Key Takeaways
- A dental practice can be named by an AI chatbot without ever being genuinely recommended to a patient searching for care
- Nearly half of US consumers now use AI tools like ChatGPT for local business recommendations, up sharply from the year before
- A simple 15-minute test using ChatGPT, Google AI Mode, Gemini, and Perplexity can reveal exactly where a practice stands on the AI visibility ladder
- Fixing AI visibility problems works best in order: correct wrong facts first, clarify services and location second, then strengthen outside corroboration
- A free scorecard from DataDriven Digitals helps practices track whether AI answers mention, cite, and correctly recommend them over time
A parent staring at a chipped tooth at 9 p.m. used to type “emergency dentist near me” into Google. Today, that same parent might just as easily ask a chatbot, “Which dentist near me takes Delta Dental and has Saturday hours?” The answer that comes back, not a list of ten blue links, decides which practice gets the call.
45% of Patients Now Ask AI First
That shift is no longer a hunch. Roughly 45% of US consumers used ChatGPT or another AI tool to get a local business recommendation in the past year, a steep jump from just 6% the year before. Trust is climbing right alongside usage: 42% of consumers now say they trust AI platforms as much as traditional reviews when deciding where to go.
ChatGPT has reached 900 million weekly active users. Google’s AI Overviews now reach more than 2.5 billion monthly users, and AI Mode has already surpassed 1 billion. The effect on click-through is measurable too: when a Google AI summary appears above the results, people click through to a regular search result 8% of the time, compared to 15% without one. Fewer clicks means the AI-generated answer itself is often the entire shortlist a patient sees.
A deeper walkthrough of the exact prompts, scorecard, and repair plan referenced here lives in a companion piece, can AI recommend your dental practice, which breaks down the full 15-minute test alongside a 30-day fix plan. The rest of this guide covers the same ground: how to run the test, read the results, and know what to fix first.
Six Levels of AI Visibility
Most practices think about AI visibility as a yes-or-no question. It works better as a ladder with six rungs, because “not showing up in ChatGPT” can mean six very different things.
- Not found: The AI has no record of the practice at all, even when asked by name. This usually points to a discovery or crawl problem.
- Branded only: AI answers correctly when asked about the practice by name, but never surfaces it for a generic search like “dentist in [city].” AI knows the practice exists but has no reason to pick it.
- Mentioned, not cited: The practice name shows up with no link back to the website, suggesting the answer relies on stored information or a third-party source instead.
- Cited, but wrong: AI links to the practice but gets hours, insurance, or services wrong, a sign of conflicting information floating around online.
- Wrong fit: The practice gets recommended for services it doesn’t offer or patients it doesn’t typically see, often because service pages or online profiles are vague or outdated.
- Accurately recommended: The practice appears for a relevant, unbranded search, with every detail correct. This is the goal.
A practice stuck at level three and a practice stuck at level zero both technically “aren’t showing up in ChatGPT.” Their fixes have almost nothing in common, which is exactly why knowing the level matters more than knowing the score.
The 15-Minute Test
Running this test takes a phone or laptop, a simple spreadsheet, and about fifteen minutes. Testing from a private or logged-out browser window helps keep past chat history from skewing the results.
Step 1: Choose Your Four Platforms
Focus on the four places patients are most likely to ask a question out loud or type it into a search bar:
- ChatGPT, with search turned on
- Google AI Mode, or an AI Overview inside regular Google Search
- Gemini
- Perplexity
Location matters here. ChatGPT, for instance, relies on an approximate location based on IP address for local results, so a search run from the front desk only reflects the office’s own neighborhood. Adding the city or ZIP code directly into each prompt keeps results consistent across platforms.
Step 2: Ask Branded Questions
Branded prompts check whether AI actually knows the practice, using its real name in each question:
- “What services does [Practice Name] offer?”
- “Is [Practice Name] accepting new patients?”
- “Does [Practice Name] offer emergency appointments?”
- “What insurance does [Practice Name] take, and do they offer financing?”
Wrong answers here tend to be the fastest wins to fix. If AI tells a patient the office is closed on Saturdays when it isn’t, that patient simply never calls.
Step 3: Ask Unbranded Discovery Questions
Unbranded prompts test something more valuable: whether AI actually chooses the practice among the competition. This is where new patients come from.
- “Dentist accepting new patients in [City, State]”
- “Emergency dentist open on Saturday in [City]”
- “Implant dentist in [City] with financing”
- “Dentist for anxious patients in [City]”
- “Family dentist near [Neighborhood or ZIP code]”
Only test services genuinely offered. If implants aren’t part of the practice, an implant search that skips right over it means the system is working correctly, not failing.
