The phone log noticed before she did.
Consider a practice like this one. Dr. Sarah Okafor had run her Columbus practice for twelve years. 512 Google reviews. A website that ranked on page one for the searches that mattered. A marketing budget she reviewed every quarter and a schedule that still looked healthy from the operatory.
Then the front desk mentioned something in a Monday huddle: eleven fewer new patient calls than the same month last year. The month after that, fourteen fewer. Nothing on her end had changed. The reviews kept coming. The rankings held. The patients just arrived less often, and nobody could say why.
The answer showed up on an intake form. One of the few new patients that month, an implant consult, had written a single word on the how-did-you-hear-about-us line: "ChatGPT."
Which raised the question that would not leave her alone. If one patient found her that way, how many others had asked the same question, and been sent somewhere else?
If your schedule still looks fine but your new patient flow has gone quiet in a way you cannot explain, you may be watching the same shift from the inside. This guide is the complete picture of dental AI search: what it is, how patients use it, why most practices are invisible in it, and what actually earns visibility. It is the cornerstone of the AI-Enabled Patient Growth topic, and everything else we have published on the subject branches from here.
Start with the scale of what changed. The Dental Index national practice audit found that 70% of US dental practices are invisible to AI-referred patients, against a backdrop of 432,000 AI dental searches every month. Your practice is almost certainly in that 70% unless you have deliberately built the signals this guide describes. Most owners have not, because until recently nobody told them these signals existed. The full breakdown of that number lives in why 70% of dental practices are invisible to AI search; this guide is concerned with what you do about it.
What Is Dental AI Search, and Why Does It Work Differently?
Dental AI search is the layer of patient discovery that happens inside AI assistants and AI-generated answers: ChatGPT, Perplexity, Google AI Overviews, Gemini, Copilot. A patient describes what they need in a full sentence, and the system answers with a small number of named practices. Not a list of ten blue links. Two or three names, with reasons.
Most dentists assume this is just Google with a new interface. That assumption is the expensive one.
Google ranks pages. AI systems evaluate entities. A traditional search engine asks, "Which web pages best match this query?" An AI assistant asks a harder question: "Which provider am I confident enough in to recommend, by name, to a person asking for help with their health?" It answers by reading everything it can verify about you: your business profile, your directory presence, your reviews, the specificity of your service descriptions, and whether all of it tells one consistent story.
That is why the results are binary. In classic search, position seven still exists. In AI search, you are either named in the answer or you are not, and the patient never knows you were close. There is no page two.
Here is the frame this entire guide rests on: AI search is where your positioning meets a patient before the first call. Every practice has a positioning, whether it was chosen deliberately or accumulated by accident. AI systems read that positioning and repeat it back to patients. If the signal is clear, you get named. If the signal is generic, fragmented, or thin, the system moves on to a practice it can describe with confidence. Visibility is earned by signal clarity, not spend. You cannot buy your way into an AI answer. You can only become legible to it.
That is also why this is a positioning story and not a technology story. The audit tracks 2.4M+ monthly dental searches across 36 keyword clusters, and the AI-driven slice of that demand is the fastest-shifting part of it. The practices capturing it are not the biggest spenders. They are the clearest signals.
How Are Patients Actually Using AI to Choose a Dentist?
The queries themselves tell you what changed. A patient no longer types "dentist near me" and does their own filtering. They ask one question that contains all of their criteria at once: "Who is the best implant dentist near Dublin, Ohio who is good with anxious patients and can see me this month?"
Notice what is inside that sentence. Service. Location. A fear. A timeline. The AI resolves all four against what it knows about the practices in that market, and the practice it names has effectively won the patient before any website is visited. The full behavioural picture, including the exact question patterns, is documented in how patients use ChatGPT to find a dentist. For this guide, three findings matter most.
First, the patients arriving through AI are pre-sold. AI-referred patients book high-value procedures at 2-3x the rate of other referral sources. That makes sense once you see the mechanism: they arrive having already been matched to your practice by name, for the specific thing they need. The convincing happened before the phone rang. If your case acceptance conversations feel like uphill work, part of the answer is that your patients are not arriving pre-sold, and AI referral is the channel that delivers patients who are.
