# Free GEO Audit: See How AI Search Reads Your Website | Koso > Run a free GEO audit on any website. We fetch it as GPTBot, the way ChatGPT and Perplexity do, score it out of 100 and list what to fix first. [Skip to content](#main) [](/) [Services](/services)[AI OS](/ai-os)[Training](/training)[Education](/education)[Agents](/agents)[Learn](/learn) … [Book a call ](/book) 1. [Home](/) 2. [AI Agents](/agents) 3. GEO Audit # Free GEO Audit: Can AI Search Read Your Site? This GEO audit fetches your website as GPTBot, exactly the way ChatGPT and Perplexity do, then scores what comes back out of 100 and tells you what to fix first. Check site See an example:koso.aistripe.comvercel.com Free to use. Results in about 30 seconds. ## Want An Agent Like This For Your Business? This tool is a multi-agent pipeline Koso builds for clients every week, pointed at websites instead of your invoices, inbox or ops. Imagine it pointed at yours. [Book a free call ](/book)[All agents ](/agents) ## We Don't Simulate An Agent. We Send One. Here's what most GEO audit tools miss: an AI engine reading your website is a plain HTTP request that never runs a line of JavaScript. So this audit doesn't model what an agent would see, it fetches your site the way an agent does and reports what came back. Five nodes, one of which is deliberately not an AI at all: true: scanned recently, answered instantlyfalseresult saved for 24 hourswebsite inputyou type a site aboveIFcache checkscanned in last 24h?access-probe3 fetches, 3 user-agentssurface-scanllms.txt, agent cardscontent-readerpages, text, actionsrubric100-point scorereport-writerprioritised fixesGEO resultscore · band · fixesas GPTBot/.well-known/\*page fetch x4{ }pure code, no AIAIAI modelresults cache Swipe the diagram sideways to follow the full flow. The actual pipeline behind every scan. Dashed wires are the tools and data each node pulls in. Note the fourth node: the score is computed in code, not by a model, which is why it is reproducible. ## What The GEO Audit Checks. 30 checks across 6 categories, weighted by how much each one actually costs you when it fails. The scoring is done in code rather than by a model, so the same site scores the same twice and you can re-run it after a fix and trust the difference: ### Crawler access 27 Can an AI crawler load your pages at all? We fetch as GPTBot and compare against a browser, then read robots.txt for blanket blocks, named AI crawler rules, noindex tags and HTTPS. This is where most sites lose points without knowing it. ### Content extractability 22 What survives a fetch with no JavaScript. Real text in the raw HTML, a sane text-to-markup ratio, semantic landmarks like main and article, and whether a crawler gets the same content a browser does. ### Machine-readable structure 18 JSON-LD structured data, a crawlable sitemap, a title and description worth quoting, one clear H1 and a canonical URL. The facts an agent can take from you without inferring them from prose. ### Agent discovery surface 15 How an agent finds out what's on offer without crawling for it: llms.txt, Link headers, the A2A agent card, an Agent Skills index, an ARD manifest, DNS-AID records and a Content-Signal policy. Almost nobody has these, which is exactly why they're worth points. ### APIs, auth and tools 10 Whether an agent can call you rather than just read you. An MCP server card, an OpenAPI description or API catalog, WebMCP tools in the page, OAuth and OpenID discovery, protected resource metadata, auth.md and a Web Bot Auth key directory. Most of these only matter if you have an API, and the fixes say so. ### Can an agent act 8 Whether an agent sent to do something could finish. Key pages at conventional paths, a contact route that exists in the HTML rather than mounting in JavaScript, and a plain-text or Markdown version for machine readers. ### What the bands mean * Invisible (0 to 39): AI agents effectively cannot use this site. * Emerging (40 to 64): Readable, but missing most of the agent surface. * Agent-ready (65 to 84): Reachable, readable and structured. * Agent-optimized (85 to 100): Built for agents on purpose. The bands are unforgiving at the top on purpose. Most well-run sites land in Emerging, because the discovery surface that separates a readable site from an agent-ready one barely existed two years ago. Grading on a curve would make a good score meaningless. ## GEO Audit FAQ ### What is a GEO audit? GEO is generative engine optimization: getting your site used as a source when AI engines like ChatGPT, Perplexity, Claude and Google's AI Overviews answer a question. A GEO audit