Glossary

The AI-search glossary.

The vocabulary of getting named by AI is new, and a lot of it is sold with more confidence than it deserves. Here are the terms that actually decide whether an answer engine says your name, defined plainly, with no jargon for its own sake.

The basics

Answer Engine Optimization (AEO)
The practice of getting a business named, and ideally cited, inside the direct answers AI assistants give. Classic SEO competes for a spot in a list of links. AEO competes to be the answer itself, which rewards different signals: answer-first content a model can quote, a complete business profile, and citations on the sources an engine already trusts.
Generative Engine Optimization (GEO)
A close cousin of AEO focused on generative engines like ChatGPT, Perplexity, and Google AI Overviews. The two terms are often used interchangeably. Where they differ, GEO leans toward the model-generated answer, and AEO toward any answer surface, including voice and featured snippets.
Answer engine
Any system that returns a synthesized answer instead of a list of links. ChatGPT, Perplexity, Google AI Overview, Google AI Mode, Gemini, Grok, and Claude all qualify. For a local business the question is simple: when a buyer asks one of these for a recommendation, does it say your name.
Brand mention vs. ranking
Ranking is your position in a list of links. A brand mention is whether an assistant says your name when someone asks a question. They are related but not the same. A business can rank well on Google and still never be mentioned by AI, and closing that gap is the entire job of AEO.

Where answers appear

Google AI Overview
The AI-generated summary at the top of many Google searches, above the traditional results. It pulls from the live web and from Google's own data, and for local queries it leans heavily on Maps and Business Profile signals. Being named here is often worth more than ranking first in the blue links below it.
Google AI Mode
A conversational search experience where a user asks follow-up questions and receives a running answer. It draws on the same systems as AI Overview but in a back-and-forth format. It is a separate surface to track, because a business can appear in one and not the other.
Local pack
The block of three map results Google shows for local-intent searches. It draws on Google Business Profile and proximity, and it heavily informs what AI surfaces say about local options. Strength in the local pack and strength in local AI answers tend to move together.
Google Business Profile (GBP)
The free listing that powers a business on Google Maps and the local pack. For local AI answers it is foundational, because engines lean on its categories, reviews, and attributes when they decide who to name. A thin or miscategorized profile is one of the most common reasons a business that ranks on Google still goes unnamed by AI.
Zero-click
When a search or AI answer satisfies the user without a click through to any website. Most AI answers are zero-click by design, which is exactly why being named inside the answer matters more than ever. The impression lands whether or not the user ever visits your page.

How we measure

AI presence rate
How often a business is named when an answer engine responds to the prompts a buyer would actually ask, expressed as a percentage with a confidence interval. A presence rate without an interval is a guess dressed as a fact, because AI answers change every time they run.
Share of voice
The portion of relevant AI mentions that go to a given business, measured against the competitors named in the same answers. If your name appears in one of every six mentions across a set of local prompts, your share of voice is about seventeen percent. It reads competitive position more cleanly than a raw presence number on its own.
Confidence interval (Wilson)
A range that expresses how sure a measurement is. An engine might name a business in four of ten runs today and six of ten tomorrow, so a single number misleads. Running each prompt repeatedly and reporting a 95% Wilson interval shows the real spread, and it is what lets you call a change confirmed only when two intervals stop overlapping.
Geo-grid
A lattice of coordinate points across a metro, each used to ask the same question from a different location. It exists because the best dentist near me resolves differently on the north side of a city than the south. A geo-grid measures local visibility the way a real buyer experiences it, block by block, instead of as one national average.
Multi-run sampling
Asking the same prompt several times and aggregating the results, rather than trusting one response. It is the only honest way to measure a system that answers differently each time, and the sample size is what feeds the confidence interval.
Tracked prompt
A single buyer question watched over time, such as who is the best orthodontist in Austin. You track presence per prompt, per engine, so you can see which questions you win, which you lose, and which moved after a fix.
Supported presence
A mention where the engine also cites the business's own website, not just a bare name-drop. Supported presence is sturdier, because it is anchored to a source the model can point to. Unsupported presence can vanish the next time the model runs.

How answers get built

Citation
The source an answer engine credits when it makes a claim, usually a linked domain shown with or beneath the text. Citations are the cause behind presence. A business tends to get named when the sites an engine already cites for that topic mention it too.
Citation graph
The map of third-party sites an engine cites across a category, such as review sites, directories, and local guides. The graph shows where presence comes from. The gaps in it, the cited sources that name your competitors but not you, become a concrete and checkable fix list.
Grounding
The step where an engine pulls real sources from the web before composing an answer, instead of answering from memory alone. Grounded answers cite sources and reflect current reality, which is why the citations behind them are the lever for visibility. The more an engine grounds, the more your off-site footprint decides whether you appear.

What actually moves visibility

Answer-first content
A page that opens with the single clearest answer to a specific question, stated so a model can lift it verbatim, before any marketing prose. Answer engines reward content that is easy to quote. Burying the answer under a story is a reliable way to be skipped.
llms.txt
A plain-text file at the root of a site that hands AI models a clean summary of what the site is and where its important pages live. It is a courtesy to crawlers, not a ranking lever. It can help a model describe you accurately, but on its own it will not make you the answer.

Now see where you actually stand.

Definitions are the easy part. Run a free check to see how often the AI engines name your business, and which competitors they name instead.

Run a free AI-visibility checkHow we measure