Score Your Results
Each prompt earns one point per “yes” across five checks, for a possible score between 0 and 5:
- Mentioned: Does the practice name appear in the answer at all?
- Linked or cited: Does the answer link to the website or a profile the practice controls?
- Details accurate: Are hours, address, phone number, insurance, and new-patient status all correct?
- Correct service fit: Is the practice recommended for what was actually asked?
- Clear booking path: Does the answer point toward a phone number, booking link, or page where a patient can act?
Running nine prompts across four platforms produces 36 scored results. That volume matters, because a single answer says very little on its own; a repeated pattern across platforms and prompts says a great deal.
What Your Score Pattern Reveals
Strong branded scores paired with weak unbranded scores usually mean AI knows the practice exists but doesn’t have enough evidence to select it for a specific service or location, pointing toward treatment pages and outside corroboration as the first things to check. A practice mentioned often but rarely linked may be leaning on stored information or third-party sources rather than the practice’s own site, which is worth checking against crawl access and indexation settings.
When a practice gets cited with incorrect details, that almost always traces back to conflicting information sitting somewhere online, across the website, Google Business Profile, Bing Places, or a directory listing. Getting recommended for the wrong treatment tends to mean service pages are too vague or mixed between providers and locations. And a practice recommended accurately but with no clear booking path has a mostly solved visibility problem sitting on top of a weak conversion path, worth checking through the phone number, appointment page, and calls to action.
Results that swing wildly between platforms usually mean each one is drawing from a different mix of indexes and sources, so recording which sources get cited helps address each gap individually. Weak scores everywhere, branded and unbranded alike, point to a broader discovery and entity-clarity problem worth tackling from the ground up.
Why AI Finds You But Won’t Recommend You
Low scores tend to trace back to one of four layers of trouble. Working through them in order matters, since a problem in an earlier layer can quietly limit everything that comes after it.
Discovery: Can AI Reach Your Information?
Key pages that aren’t indexed or crawlable simply can’t show up in an AI answer. Google’s own guidance states that content only needs to be indexed and eligible to appear in regular Search to also surface in AI Overviews and AI Mode, no extra step required, and no workaround makes up for a page Google can’t index in the first place.
Blocking AI crawlers is a quieter, more common problem. OpenAI documents OAI-SearchBot for search and GPTBot for model training as two separate controls: blocking OAI-SearchBot keeps a site out of ChatGPT search answers beyond basic navigational links, while blocking GPTBot simply tells OpenAI not to use that content to train its generative AI foundation models. A practice can block training access while still allowing search access, a distinction plenty of websites get backward. Perplexity’s guidance works a little differently: PerplexityBot respects robots.txt, and blocking it removes a page from Perplexity’s citation-driven answers, though Perplexity may still index the domain, headline, and a brief factual summary even when a page is blocked. Aggressive firewall or bot-protection settings can block any of these crawlers without anyone on staff ever noticing.
Important facts also need to live where a machine can actually read them. Hours buried in an image, an insurance list locked inside a PDF, or services hidden inside a booking widget are all invisible to AI even when a human patient can see them perfectly well.
Understanding: Does AI Know What You Do?
Vague service descriptions leave AI with nothing to match against a specific search. A phrase like “family and cosmetic care” gives an AI system no way to connect the practice to a search for “implant dentist with financing.” Location clarity matters just as much; if the city, neighborhood, or service area is buried in a footer or missing altogether, AI has less to work with when narrowing down a local answer.
Confusion between a dentist’s personal profile, the practice brand, and a second location can also blur AI’s understanding of what’s actually being offered where. And useful facts sometimes get buried under marketing language: a line like “we’re passionate about your smile” pushes something genuinely useful, like “new patients seen within 48 hours,” further down the page where it’s harder for both patients and AI to find.
Verification: Do Other Sources Agree?
AI tools often draw on more than one source when building an answer, and patients tend to double-check anyway. Common conflicts that trip up an AI answer include a name, address, phone number, or set of hours that differs between the website, Google Business Profile, Yelp, and dental directories. A website that says “accepting new patients” while a directory says otherwise creates the same kind of confusion, as does an insurance list that doesn’t match across sources or a claimed specialty with no review or profile to back it up.
When sources disagree, AI doesn’t always pick the correct one. It might repeat the wrong version, hedge with uncertainty, or leave the practice out of the answer entirely.
