AI search visibility glossary: GEO, AEO, LLMO and the rest
Plain definitions of the terms used in AI search visibility: GEO, AEO, LLMO, prompt, run, answer, mention, citation, Visibility, sample size and verdicts.
- The industry uses several names for the same discipline: GEO, AEO and LLMO all point at showing up in AI-generated answers.
- Three words carry the daily reading: a prompt is the question, a run is one scheduled collection, an answer is what one engine replied.
- Mentions and citations are counted separately, and no metric means anything without the number of answers behind it.
- Brand perception is what answers say about you, brand facts are what you registered to check it against, and each check ends in one of four verdicts.
Names for the discipline
The industry uses several names for substantially the same work. GEO, AEO and LLMO all point at showing up correctly in AI-generated answers, and vendors add their own variations on top.
Generative engine optimization (GEO)
The work of being mentioned, described correctly and cited in answers written by generative engines. The term comes from a 2023 Princeton paper.
Answer engine optimization (AEO)
The same discipline named for answer engines, the tools that write one answer instead of ranking links. Vendors sometimes split AEO for citation work and GEO for mention work; most teams do not keep that line.
Large language model optimization (LLMO)
An engineering-flavoured name for the same work, used for anything touching model behaviour rather than search interfaces. It overlaps GEO almost completely.
Answer engine
An engine that answers a question directly instead of ranking links: ChatGPT, Perplexity, Gemini, Grok and Claude are the ones most buyers use today. The terms answer engine and generative engine describe the same products.
The daily objects: prompts, runs and answers
A prompt is the question you ask, a run is one scheduled collection of answers, an answer is what one engine replied. Get these three straight and every metric sentence parses.
Prompt
A question you send to AI engines, written the way a buyer would type it. Prompts replace keywords in this discipline.
Run
One scheduled collection: a single prompt, on a single engine, in a single market, on a single day. Runs are what turn a prompt into a growing sample.
Answer
What one engine replied to one prompt on one day. Answers stay unedited so better analysis can be applied to history later.
Model
The specific language model behind an engine, such as a GPT, Claude or Gemini version. Vendors update models on their own schedule, which moves answers without warning.
Retrieval
The step where an engine gathers candidate material before writing: searches, index lookups, passages from the web. What retrieval shortlists constrains what the answer can say.
Query fan-out
The behaviour of rewriting one prompt into several shorter searches and merging the results. It is why an answer can depend on pages that never mention the words you typed.
Mentions, citations and sources
A mention is your name inside the answer text. A citation is a link to a page the answer drew from. They are counted separately because an answer can do one without the other.
Mention
Your brand named in the answer text, in a list or a sentence. The basic unit behind Visibility.
Citation
A link from the answer to a source page it drew from. Citations are most visible in Perplexity and less visible in a chat window with search switched off.
Citation share
How often your domains appear among the sources cited, as a share of all citations in the same answers. It measures being trusted as a source, which is a separate prize from being named.
The four headline metrics
Visibility, Average position, Sentiment and Share of voice are the four names used across this field, each is one sentence long, and none of them means anything without its sample size.
AI search visibility
The umbrella measurement: how often your brand shows up when buyers ask AI engines about your category. Visibility is the headline number inside it.
Visibility
The share of answers that mention your brand, where prompts you mark as more important count for more. Always read it as “4 in 10 answers, based on 90 answers”.
Average position
How high your brand appears in the lists AI gives, on average. 1 is the top, so lower is better.
Sentiment
How positively answers talk about your brand, from 0 to 100, where 50 is neutral.
Share of voice
Your brand's mentions as a share of all tracked brands and competitors in the same answers. Useful when category attention shifts rather than your own mention count.
Reading the numbers
Sample size says how many answers a figure rests on, a confidence interval says how wobbly it is, and est. marks a figure that was modelled rather than counted.
Sample size
The count of answers behind a figure, written as “based on N answers”. A percentage without it is not a measurement, it is a mood.
95% confidence interval
The range the true value most likely falls in, given the answers you collected. When two windows overlap heavily, the move you think you see is not there yet.
est.
The label on an approximate figure: a relative estimate computed from the data rather than a counted total. Estimates let a relative claim stand on a small sample.
Non-determinism
The property that the same prompt can produce different answers on different runs, because models sample text instead of reading it back. It is the reason daily runs exist.
Brand perception, brand facts and verdicts
Perception is what the answers say about you, brand facts are the statements you registered so those claims can be checked, and every check ends in one of four verdicts.
Brand perception
How answers describe your brand when they mention you: tone, framing, the objections they raise and the claims they repeat. Visibility says whether you appear, perception says what is said.
Brand facts
Statements you register about your brand, such as pricing model or who you serve, so answers can be checked against them.
Fact-checking verdicts
The four outcomes of checking an answer against your brand facts:
- Contradicted: the answer says something your fact rules out.
- Supported: the answer matches the fact.
- Inconclusive: the answer says something related that neither confirms nor contradicts the fact.
- Not covered: the answer never touches the subject of that fact.
How the tools work
Bring your own key means you connect your own provider key, so answers come straight from each engine's API and you pay the provider directly for what you use.
Bring your own key
You supply a provider API key and the tool spends it without marking up tokens. In advisible there is no token markup.
Sample data
The badge on demo content generated by advisible instead of a real engine reply. It appears only in the demo, never in a live workspace.
For the metric definitions behind this page in longer form, see Visibility, position, Share of voice and Sentiment.
Common questions
Is GEO the same as AEO?
In practice, yes. AEO (answer engine optimization) and GEO (generative engine optimization) describe the same prompts, pages and measurements, and vendors pick whichever name matches their angle. LLMO and AI SEO are two more aliases.
What is the difference between a mention and a citation?
A mention is your brand named in the answer text. A citation is a link to a page the answer drew from. An answer can cite you without naming you, or name you without citing you.
Which metric should I report first?
Visibility, because nothing else matters when you are absent from the answer. Once mentions are steady, add Average position to see where you appear in the list, Share of voice to see your slice against competitors, and Sentiment to see how you are described.
What counts as a good Visibility score?
There is no universal good score. It is a share of answers, so the useful readings are your gap to the competitors in the same answers and the direction of the trend over weeks, always with the sample size attached.
Why are there so many names for this field?
The discipline is new and each vendor names its own angle. Underneath, everyone is counting the same objects: prompts, runs, answers, mentions and citations. Translate rather than choose.
What is AI search visibility?
AI search visibility is how often your brand shows up when buyers ask ChatGPT, Perplexity or Gemini about your category. What it means and how to measure it.
What is generative engine optimization?
GEO is getting mentioned, described correctly and cited in AI answers. How engines build an answer, what GEO includes, and what you can measure.
GEO vs SEO: what changes and what carries over
GEO optimises how AI engines describe and cite you. SEO optimises ranking. The buyer is the same and much of the work carries over. A side-by-side comparison.
Visibility, average position, share of voice and sentiment explained
What each AI visibility metric counts, worked examples with invented numbers, what each one cannot tell you, and which to put in front of your boss.
Track this for your own brand
advisible runs your prompts on ChatGPT, Perplexity, Gemini, Grok and Claude every day, and every number shows how many answers it is based on.