This AI search glossary defines 40 terms used in AEO, GEO and AI visibility work, from share of voice to GPTBot. Each term gets one plain definition, a link to the full guide where one exists, and a note where Depra does not measure it.
AEO and GEO are two names for the same job: getting your brand named and cited when a buyer asks ChatGPT, Gemini, Perplexity or Google AI Overviews what to buy. The terms below are grouped by strategy, engines, how answers are built, prompts, metrics, sources and crawlers.
Strategies: AEO, GEO and AI SEO
- AEO (answer engine optimization): the work of getting your brand named and cited in the answers that ChatGPT, Gemini, Perplexity and Google AI Overviews write for a question. Full guide: what is AEO.
- GEO (generative engine optimization): the same goal under a name taken from a 2023 research paper, focused on making your pages the sources AI answers draw on. Full guide: what is GEO, and how the terms relate in AEO vs GEO vs SEO.
- LLM SEO: optimising your site and reputation so large language models, the AI models behind tools like ChatGPT and Gemini, name your brand when they answer. Full guide: LLM SEO explained.
- AI SEO: a broad label for search work aimed at AI answers, used for both AEO and GEO. How it differs from classic search work: GEO vs SEO and AEO vs SEO.
AI engines and surfaces
- AI Overview: the AI-written summary Google shows above some search results, with links to pages it drew on. Google does not show one for every search, so Depra records whether an Overview appeared as well as whether it named you (Google AI Overview tracker).
- Google AI Mode: a separate Google search mode where you ask follow-up questions in a chat and get a longer AI answer with links. Depra does not track AI Mode today.
- ChatGPT search: ChatGPT's web search feature, which looks up current pages before it answers and shows the sources it used. Depra captures ChatGPT answers with search switched on (ChatGPT visibility tracker).
- Perplexity Sonar: the answer model Perplexity sells through its API, which searches the live web and cites sources. Depra uses Sonar for Perplexity answers and checks them weekly (Perplexity visibility tracker).
- Gemini: Google's AI assistant at gemini.google.com. Depra checks it daily in the same web app people use (Gemini visibility tracker).
How AI engines build an answer
- Retrieval: the step where an AI engine searches the web or an index and pulls in pages before it writes. Depra stores the pages an answer cites. It cannot see pages an engine read and chose not to cite.
- Grounding: tying an AI answer to real sources, such as search results, so the model writes from current pages in place of its memory alone.
- RAG (retrieval-augmented generation): a design where a system first fetches relevant documents and then has the language model write its answer from them. AI search features that look up web pages before answering work this way.
- Knowledge cutoff: the date after which a model has no training data, so without web search it cannot know about newer products, prices or brands.
- Query fan-out: when an AI engine splits one question into several related searches and combines the results. Google uses the term for AI Overviews and AI Mode (Google Search Central). Depra does not capture fan-out searches.
Prompts and languages
- Prompt: a question your buyers type into an AI tool, such as "best sunscreen for oily skin under ₹500", which a tracker asks on a schedule to see which brands the answer names. Depra suggests 5 topics with 8 questions each.
- Branded vs non-branded prompt: a branded prompt contains your brand name and almost always returns it, and a non-branded prompt does not. Depra leaves branded prompts out of your scores by default.
- Hinglish: Hindi and English typed in English letters, such as "sabse accha protein powder kaunsa hai". Depra tracks Hinglish versions of your questions next to English (Hinglish AI visibility), and our English vs Hinglish study shows how the answers differ.
- Answer language: the language an AI engine replies in, which can differ from the language of the question. In our study, Gemini answered 8 of 80 Hinglish questions in Hindi script. Depra detects the reply language on every answer.
AI visibility metrics
- AI visibility: how often AI names your brand: the share of AI answers to your buyers' questions that mention you (AI visibility tracker).
- AI share of voice: your share of all brand mentions in the AI answers to your category's questions: your mentions divided by the mentions of all tracked brands, times 100 (AI share of voice formula).
- Visibility score: one percentage for AI visibility over a date range. In Depra it is the answers that name you divided by all answers in the period (how Depra measures AI visibility).
- Average position: where your brand comes when an answer lists several options, averaged across answers, where 1 means named first.
- Sentiment: whether an answer describes your brand positively, neutrally or negatively. Depra keeps the quoted sentence so you can check each one.
- Confidence range: the band a percentage most likely sits in, given the number of answers behind it. Your Depra report shows a 95% range for visibility.
- Rerun variation: the change in the brands named when the same question is asked again. In our test, two ChatGPT answers to the same question shared 54.7% of their brands on average, which is why one manual check is not enough (does ChatGPT give the same answer to everyone).
Content and sources
- Citation: a link to a web page that an AI answer shows as a source.
- Cited source: a website AI answers link to. Depra labels each one by type, such as editorial, forum, marketplace, directory, company site or your own site.
- Citation gap: a website AI cites when it recommends your competitors but never when it names you, which makes it a place to get mentioned (citation gap analysis).
- Entity: a distinct thing, such as a brand, person or product, that search engines and AI models recognise and link to facts about it. Depra does not audit entity data.
- Listicle: a "best X for Y" list article. Depra labels cited pages as listicles, comparisons, how-tos, product pages and more, so you can see which formats AI cites in your category.
- Zero-click search: a search where the person gets what they need on the results page, from a snippet or an AI Overview, and clicks no link.
Crawlers and files
- robots.txt: a text file at the root of your site that tells crawlers which pages they may fetch. OpenAI and Perplexity say the fetchers they send when a user asks a question may not follow it (AI crawlers guide).
- llms.txt: a proposed markdown file at yoursite.com/llms.txt that points AI tools to your most useful pages. Google says Google Search ignores it (does llms.txt work).
- GPTBot: OpenAI's crawler for content that may be used to train its AI models (OpenAI crawler docs).
- OAI-SearchBot: OpenAI's crawler for ChatGPT search, used to surface websites in ChatGPT's search results. OpenAI says sites that opt out of it are not shown in ChatGPT search answers, though they can still appear as navigational links (OpenAI crawler docs).
- ChatGPT-User: the agent ChatGPT uses when it visits a page because a user asked something. OpenAI says robots.txt rules may not apply to it (OpenAI crawler docs).
- Google-Extended: a robots.txt token with no crawler of its own that controls whether Google may use your content to train Gemini models and for grounding. Google says it does not affect inclusion in Google Search (Google crawler docs).
- PerplexityBot: Perplexity's crawler that surfaces and links websites in Perplexity search results. Perplexity says it is not used to train AI foundation models (Perplexity crawler docs).
- Training crawler: a bot that collects pages to train AI models, such as GPTBot. Blocking a training crawler is a separate choice from blocking the same vendor's search crawler.
- Search crawler: a bot that fetches pages so an AI engine can cite them in answers now, such as OAI-SearchBot or PerplexityBot. Depra's AI crawler checker shows which of 41 AI crawlers your robots.txt allows.