checks whether those engines can actually use your site. Most of them fetch plain HTML and never run JavaScript, so a site that looks perfect to a person can be close to empty to them. This audit measures whether they can reach your pages, read the content, understand what you offer and act on it. ### What's the difference between GEO and SEO? They share foundations like crawlable pages and structured data, but the reader is different. SEO is about ranking in a list of links that a person then clicks. GEO is about being the source an AI answer is built from, often with no click at all. The crawlers differ too: GPTBot, OAI-SearchBot, PerplexityBot and ClaudeBot mostly don't render JavaScript the way Googlebot does, and CDNs often block them by default. That's why a site can rank well on Google and still be invisible in ChatGPT, which is exactly the gap this audit looks for. ### How does this GEO audit actually test my site? It fetches your site the way a crawler does, and that is the whole trick: an AI agent reading a website is a plain HTTP request with no JavaScript. We request your homepage three times, once as a normal browser, once as GPTBot and once asking for Markdown, then compare the answers. If your CDN refuses our GPTBot, we ask OpenAI to open the page from its own network, which tells a real block apart from spoof protection. We also read robots.txt, look for llms.txt, your sitemap, agent and MCP cards, and try the conventional paths agents guess at, like /pricing and /contact. ### Is the GEO score generated by AI? No, and that's deliberate. The 100-point score comes from a fixed rubric in code, so the same site scores the same twice and every number traces back to a real HTTP response. AI writes only the explanation and the ordering of the fixes. A score a language model invented would not be something you could act on or re-measure. ### My site scores badly but looks fine in a browser. Why? Almost always one of two reasons. Either your pages are rendered by JavaScript in the browser, so the raw HTML an agent receives is an empty container, or your CDN's bot protection is refusing AI crawler user-agents while letting people through. Both are invisible from a browser and both make you unreadable to AI search. The report tells you which one you have. ### Should I block AI crawlers instead? That is a real choice, and this tool does not pretend otherwise. But it is rarely all-or-nothing. The crawlers that gather training data are different from the ones that fetch a page because a user asked about you right now. Blocking OAI-SearchBot, ChatGPT-User or Claude-User removes you from answers people are actively asking for. Blocking GPTBot only opts you out of training. Content-Signal in robots.txt lets you state the terms without blocking anyone. ### What is llms.txt and do I need one? It's a short Markdown file at your site root that lists your most important pages with a line of description each: a curated map for models, the way a sitemap is a map for search engines. It is not an official standard, it takes about twenty minutes to write, and it is the highest-value item on this list relative to effort. Yes, you should have one. ### How is this different from other GEO audit tools? Most free GEO tools parse your robots.txt and tell you which crawler names appear in it, or ask a model to guess how visible you are. Reading robots.txt is one check out of the 30 here. This one actually fetches your site as GPTBot and compares what it gets to what a browser gets, which is how it catches silent CDN blocking and cloaking that a robots.txt reader cannot see. It also scores the agent surface (MCP server card, A2A agent card, OpenAPI, OAuth discovery, WebMCP, Agent Skills) that separates a site an agent can read from one it can use, benchmarks your score against every site the checker has scanned, and shows you the full pipeline it ran. ### Should I run a GEO audit or an AEO audit? Start with this one. The GEO audit checks your whole site's plumbing, from whether AI crawlers can get in at all to whether your llms.txt and structured data exist. Koso's AEO page audit goes one level down and checks whether a single page's content is written in a way an AI answer would quote. If your GEO score is under 65, fix that first, because a page an AI engine can't read won't get quoted however well it's written. ### What is a good GEO score? Above 65 is Agent-ready and above 85 is Agent-optimized, and the bands are deliberately hard: most well-run sites land in Emerging (40 to 64), because the agent discovery surface barely existed two years ago. To make