GEO & LLMO glossary
- AEO (Answer Engine Optimization) Optimizing so that 'answer engines' — AI search, voice assistants, featured snippets — adopt your information as the answer. Overlaps heavily with GEO and LLMO.
- AI crawler Bots that AI companies use to fetch web content: GPTBot and OAI-SearchBot (OpenAI), ClaudeBot (Anthropic), PerplexityBot and others. Major AI crawlers do not execute JavaScript.
- AI Overviews Google Search's AI-generated summary shown at the top of results, answering the query directly while citing multiple sources. Being cited in it is a new visibility surface.
- AI share of voice The share of AI answers on a topic that mention your brand, compared with competitors. The central KPI of "how loud you are" in the AI-search era.
- Answer capsule A self-contained, direct answer of roughly 40–60 words placed immediately under a heading. Writing conclusion-first makes the passage easy for AI to extract and cite.
- Citation When AI references a specific page or domain as the basis of its answer and presents it as a source. The AI-search equivalent of ranking — the central KPI of GEO.
- E-E-A-T Experience, Expertise, Authoritativeness, Trust — Google's framework for judging content quality, with Trust at the core. Not a direct ranking factor but the quality lens.
- GEO (Generative Engine Optimization) The practice of optimizing how AI assistants like ChatGPT and Gemini mention your brand in generated answers — accurately, favorably and prominently. Coined by a Princeton-led study (KDD 2024).
- Hyokiyure (表記ゆれ, spelling variants) hyōki-yure Multiple ways of writing the same name — e.g. キヤノン, キャノン, Canon, CANON. Mention tracking in AI answers misses real mentions unless variants are matched together.
- LLMO (Large Language Model Optimization) Optimizing for brand presence in the answers and recommendations of large language models. The term most widely used in Japan's marketing industry; effectively synonymous with GEO.
- RAG (Retrieval-Augmented Generation) The mechanism by which AI retrieves relevant information from web search or databases before generating an answer. AI-search citations happen on top of this mechanism.
- Sentiment analysis Judging the tone — positive, neutral, critical — in which AI talks about a brand inside its answers. For Japanese, accuracy depends on handling keigo and indirect phrasing.
- Structured data (JSON-LD) Machine-readable markup (Schema.org vocabulary as JSON-LD) describing page content. Helps AI and search engines understand and extract — but by itself does not guarantee citations.
- Visibility score A 0–100 composite of brand mentions across prompts, AI engines and samples, weighted by position. The central dashboard metric in Suparanku.
- Zero-click search When the search result or AI answer satisfies the need and no site gets clicked. Growing with AI search; the force shifting KPIs from traffic to representation inside answers.