Second, AI search and Google Maps are one system, not two. 82% of dental searches end in a Google Maps interaction, and AI assistants lean on the same underlying business data that Maps does. The practice that wins the AI answer is usually the practice whose Maps presence is complete, active, and specific. If you want the mechanics of that side, how dental Google Maps ranking works in 2026 covers it in depth. The point here: the signal you build for one is the signal the other reads.
Third, the highest-value demand is the most AI-shaped. Discovery patterns differ by service. Emergency patients skew hard toward Maps: 52% find a practice there. Implant patients spread across referral, word-of-mouth, and Maps, and they research longest, which is exactly the behaviour AI assistants absorb first. With implant demand growing 8.5% per year at a $4,500 average case value, and cosmetic demand growing 6.8% at $3,800, the procedures that fund your practice are precisely the ones patients now research conversationally.
Land that on your own numbers. The average solo practice leaves $147,000 in annual production unrealised, not from empty chairs, but from a case mix that does not match the demand around it. AI search is where that mismatch now gets decided, silently, before you ever meet the patient.
What Is Actually Keeping Your Practice Out of AI Answers?
Dr. Okafor's turning point was not the intake form. It was the night she asked the machines directly.
She opened ChatGPT and typed the question her own patients would type: best implant dentist in her part of Columbus. Three names came back. Hers was not one of them. She tried Perplexity. Same result, different order. She tried five variations of the question. She appeared in none of them. Then she ran her practice through an audit and got a number for the feeling: AI readiness score, 33 out of 100. Twelve years of clinical work, and the systems patients now ask first could barely see her.
Her score was not unusual. It was almost exactly average. The national average AI readiness score sits below 40 out of 100, and fewer than 8% of US practices score above 65. What that score measures, and how each factor is weighted, is broken down in the dental AI readiness score explained. What it reveals, in practice after practice, is one of three signal failures.
- A fragmented identity. Your practice name, address, and phone number differ slightly across the platforms an AI checks. Each variation reads as a separate, smaller entity. None of them accumulates enough evidence to be recommended. You are not absent from the internet. You are scattered across it.
- A profile that answers nothing. Your Google Business Profile exists but is partial: missing services, thin descriptions, unanswered questions, stale photos. To a patient it looks like a listing. To an AI it looks like a provider it cannot describe, and an AI never recommends what it cannot describe.
- Language with nothing to cite. "Comprehensive dental care for the whole family" appears, in some variation, on the majority of practice websites in the audit database. An AI cannot repeat it because it distinguishes nothing. "Same-day crowns, most completed in under two hours" is a sentence a machine can carry to a patient. Generic language is invisible language.
None of these are effort failures. They are clarity failures. Which is why the performance gap between clear-signal and generic-signal practices is not a gap of degree. It is a gap of kind:
| Practice Signal State | AI Readiness Score | AI-Referred Patient Flow | Observed Pattern |
|---|---|---|---|
| Clear positioning: consistent identity, fully completed GBP, specific citable services | 65+ (fewer than 8% of US practices) | 7x more AI-referred clicks than partial profiles | Named in AI answers; AI-referred patients book high-value cases at 2-3x the rate of other sources |
| Generic positioning: partial profile, vague service language, uneven directory presence | Below 40 (national average) | Baseline clicks; rarely named in AI answers | Visible in Maps but not cited; demand routes to clearer competitors |
| Fragmented positioning: identity conflicts across platforms | Bottom tier | Effectively none; part of the invisible 70% | Even top-ranked practices in Maps capture only 2.3% of available patient demand; fragmented practices capture almost none of the AI-routed share |
Source: The Dental Index national practice audit · 2026
Read that table as a mirror, not a chart. If you have never verified your identity consistency or your profile completeness, the middle row is statistically where you sit. And the practice two miles away that keeps appearing in the answers is probably not better at dentistry. It is better at being read.
You cannot buy your way into an AI answer. You can only become clear enough to be read, and specific enough to be repeated.
What Do the Practices Getting Recommended Do Differently?
Strip away the platform details and the visible 8% share one habit: they decided what their practice is known for, and they say it identically everywhere a machine might look. The implant practice reads as an implant practice on its profile, in its reviews, on its service pages, and in its directory listings. There is nothing for an AI to reconcile, so there is nothing to doubt.
