Glossary

AEO — Answer Engine Optimisation. Optimising to be the source of a generated answer rather than a ranked link. Used more or less interchangeably with GEO.

Alt text. The text alternative for an image. How a text-based model perceives your visual content. See alt text and media.

Answer engine. Any system that responds to a question with a synthesised answer instead of a list of links. ChatGPT, Perplexity, Google's AI Overviews.

ClaudeBot. Anthropic's web crawler.

Chunking. Splitting a document into passages that can be retrieved independently. Your headings and paragraph structure determine your chunks, whether you meant them to or not.

Citation. A named source attached to a generated answer. The unit of success in AI visibility.

Entity. A distinct thing a knowledge system recognises — a business, a person, a place. Being a clear entity is why consistent naming matters. See entity clarity.

GEO — Generative Engine Optimisation. The practice of improving how often and how favourably a brand appears in AI-generated answers.

Google-Extended. A robots.txt token controlling Google's use of your content in Gemini and AI training. It does not affect Google Search ranking.

GPTBot. OpenAI's crawler for model training. Distinct from OAI-SearchBot, which fetches pages for live search.

Grounding. Anchoring a generated answer in retrieved source material rather than model memory. The stage where most pages get discarded. See how assistants choose sources.

Hallucination. A confidently stated but false output. Grounding reduces it; being clearly stated on your own site reduces the chance a model invents your details.

JSON-LD. The recommended format for structured data — a script block of JSON in your page. See JSON-LD for AI.

Knowledge graph. A structured store of entities and their relationships. Your sameAs links help place you in one.

llms.txt. A proposed markdown file at your website root summarising your website for language models. See llms.txt explained.

RAG — Retrieval-Augmented Generation. The general pattern of searching for documents and then generating an answer from them. The technical name for what most assistants do when they answer a current-events question.

Retrieval. The search step that pulls candidate documents before an answer is generated.

Rich result. An enhanced search listing produced from structured data.

Schema.org. The shared vocabulary used by structured data, maintained collaboratively by the major search engines.

Server-side rendering (SSR). Producing complete HTML on the server so content exists without running JavaScript. Strongly preferred for AI visibility.

Share of voice. How often your brand is named relative to competitors across a set of prompts. Measured by sampling, not census.

Structured data. Machine-readable facts embedded in a page, usually as JSON-LD.

User-agent. The identifying string a crawler sends. What you target in robots.txt rules.