard, based on a seven-point checklist I published on the blog. It measures whether an AI agent browsing on a customer's behalf can read your content, find the data people decide with, and get to the end without running into a captcha or a PDF. I explain it in full in this article.
Want someone to fix it?
I built this tool end to end: the scoring engine, the backend, the cost control and the legal layer. I do the same —and the work of fixing whatever comes out red— for agencies and hotels.
See services