That clarity then gets expressed platform by platform, because each system reads you through a different door:
- ChatGPT reads Bing's index, not Google's. This is the single most common blind spot in the audit data. A practice can be strong on Google and simply not exist in the index ChatGPT draws from. The fix, and why it matters more than its ten-minute cost suggests, is covered in why Bing Webmaster Tools controls your ChatGPT visibility. What earns the actual recommendation, the shape of a practice ChatGPT is willing to name, is the subject of how to get your dental practice listed in ChatGPT's answers.
- Perplexity favours evidence it can cite. It builds answers from sources it can attribute: specific service pages, review patterns, structured content. Practices with citable specifics get quoted; practices with slogans get skipped. The platform-specific playbook is in how dental practices rank on Perplexity.
- Google AI Overviews and Gemini lean on your Business Profile. Practices with fully completed profiles receive 7x more AI-referred clicks than practices with partial ones. Not marginally more. Seven times. Completeness is not an administrative chore; it is the difference between a provider the system can describe and one it cannot.
- Reviews function as positioning evidence, not a score. AI systems read what your patients say you are known for, not just how many stars they left. Two hundred reviews that mention gentle implant work build a recommendable identity. Eight hundred that say "great staff, clean office" build a pleasant blur.
Notice what is absent from that list: ad budget, follower counts, website redesigns. Your marketing still matters, and it converts better once these signals are in place, because every dollar you spend now points at a practice AI systems can verify. Clear positioning is what makes the marketing you already do work harder. But the entry ticket to the AI answer itself is signal clarity, and signal clarity is a set of decisions, not a line item.
What Separates the Owners Who Fix This From Those Who Don't?
After enough audits, a pattern emerges that has nothing to do with market size or tenure. The owners who close the visibility gap are not the ones who work hardest at it. They are the ones who reframe it.
The owners who stay invisible treat AI search as one more marketing channel to fund eventually, somewhere behind the ads and the mailers. The owners who become visible see it differently: AI search is the public record of their positioning. It is what every prospective patient, every potential associate, and eventually every potential acquirer sees before speaking to them. You do not delegate your public record and check on it quarterly. You decide what it should say.
That reframe changes the question from "how do I rank?" to "what should a machine say about my practice when a frightened person asks at 11pm?" Answer that question precisely, and the tactics almost arrange themselves. The six moves below are how the reframe becomes action. Each one is a positioning decision, and each one works because of what it changes in the patient before they ever call you.
Name one specific thing your practice is known for, everywhere
A patient who arrives knowing you as "the implant practice that handles anxious patients" has already answered their own biggest question: is this the right place for someone like me? That patient books, accepts, and refers differently than one who arrived at "a good dentist nearby." AI systems amplify whichever version of you exists. If your identity is one clear sentence repeated across every patient-facing channel, that sentence is what gets carried into the answer. If it is five slightly different descriptions, the machine averages you into the invisible middle.
Make your practice identity read identically on every platform
Consistency is how trust works for machines and patients alike. A patient who sees the same name, the same story, and the same specialty everywhere they check concludes, without thinking about it, that you are established and exactly what you claim to be. Conflicting versions produce the opposite instinct: a flicker of doubt they never mention and never get past. AI systems formalise that instinct. Fragmented identities read as small and uncertain; consistent ones accumulate every review and citation into a single recommendable entity.
Answer the patient's real question before they ask it
Patients do not search for "comprehensive care." They search from inside a fear: how much will this cost, will it hurt, can they see me soon, do they treat people like me. A practice whose visible presence answers those questions specifically becomes the answer to them. This is why complete profiles receive 7x more AI-referred clicks: completeness is not bureaucratic thoroughness, it is the machine-readable version of "yes, we handle exactly your situation." Every unanswered question on your profile is a patient question your competitor gets to answer instead.
Replace vague authority language with citable specifics
"We offer state-of-the-art implant dentistry" asks the patient to take your word. "Full-arch implant cases restored in-house, from scan to final prosthetic" gives them a fact to hold onto, and gives an AI a sentence it can repeat with attribution. Specificity reads as competence to a human and as citability to a machine. The patient who arrives holding one of your specifics has already rehearsed choosing you. That is what pre-sold means, and it is why specific practices convert high-value consults at rates generic practices interpret as luck.
