Short definitions of the terms you will meet when tracking how AI assistants talk about your brand. Where a term is also an AEOmeter metric, the definition says exactly how AEOmeter counts it. The methodology page has the full detail.
The practice
- AEO (answer engine optimization)
- The work of making your brand appear, and be described accurately, in answers from AI assistants such as ChatGPT, Claude, Perplexity and Google AI Overviews. The term predates chatbots and was first used for featured snippets and voice assistants. See the AEO guide.
- GEO (generative engine optimization)
- Another name for the same work as AEO, popularized by a 2023 research paper of that name. In practice GEO and AEO are used interchangeably.
- LLMO (large language model optimization)
- A third name for optimizing how large language models mention and describe a brand. Same idea as AEO and GEO.
- AI visibility
- How often, and how, AI assistants mention or cite your brand in answers to the questions your buyers ask. Measured by sampling answers to a fixed set of prompts. See what is AI visibility?.
- AI Overview
- The AI-generated summary Google shows above regular search results for some queries, with links to the pages it drew on. Many queries show no AI Overview at all; AEOmeter records those as "no AI Overview" and does not count them as misses. See Google AI Overviews tracking.
Metrics
- Mention
- An answer that names a brand. AEOmeter detects mentions with deterministic, case-insensitive matching of the brand’s aliases on word boundaries, and counts a brand at most once per answer.
- Mention rate
- Answers that mention any alias of your brand ÷ answers successfully captured. Shown as "Visibility" in AEOmeter. When there are no captured answers, it shows a dash instead of 0%.
- Citation
- A source the AI engine itself attributes to an answer, such as the source links ChatGPT, Perplexity or an AI Overview show. A citation tells you which page was used. It does not prove that page caused a mention. See citation tracking.
- Owned citation rate
- Answers that cite at least one URL on your own domain or its subdomains ÷ answers successfully captured.
- Inline link
- A URL that appears inside the answer text rather than in the engine’s list of sources. AEOmeter keeps inline links separate from citations, because the engine has not attributed the answer to them.
- Gap
- An evidence-backed finding such as "a competitor is mentioned in this answer and you are not". Each gap links to the answer and its sources, so you can see what to work on.
How tracking works
- Prompt
- A question sent to an AI assistant with the same wording every time, such as "best product analytics tool for a seed-stage startup". See choosing prompts.
- Baseline
- The first set of results for a prompt, which later results are compared against. In AEOmeter, editing a prompt’s wording starts a new baseline, so trends only compare answers to the same text.
- Check
- One prompt sent to one engine, producing one answer, or a recorded failure such as a login wall, verification page or timeout. Failed checks are shown and never counted as misses.
- Run
- A batch of checks: the enabled prompts, each sent to its selected engines. A run can be started by hand or queued by a daily schedule.
- Engine
- An AI product that answers prompts, also called an answer engine. AEOmeter supports four: ChatGPT, Claude, Perplexity and Google AI Overviews. See engines.
- Grounding
- When an assistant runs a web search and bases its answer on the pages it finds, often listing them as sources. Grounded answers can change as the web changes, without any change to the model.
- Temporary chat
- A ChatGPT mode that does not use or create memories and keeps the conversation out of your chat history. Useful when checking visibility, so your own history doesn’t shape the answer. AEOmeter turns on each provider’s private mode where it can.
Technical
- llms.txt
- A proposed convention from 2024: a Markdown file at a site’s root that lists its important pages for language models. Some coding agents and tools read it. No major AI assistant has documented using it to choose which brands to mention or cite. AEOmeter publishes one at /llms.txt.
- MCP (Model Context Protocol)
- An open protocol that lets AI assistants connect to outside tools and data. AEOmeter runs a remote MCP server, so you can ask your AI assistant about your prompts, runs and AI visibility results. See MCP and API.
