Service · 1.000 domains tested
Generative engine optimization: what ChatGPT actually reads from your website
What ChatGPT, Perplexity and agents read from your site.
According to public descriptions, ChatGPT reads a website differently from a browser: structure is dropped, text is split into blocks, and links shift the cuts. In my simulation across 1,000 domains, the cuts landed elsewhere on 363 of 376 multi-chunk pages compared with a plain text conversion. I check what of this reaches your website and align it accordingly.
How an AI reads your page
What it covers
- llms.txt, ai.txt, entity map and agent-readable interfacesllms.txt of this domain ↗Verifiable
- Identity consistency across JSON-LD, llms.txt and DNSQuery the _sovp DNS record yourself ↗Measurable
- Rendering probe: what arrives without JavaScriptShown live in a call →Demo
- Quick check of infrastructure signalsTool on GitHub (MIT) ↗Verifiable
- AI crawler policy (robots.txt, ai.txt)AI policy ↗Verifiable
- Benchmarks across domain listsBenchmark report ↗
- Web infrastructure audit of 559 domains ↗
- Nearly 1 in 4 Cloudflare-protected domains block verification bots ↗
Verifiable: public source. Measurable: check it yourself with an independent tool. Demo: runs in my own systems and is shown in a call.
Built with
- Node
- Playwright
- html2text
- llms.txt
- JSON-LD
Connected to
GEO & AI Visibility
- Technical SEOGEO builds on technical SEO
- SEO Reporting
- Agentic SEOAgents need what AI search reads
- Custom SoftwareMachine-readable from day one
- AI Consulting & Automation
- Web Development & Operations
What you end up with
- Your AI visibility in numbers: for 10 to 15 customer questions I count how often ChatGPT, Perplexity and AI Overviews cite your website as a source.
- The rendering check: which content on your pages arrives without JavaScript.
- A work list with the points that make your site more readable for AI search.
What the list shows is implemented by the module for llms.txt, identity and AI crawler rules.
Who it is for
- Agencies: clients ask about ChatGPT and Perplexity, and you need a measurement instead of a guess.
- SaaS and software: your customers ask AI assistants for solutions, and you want to know whether your product is named there.
- Publishers and content teams: you want to know whether AI answers cite your articles as a source.
Frequently asked questions
What is generative engine optimization?
GEO stands for generative engine optimization. It makes sure that AI search tools such as ChatGPT, Perplexity or Google's AI Overviews read your website, understand it and name it as a source. Classic SEO aims at positions in the result list, GEO at mentions in the answer itself.
How do I know whether ChatGPT names my website?
I ask ChatGPT and Perplexity the questions your customers ask, usually ten to fifteen, and record which sources are named. In addition, I check what arrives from your site without JavaScript. Together they show whether the cause is content or technology.
What can I do myself right away?
Check three things: whether your key content is in the HTML without JavaScript, whether every page has a clear heading structure, and whether structured data (JSON-LD) describes your company. For the first point, your browser's page source view is enough: what is not there, a fetch without JavaScript does not see.
How much does GEO cost?
The entry point is a GEO measurement for €1,500: 10 to 15 customer questions in ChatGPT, Perplexity and Google's AI Overviews, rendering check and work list. Together with the agentic SEO measurement it costs €2,500. Implementation modules start at €960. All prices net. All prices →
How we work together
A first call costs nothing. Afterwards you know whether and how we work together.