Let your reviews prove the positioning you claim
Patients trust other patients to tell them what you are actually like, and AI systems read reviews the same way: as third-party evidence of what you are known for. When the thing you claim and the thing your patients describe match, both audiences believe it. The positioning move is to earn reviews around your signature work, by asking the patients whose cases embody it, so that the public record of your practice tells the story you chose rather than a generically pleasant one.
Be present where the machines verify, not just where patients browse
A patient asking ChatGPT never sees Bing, but ChatGPT does. A patient asking Perplexity never reads your schema markup, but Perplexity does. The visible 8% understand that part of their audience is now machine intermediaries who check credentials, indexes, and directories before ever putting a practice in front of a human. Being verifiable in those places is what lets your name survive the machine's confidence test. The patient only ever sees the result: your practice, recommended by name, with reasons.
How Do You Test Where Your Practice Stands Today?
You do not need to take any of this on faith. Dental AI search has a property most marketing questions lack: it is directly observable, tonight, for free.
Ask the machines what they tell patients about your market. Open ChatGPT, Perplexity, and Google, and pose the questions your highest-value patients pose: best implant dentist in your city, a dentist for someone terrified of dentists, who can place an implant near your suburb this month. Then read the answers the way a patient would, which means noticing only who is named.
There is a structured version of this: the five-prompt dental AI search test walks through the exact prompts, the platforms, and how to score what comes back. Most owners who run it experience the same thirty seconds Dr. Okafor did: the market answers are confident, specific, and do not include them.
Two cautions as you run it. First, do not test with your own name. "Dr. Okafor Columbus" will find you; patients who already know your name were never the question. Test the way strangers search: by service, by fear, by location. Second, do not stop at one platform. Each system reads a different door of your presence, and being visible on one while absent from another tells you precisely which signal is weak.
What the test gives you is a baseline. What the baseline gives you is the ability to watch it move.
How Long Before You See a Difference?
Faster than positioning work usually pays off, because AI systems re-read the world continuously.
In the audit data, practices that repair their signals tend to see movement in stages. Identity consistency and profile completeness register first, often within weeks, because they resolve the machine's most basic doubts about who you are. Citable service content and review evidence build more gradually, and the named-recommendation threshold, the moment you start appearing in answers, typically follows the signal work rather than accompanying it. Practices in this pattern have generally measured the shift in months, not years. No one can promise you a date, and you should be suspicious of anyone who does. What the data supports is direction: clear signals get read, and read signals get recommended.
The part worth sitting with is what compounds while you wait. AI systems weight established, consistently cited entities more heavily over time. Every month your signal stays generic, the practices being named in your market accumulate exactly the kind of authority that gets harder to displace. The cost of starting later is not the months you waited. It is the position someone else consolidated during them.
Dr. Okafor started with the test, because the test was free and the silence was not.
She made her decisions in one week. One identity, everywhere. One sentence about what the practice was known for. Service descriptions a machine could quote. A profile with nothing left blank.
Ninety days later she ran the five prompts again.
Named in three of five.
The month after that, the front desk flagged the intake forms.
"ChatGPT" had stopped being a surprise.
The implant consults arriving that way did not ask her to convince them. They arrived already knowing why they had chosen her, because a machine had told them, accurately, what her practice stood for.
Her schedule was no fuller than the year before. It was fuller of the right patients.
If your intake forms have never once said "ChatGPT," that is not evidence the shift missed your market. It is evidence of which side of the answer you are on.
Here is the thread that ties this whole guide together. Everything you want more of, higher case acceptance, a better case mix, patients who arrive pre-sold rather than price-shopping, runs through the same gate: whether the places patients now ask first can see your positioning at all. AI search and Google Maps are that gate. 82% of dental searches end in a Maps interaction, and the AI layer reads the same signals Maps does. One clear identity feeds both.
You have spent years building something worth finding: the clinical skill, the team, the reviews, the reputation that walks in the door with every referral. None of that is the problem. The problem is that a new layer of patient discovery grew up between you and your market, and it can only recommend what it can read.
The practices that win the next five years of patient flow will not be the ones that spent the most. They will be the ones whose positioning was clear enough for a machine to repeat. That is a fixable state, and the fix starts with knowing exactly where you stand.