A patient in Columbus opens ChatGPT and types "best periodontist near me for gum recession." The assistant names three practices, confident and specific, no hedging. One of them opened eighteen months ago and gets named first. A fourth practice down the street — one that spent years building its reputation the traditional way — never appears in the answer at all. Nobody told that practice this was happening. There was no notification, no dropped ranking to investigate, no red line on a dashboard. Just a quiet reassignment of who gets recommended, playing out thousands of times a day across seven different AI assistants.
That reassignment runs on a vocabulary most local professionals have never been taught — and the ones who've learned it are the ones deciding how their category gets described to the machines, while everyone else just watches the names change. Below is a working reference for that vocabulary: defined plainly, cross-linked, and written so you can actually use it in a conversation with a client or a competitor.
Why a glossary matters right now
You can't manage what you can't name. Most local professionals are still describing their visibility problem in SEO terms — rankings, keywords, backlinks — while the thing quietly happening to their business runs on an entirely different set of concepts. When a dentist says "I'm on page one," they're answering a question ChatGPT never asked. The assistant doesn't page through results. It composes an answer from what it already believes to be true about the practices in that zip code, and belief is built differently than ranking.
So the terms below aren't academic. Each one names a lever, a signal, or a measurement that decides whether an AI assistant says your name or someone else's. Read them as a map of where the ground moved.
The core vocabulary of AI visibility, A to Z
AI visibility score. A composite measure of how likely AI assistants are to surface your business across relevant queries and across platforms. Where a rank-tracking tool gives you a position for a keyword, an AI visibility score reflects presence inside generated answers — a fundamentally different quantity. It's the number that tells you whether you exist in the world the machines describe. See our full breakdown at /blog/what-is-ai-visibility-score.
Citation. A specific instance of an AI assistant naming or referencing your business inside a response. One citation is one moment where the model chose you as part of its answer. Citations are the raw events that everything else measures.
Citation rate. The percentage of relevant queries where you appear across assistants. If a plausible set of client questions returns your name in a fraction of cases, that fraction is your citation rate. It's the headline KPI of AI visibility because it measures the outcome — being recommended — rather than a proxy for it.
Co-citation. When your business appears alongside other established names in the same context, either within an AI answer or across the sources a model draws from. Being named in the same breath as recognized practices in your field signals to a model that you belong in that set. Related: entity authority, topical authority.
E-E-A-T. Experience, Expertise, Authoritativeness, Trustworthiness — the qualities that make a source credible. Originally a search-quality concept, it maps cleanly onto how models decide whom to trust. A profile that demonstrates real, corroborated expertise reads differently to an AI than one that merely asserts it.
Entity. The thing an AI understands your business to be — a distinct real-world actor with a name, a location, a specialty, and a reputation, as opposed to a string of keywords on a page. GEO is largely the work of making that entity coherent. Related: entity graph, knowledge graph.
Entity authority. The accumulated weight of evidence that establishes your business as a credible, real, notable entity in its field. It grows when independent sources describe you consistently and describe you as significant. Low entity authority is why a technically excellent website can still go unnamed.
Entity disambiguation. How AI systems tell you apart from everyone with a similar name, location, or specialty. If there are three Dr. Sarah Chens practicing dermatology in the greater Phoenix area, a model needs enough consistent, distinguishing information to know which one to recommend — and confusion here often reads as absence.
Entity graph. The web of relationships connecting your entity to others — the directories that list you, the practices you're mentioned alongside, the professional bodies you belong to. Models reason over these relationships to place you in a category and gauge your standing within it. Related: knowledge graph, co-citation.
FAQPage schema. A structured-data format that labels question-and-answer content on your site so machines can read it as such. Because assistants often answer in the shape of a direct reply, clearly marked Q&A content is unusually legible to them.
GEO (Generative Engine Optimization). The practice of optimizing your presence so generative AI assistants recommend you. The AI-era sibling of SEO, aimed at the trust and entity signals models rely on rather than the link and keyword signals a search engine ranks. Our complete guide lives at /blog/generative-engine-optimization-guide.
