AI search & GEO glossary
Plain-English definitions of the terms used in AI search optimization, GEO, AEO and LLM SEO — mention rate, citation sources, entity consistency and more.
AI agent
An AI agent is a program that plans its own steps, calls tools and adjusts based on results to reach a goal, rather than just generating a block of text.
AI hallucination
AI hallucination is when a language model produces fluent, confident text that is factually wrong, because it predicts likely words rather than checking facts.
AI search optimization
AI search optimization is the umbrella term for GEO, AEO and LLM SEO: work that gets a brand mentioned, recommended and cited accurately in AI-generated answers.
AI search platform
An AI search platform answers questions conversationally and may cite source links, replacing the list of blue links as the place people get conclusions.
AI visibility
AI visibility is how often and how accurately a brand appears in answers from AI assistants such as ChatGPT, measured with a fixed question set and mention rate.
AI visibility audit
An AI visibility audit measures how AI answers describe your brand today and pinpoints whether problems sit in the entity, content or crawl layer.
Answer engine optimization (AEO)
Answer engine optimization (AEO) is optimizing content for engines that answer directly, so your content becomes the answer itself or a cited source for it.
Answer position
Answer position is where a brand ranks in the list order of an AI answer — the middle metric between mention rate and first recommendation rate.
Answer-first writing
Answer-first writing puts the conclusion in the first sentence and the reasoning after it, so readers and machines can take a complete answer from the top.
Baseline and retest
A baseline retest measures AI answers with a fixed question set before optimizing, then repeats the exact same method later and compares the difference.
Brand entity
A brand entity is the record a machine uses to treat a brand as one uniquely identifiable object, not just a string of characters that appears in text.
Brand knowledge base
A brand knowledge base is one master document holding a company's facts, definitions and standard answers, so every external piece is rewritten from it.
Canonical link
A canonical link (rel="canonical") declares which URL in a set of identical or near-identical pages is the official version, consolidating signals onto it.
Category keyword
A category keyword describes a type of need without naming a brand, such as "GEO services" — AI answers to these questions pick and list brands on their own.
Citation source
A citation source is a page an AI retrieves and credits when answering with web search. It decides whose version of the facts ends up in the answer.
Content factory
A content factory produces content at volume with a fixed process plus AI tools, splitting topics, writing, review and distribution into repeatable steps.
Content farm
A content farm mass-produces low-value content purely to capture traffic, mostly stitched together and reworded, without adding any new verifiable information.
Crawl budget
Crawl budget is how much a crawler will fetch from a site over time, set by what the server can handle and whether the content is worth re-crawling.
Decision query
A decision query is a question asked while choosing — "how do I choose," "which one fits," "is it worth it" — and the answer shapes the shortlist and its order.
Doubao
Doubao is ByteDance's AI assistant, offering chat and web search, and one of the platforms most often measured when checking a brand's AI visibility in China.
E-E-A-T
E-E-A-T is the four page-quality dimensions in Google's rater guidelines: experience, expertise, authoritativeness and trust, with trust at the center.
First recommendation rate
First recommendation rate is the share of test answers where an AI recommends your brand first: whether you come to mind first, not just at all.
Forward deployed engineer (FDE)
A forward deployed engineer (FDE) works on site with a client to fit a product or AI capability into real workflows, combining engineering with process work.
Generative engine optimization (GEO)
Generative engine optimization (GEO) improves how AI answers describe, mention and recommend a brand, measured with fixed questions and retests.
Knowledge graph
A knowledge graph stores facts as entity–relationship–entity triples, so machines can reason along connections instead of only matching strings of text.
LLM SEO
LLM SEO is optimizing a brand's content and entity signals so large language models like ChatGPT mention it accurately and cite its pages in their answers.
llms.txt
llms.txt is a plain-text Markdown file at a website's root that tells large language models what the site is and which key pages are worth reading first.
Long-tail question
A long-tail question is a specific, low-volume query, such as GEO compliance limits for pharma, that adds up to more total demand than broad category terms.
Mention rate
AI mention rate is the share of valid AI answers to a fixed question set that correctly mention your brand: correct mentions divided by valid answers.
Name confusion rate
Name confusion rate is the share of AI answers using your brand name that actually describe a different company with the same or a similar name.
One-source, multi-platform distribution
One-source, multi-platform distribution rewrites a single fact base to fit each platform's format, so facts stay identical everywhere while the form changes.
Question set
A question set is the fixed list of prompts used to measure AI answers. It is the denominator of every GEO metric; change the questions and you change the ruler.
robots.txt
robots.txt is a plain-text file at a site's root that uses User-agent and Disallow rules to tell crawlers which paths not to fetch. Compliance is voluntary.
sameAs
sameAs is a Schema.org property listing other authoritative URLs for the same entity, binding a brand's profiles across platforms into one recognizable entity.
Search engine optimization (SEO)
Search engine optimization (SEO) improves organic search performance through technical fixes, useful content and external links, measured by clicks and leads.
Server-side vs. client-side rendering (SSR vs. CSR)
Server-side rendering (SSR) sends fully built HTML; client-side rendering (CSR) sends an empty shell filled in by JavaScript, which many AI crawlers never run.
Structured data (Schema markup)
Structured data labels page facts in a machine-readable format, usually JSON-LD using the Schema.org vocabulary, so software does not have to guess entities.
Tencent Yuanbao
Tencent Yuanbao is Tencent's AI assistant, launched in 2024. It can search WeChat Official Account content as well as the web, making it key to China GEO.
Third-party evidence
Third-party evidence is any verifiable, clickable record about a brand published by someone else, such as regulatory filings, press, directories and encyclopedias.
XML sitemap
An XML sitemap (sitemap.xml) lists a site's indexable URLs and when each last changed, telling crawlers which pages exist and which were recently updated.
Looking for an agency rather than a definition? See Moments, the AI SEO agency.