Patient Fit: Does AI Have a Reason to Choose You?
An AI system answering “dentist for anxious patients” needs actual evidence that a practice fits that description. That evidence tends to look like a page explaining sedation options and what a first visit feels like, reviews that genuinely mention anxiety or feeling at ease, and clear, unscripted answers about cost, insurance, and availability. Without that evidence, a practice is simply a dentist in the right city. With it, a practice becomes the dentist that actually fits the question being asked.
Confirmed Facts vs. Unproven Claims
A lot of advice floating around about AI visibility is confident guesswork dressed up as fact. Google’s AI optimization guide draws some clear lines worth repeating.
Confirmed by Google: pages must be accessible and indexable to appear in AI features, unique and useful content genuinely helps visibility, and optimizing for generative AI search is still fundamentally optimizing for the search experience, meaning it’s still SEO. Also confirmed, from OpenAI directly, is that allowing its OAI-SearchBot crawler keeps a site eligible for ChatGPT search citations.
Treat with more caution: claims that any single directory controls all AI recommendations, fixed promises like “rank in ChatGPT within 30 days,” and the idea that a special llms.txt file boosts AI visibility. Google has been especially direct here, stating plainly that no new machine-readable files, AI-specific text files, or special markup are needed to appear in AI features, and that these files “will neither harm nor help” visibility. The same guide states that structured data “isn’t required for generative AI search, and there’s no special schema.org markup you need to add.” Schema still helps search engines read an address, set of hours, or service list cleanly, so it’s worth keeping in place, but it won’t leapfrog a practice with clearer, more useful pages past better competition.
A 30-Day Plan to Improve Your Standing
Fixing AI visibility works best in a specific order: facts first, clarity second, corroboration third, then a fresh measurement to confirm progress.
Week 1: Fix Factual Conflicts
Audit and align business details everywhere they appear: the website, Google Business Profile, Bing Places, Yelp, Healthgrades, Zocdoc, insurer directories, and any relevant dental association listing. Check the practice name, address, phone number, and hours (including holiday hours), dentist names and credentials, services and locations, new-patient status, and insurance or financing wording. Bing Places deserves particular attention, since Microsoft is listed among ChatGPT search’s third-party providers, and plenty of practices have never touched that listing at all.
Week 2: Clarify Treatments and Location
Strengthen the highest-value treatment pages first, things like implants, emergency care, clear aligners, sedation, and pediatric dentistry. State the location and service area plainly on every relevant page rather than burying it in a footer, and answer the questions patients ask before booking: cost range, insurance, what a first visit looks like, and how soon they can get seen. Connecting treatment pages directly to location pages helps make the service-and-place relationship obvious to both patients and AI systems parsing the content.
Week 3: Strengthen Corroboration
Review the accuracy and completeness of key dental and local profiles, then work on earning genuine local mentions through community sponsorships, local news coverage, school partnerships, or charity involvement. Review recency matters more than most practices assume, since a large share of consumers specifically look for reviews written within the last few months before trusting them. Asking every patient consistently for a review, without ever scripting what they say, builds that recency and detail naturally over time.
Week 4: Retest and Compare
Retesting only tells a useful story if the conditions stay fixed. Use the same platforms, the same prompt wording, and the same testing location, then record mentions, citations, accuracy, and service fit in the same scorecard used during the original test. Some fixes show results within days; others take longer while crawlers and directories catch up, so a 30-day retest shows direction more than a finished outcome.
Being Found Is Not Being Chosen
A screenshot of ChatGPT naming a practice by name feels like a win, but it proves very little on its own, since the next patient asking the same question might get a completely different answer. Real progress shows up in a pattern: mention rates climbing across branded and unbranded prompts, citations linking back to the practice’s own site, and factual accuracy holding steady across repeated tests.
The measure that matters most, though, sits outside any chatbot window entirely. Visibility that never turns into a phone call or a booked appointment leaves the work unfinished, no matter how good the AI answer looks. Adding a simple “How did you hear about us?” option at the front desk, with “ChatGPT or AI assistant” as a choice, connects everything measured online back to what’s actually happening in the operatory.
For a practical next step, working through a free AI visibility check from DataDriven Digitals can show exactly where a practice stands across a broader set of buyer questions and AI platforms before committing time to a full 30-day fix plan.
DataDriven Digitals
hello@datadrivendigitals.com
8005 Financial Drive, Unit A4
Brampton
Ontario
L6Y 6A1
Canada