JSON-LD. The preferred syntax for schema markup — a block of structured data that states plainly who you are, where, and what you do, in a format machines parse without ambiguity. It's how you hand a model facts instead of hoping it infers them.
Knowledge graph. The large-scale map of entities and their relationships that systems consult to understand the world. Your presence and accuracy within these graphs shapes whether a model treats you as a known quantity or a stranger. Related: entity, entity graph.
LLM (Large Language Model). The class of AI that powers ChatGPT, Claude, Gemini, Perplexity, Google AI Overviews, Grok, and Copilot — trained on vast text, generating answers by predicting language rather than retrieving ranked links. Understanding that it composes rather than lists is the mental shift the whole glossary supports.
NAP consistency. Whether your Name, Address, and Phone number match everywhere they appear online. Small mismatches — a suite number here, an old number there — quietly erode a model's confidence that it knows who you are. This one earns a closer look below.
RAG (Retrieval-Augmented Generation). A method where an AI fetches live information from external sources before answering, rather than relying only on training data. Perplexity and Google AI Overviews lean heavily on it, which is why fresh, retrievable, consistent information about your business can matter as much as anything baked into a model months ago.
Schema markup. Structured tags that describe your content to machines in a standardized vocabulary. It doesn't change what a human sees; it changes what a machine understands. Related: JSON-LD, FAQPage schema.
Topical authority. The depth and consistency with which you're associated with a particular subject or specialty. A practice known specifically for pediatric orthodontics, described that way across many sources, reads as an authority on it — and gets recommended for it. Related: entity authority, co-citation.
Why consistency does so much work: a closer look at NAP
Here's the concept most people underestimate, so it's worth slowing down on one term.
AI systems don't take your word for who you are. They can't. A model has no way to verify a claim a business makes about itself except by checking whether the same claim shows up, unchanged, somewhere it didn't control. Agreement across independent sources is what turns an assertion into a fact a model will repeat. Contradiction does the opposite.
That's why NAP consistency punches so far above its weight. Picture an oral surgeon in Sacramento whose address reads one way on Yelp, another on an old health directory, and a third on the practice website after a move nobody fully updated. To a person, that's a minor mess. To a model trying to decide whether this is one trustworthy business or possibly two half-real ones, it's a reason to hesitate — and a hesitating model reaches for a name it feels surer about. The surgeon down the road, whose details line up cleanly everywhere they appear, becomes the safer recommendation. Not because they're better. Because they're legible.
The unsettling part is how invisible this is from the inside. Nothing on your own website looks broken. The inconsistency lives out in the places you stopped checking years ago, and it costs you the one thing you'd most want back: the patient who asked and got sent elsewhere.
The scale of the shift
This isn't theoretical. RankCommander's AI Visibility Index has evaluated more than 66,000 real AI answers across local professional verticals — and still counting — and the pattern holds across dentistry, medicine, law, and real estate: within the same city and specialty, some established businesses get named repeatedly while comparable ones never surface at all. The full benchmarks by vertical are at /ai-visibility-index.
If you want to go deeper on why this diverges so sharply from Google, we cover the mechanics in /blog/ai-search-vs-google-search.
Don't let a competitor become your category's answer
Every term in this glossary points at the same reality: the machines are already deciding whom to recommend in your city, and they're deciding without you in the room. Somewhere in your zip code, a competitor is being named first while your years of work sit outside the conversation — and each answer that skips you is a real person who wanted exactly what you do and was handed to someone else. What separates the practices getting recommended from the ones going unnamed usually isn't quality; it's whether anyone is watching what the assistants actually say.
That's the gap RankCommander closes. Run a free AI visibility scan and see exactly how ChatGPT, Claude, Gemini, Perplexity, Google AI Overviews, Grok, and Copilot describe your business right now — before the answer hardens around someone else's name.