your number concrete, the score card also shows the share of every site this checker has scanned that yours beats, and when you re-scan after fixes it shows exactly how many points you moved since the last scan. ### Can I put my score on my site? Yes. Every scan gets a live SVG badge showing the score out of 100, colored by band. The Embed badge button copies a one-line HTML snippet for your README or footer, and the badge links back to your full report so anyone can verify it. Re-scan after improvements and the badge updates within a day, no code change needed. ### Is this free, and how often can I run it? Free. Your first scan needs no sign-in; after that we ask for a name and email, no password, no credit card. Up to 10 scans an hour from one address, and a repeat scan of the same site inside 24 hours is answered instantly from the saved result. If you've just fixed something, hit Check again now on the result to skip the saved copy and scan it fresh. ## Try The Other Agents. Every tool here is a live multi-agent pipeline, free to run, reasoning on display. [All agents ](/agents) [LiveDomain AppraisalSix agents value any domain: real web history, a 1000-point score, real sales, retail and wholesale prices.Appraise a domain ](/agents/domain-appraisal)[LiveCompany ResearchSix agents build a dossier on any company and benchmark it against its rivals.Research a company ](/agents/company-research)[LiveAEO Page AuditChecks whether ChatGPT, Perplexity and Google's AI answers would quote one of your pages, asks ChatGPT, and scores it out of 100.Audit a page ](/agents/aeo-audit)[LiveProduct Market FitA manager agent hires analyst agents to test whether your product fits its market.Analyze your product ](/agents/product-market-fit)[LiveAgent Workflow GeneratorDescribe the agent you want. It asks 2 or 3 sharp questions, then draws the workflow and writes the build spec.Design your agent ](/agents/workflow-generator)[LiveICP GeneratorReads your real pricing page and job board, then builds the ideal customer profile: firmographics, the buying committee, triggers, and who to say no to.Build your ICP ](/agents/icp-generator)[LiveProspect ResearchGive a name and a company. Agents read the public professional web and write your call-prep note: who they are, how to talk to them, and what to open with.Research a prospect ](/agents/prospect-research)[LiveAI Skill BuilderDescribe a capability and it writes you an installable SKILL.md for Claude Code, Cursor and Copilot, after showing you what already exists on skills.sh.Build a skill ](/agents/skill-builder)[LiveCan I Run This LLMReads your graphics card from the browser, then measures every popular open model against it: what runs, how fast, and the command to pull it.Check your PC ](/agents/can-i-run-this-llm)[LiveAI Slop DetectorRenders your landing page in a real browser, boxes every AI-template tell on a screenshot and scores it out of 100.Scan your page ](/agents/ai-slop-detector)[Livellms.txt GeneratorReads your sitemap and pages, sorts them into sections and writes the llms.txt file AI models look for. Edit it, then download.Generate your llms.txt ](/agents/llms-txt-generator)[Coming soonBuild Your Own AgentSoon you'll design an agent like these yourself, right here. Sign in and you're on the list for early access.Get on the list ](/login) A custom AI development agency. Bespoke agents, copilots and products engineered for teams that want a moat, not a subscription. Ideanab Private Ltd is a company registered in England and Wales, no. [16344048](https://find-and-update.company-information.service.gov.uk/company/16344048). Registered office: 124 City Road, London, EC1V 2NX, United Kingdom. ### Company * [Services](/services) * [Custom AI Agency](/custom-ai-agency) * [AI Product Development](/ai-product-development) * [AI OS](/ai-os) * [AI Engineering](/ai-engineering-company) * [AI Consulting](/ai-consulting-company) * [AI Consulting Services](/artificial-intelligence-consulting-services) * [Training](/training) * [Education](/education) * [Free AI Agents](/agents) * [Learn AI](/learn) * [Guides](/guides) * [Fractional CAIO](/caio) * [About Us](/about) * [The Koso Deck](/deck) * [Careers](/careers) * [Docs for agents](/docs) ### Contact * [Contact us](/contact) * [Book a call](/book) * [LinkedIn](https://www.linkedin.com/company/koso) * [Instagram](https://www.instagram.com/okaashish) ### Legal * [Terms of Service](/terms) * [Privacy Policy](/privacy) * [Cookie Policy](/cookies) * [Acceptable Use Policy](/acceptable-use) * [KosoBot](/bot) © 2026 Koso, a brand of Ideanab Private Ltd. All rights reserved.