GEO glossary
By the HowsMyGEO team · Published 2025-11-03 · Updated
The vocabulary of AI visibility, defined plainly. These are the terms our scan reports use.
- GEO (Generative Engine Optimization)
- GEO is the practice of making a website findable, readable and citable by AI answer engines such as ChatGPT, Claude, Perplexity and Google AI Overviews. Where SEO optimises for ranked lists of links, GEO optimises for being quoted inside generated answers. Core GEO work includes allowing AI crawlers in robots.txt, server-rendering content, publishing llms.txt, and structuring content so it can be lifted verbatim into an answer.
- llms.txt
- llms.txt is a plain-text file served at a site's root (/llms.txt) that describes the site and its most important pages for large language models, in Markdown. Proposed in 2024 by Jeremy Howard, it is an emerging standard: an AI-oriented counterpart to robots.txt that tells models what a site is about rather than what they may crawl.
- GPTBot
- GPTBot is OpenAI's web crawler. It gathers content used by ChatGPT and OpenAI's models. A robots.txt rule of 'User-agent: GPTBot / Disallow: /' removes a site from that pipeline, which eliminates the site's chance of being cited by ChatGPT's browsing and search features.
- ClaudeBot
- ClaudeBot is Anthropic's web crawler, used to gather content for Claude. Like GPTBot, it obeys robots.txt, so an accidental disallow rule silently removes a site from Claude's visibility.
- PerplexityBot
- PerplexityBot is the crawler for Perplexity, an AI answer engine that cites its sources prominently. Because Perplexity links every answer to the pages it cites, being crawlable by PerplexityBot converts directly into referral traffic.
- Google-Extended
- Google-Extended is a robots.txt token that controls whether Google may use a site's content for its Gemini models and AI features. Blocking it does not affect classic Google Search rankings, but it reduces the site's presence in Google's generative AI experiences.
- CCBot
- CCBot is the crawler of Common Crawl, a nonprofit that publishes a free archive of the web. Many AI models are trained on Common Crawl data, so blocking CCBot reduces the chance of a site being represented in future model training sets.
- AI Overviews
- AI Overviews are the generated summaries Google shows above traditional results for many queries. They synthesise and cite web sources; sites that are extractable and factually dense are more likely to be linked inside them.
- Content extractability
- Content extractability is whether a page's main content exists in the HTML the server sends, rather than being assembled in the browser by JavaScript. Most AI crawlers do not execute JavaScript, so a client-side-only page reads as empty to them — the single most damaging GEO failure.
- Answer-shaped content
- Answer-shaped content is writing structured the way answers are: a question as a heading, a direct one-paragraph response, definition sentences ('X is a…'), and FAQ schema. AI engines preferentially lift this shape of content into their answers because it requires no re-assembly.
- Structured data (JSON-LD)
- JSON-LD is machine-readable metadata embedded in a script tag, using schema.org vocabulary, that declares what a page is: an Article, a Product, an Organization, an FAQPage. Search engines use it for rich results; AI engines use it to ground entities and attributions.
- TTFB (Time to First Byte)
- TTFB is the time between requesting a page and receiving the first byte of the response. It measures server responsiveness; under 500ms is good. Slow TTFB wastes crawl budget and correlates with poor user experience metrics.