Glossary
Short, precise explanations of GEO/AEO terms. Where we have a detailed, sourced article on the topic, the entry links to it.
- GEO (Generative Engine Optimization)
- Optimizing a website so generative AI systems (ChatGPT, Perplexity, Gemini, AI Overviews) can read and cite it in their answers. Unlike classic SEO, success isn't measured by ranking position but by whether and how a site shows up in text a model generates.
- AEO (Answer Engine Optimization)
- A close relative of GEO that emphasizes answering specific questions, that is phrasing content so an assistant can use it directly as an answer. In practice, GEO and AEO are often used interchangeably; Torumata uses "AI search readiness" as an umbrella term for both.
- llms.txt
- A markdown file at the root of a site (
/llms.txt) giving language models a short overview: a heading, a summary, and links to key pages. According to primary sources, Google Search doesn't use it, and neither OpenAI nor Anthropic declares reading it; coding tools do demonstrably fetch it. Details in the dedicated article. - llms-full.txt
- An extended version of
llms.txtcontaining the full text of linked pages in a single file. It's not part of the corellmstxt.orgspecification; it's a convention introduced by tooling aroundllms_txt2ctx(FastHTML), not an official standard. - AI Overviews
- A Google Search feature that shows a generated summary with supporting links above the classic results. Per Google's documentation, the only requirement for a page to be eligible is that it's indexed and eligible for regular search with a snippet; no additional technical requirements are stated.
- AI crawler
- A program that fetches web pages on behalf of an AI system: for model training, for a search index (e.g.
OAI-SearchBot,PerplexityBot), or to fetch a page on a specific user's request (e.g.ChatGPT-User).GPTBot(OpenAI),ClaudeBot(Anthropic) andPerplexityBot(Perplexity) are among the best known; a full overview is in AI crawlers in 2026. - Indexability
- A page's ability to be included in a search index: it must not be blocked by
robots.txt, must return a 200 status code, and must not carry anoindexdirective. Per Google's documentation, this is a necessary condition for inclusion in AI Overviews and AI Mode too, not just classic results. - JSON-LD
- A structured-
data format (JSON for Linking Data) that describes a page's content machine- readably using the schema.org vocabulary: page type, author, publication date and other attributes. Google uses it to qualify pages for rich results in classic search; whether AI assistants read it or whether it increases citation rates is not documented. - Citation / source in an answer
- A link to your site that appears as a supporting source in an AI assistant's generated answer or an AI Overview. The field's one peer-reviewed experiment (Princeton, KDD 2024) measured what increases citation odds: in-text citations (+41%), concrete statistics (+33%), and links to sources (+28%).
- robots.txt
- A text file at the root of a site (
/robots.txt) declaring which crawlers may or may not access which parts of the site. Most AI crawlers respect it, but not all:Perplexity-andUser Meta-documentedly may bypass it, andExternalFetcher Bytespiderrespects it unreliably. - Sitemap (sitemap.xml)
- An XML file listing the URLs a site owner wants offered for indexing, optionally with last-modified dates. It helps crawlers find pages that aren't well linked internally, but being listed in a sitemap guarantees neither indexing nor citation.
- Render (SSR vs. JS)
- How a page's content ends up in the HTML a crawler sees. Server-side rendering (SSR) and static generation put content directly into the initial HTML response; with client-only JavaScript (an SPA without SSR), content is added only after the page loads in a browser. Per Vercel's analysis (December 2024), seven of nine tracked AI crawlers don't execute JavaScript at all, so that content doesn't exist for them. Exceptions: Gemini and Applebot render.
- Credit
- The unit Torumata bills paid audits in; 1 credit ≈ €1. Each audit step's price is a flat fee and/or a rate for every batch of pages it actually processes; you're billed for the real consumption, and unused credits stay on your account. The result is rounded up to whole credits once per purchase: once for a whole bundle, or per step when you run a step on its own. That is why a bundle never costs more than the same steps run one by one. Credits are topped up directly inside the app after the free audit.
- Scope estimate
- An estimate of a site's page count that Torumata computes from several sources at once (sitemap, an exploratory crawl with extrapolation, and optionally the customer's own figure) during the free audit. When the sources disagree significantly, that disagreement is itself a finding about the site, not just a pricing detail. The estimate is what determines how many credits a full-site audit would cost.