|
Level |
What happens |
What it usually signals |
|||
|---|---|---|---|---|---|
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
Short answer: Yes. ChatGPT, Google AI Mode, Gemini and Perplexity can all name, cite or describe a dental practice. Being found is a different result from being recommended, though, and answers shift by platform, prompt wording, location and date. No one can promise you a recommendation. You can measure where you stand, fix what’s blocking you, and test again.
This guide gives you that test.
Think of a parent at 9 p.m. with a child’s chipped tooth. Five years ago, they typed “emergency dentist near me.” Today, there’s a good chance they ask a chatbot: “Which dentist near me takes Delta Dental and has Saturday hours?”
That shift is showing up in the data:
That last number is the one to sit with. If fewer people click through, the answer itself becomes the shortlist. Your practice is either in it, described correctly, or it isn’t.
Most advice treats AI visibility as yes or no. In practice, there are six distinct outcomes, and they need different fixes. Think of them as a ladder:
|
Level |
What happens |
What it usually signals |
|||
|---|---|---|---|---|---|
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
A practice at Level 3 and a practice at Level 0 both “aren’t showing up in ChatGPT.” Their problems have almost nothing in common. Knowing your level tells you where to start.
You need a phone or laptop, a spreadsheet and 15 minutes. Use a private or logged-out browser window where you can, so past chats don’t skew results.
Test the four places patients are most likely to ask:
Note where you’re testing from. ChatGPT, for example, uses an approximate location based on your IP address for local results. A search from your office only tells you about your office’s neighborhood.
Add the city or ZIP code to each prompt to keep things consistent.
These check whether AI knows you. Swap in your real practice name:
Wrong answers here are your fastest wins. If AI says you’re closed on Saturdays and you’re not, a patient who believes it never calls.
These check whether AI chooses you. This is where new patients come from:
Only test services you genuinely provide. If you don’t place implants, a missing implant recommendation is the system working correctly. Testing for services you don’t offer produces noise and tempts you to “fix” something that shouldn’t change.
Give each prompt one point per “yes.” Every result scores between 0 and 5.
|
|
|
|
|
|
|
|||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| |||||||||||||
| |||||||||||||
| |||||||||||||
| |||||||||||||
| |||||||||||||
| |||||||||||||
| |||||||||||||
| |||||||||||||
|
How to read the columns:
Run nine prompts across four platforms and you have 36 scored results. The pattern tells you more than any single answer. Strong branded scores with weak unbranded scores point to a different problem than weak scores everywhere.
DataDriven Digital’s Search Visibility Snapshot applies the same evidence-first principle to broader Google, Bing, local and AI Overview visibility. The test in this article focuses specifically on how prompt-based AI surfaces find, describe and recommend a dental practice.
A single score is less useful than the pattern across your results. Compare branded and unbranded prompts, then look for the closest match below.
|
|
|
|||
|---|---|---|---|---|---|
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
Do not panic over one poor answer. Look for problems that repeat across several prompts or platforms. Those repeated gaps are the ones worth fixing first.
Prefer an automated baseline?
Run the free AI Visibility Preview to check your business across a broader set of buyer questions and AI platforms.
When a practice scores low, the cause usually falls into one of four layers. Work through them in order, because a problem in an early layer can limit everything after it.
robots.txt to make sure OAI-SearchBot has access. Perplexity gives similar guidance for PerplexityBot. Aggressive bot protection or firewall settings can block these crawlers without anyone on your team noticing.A useful distinction: OpenAI runs a separate crawler, GPTBot, for model training. You can block training and still allow search. Many practices block both without meaning to.
AI tools may draw on several sources when assembling an answer. Patients check other sources too: BrightLocal found that most people who use AI for local recommendations go on to verify what it tells them.
Common conflicts include:
When sources disagree, an AI answer may repeat the wrong version, express uncertainty or omit the practice.
An AI answering “dentist for anxious patients” needs evidence you’re a good fit for anxious patients.
That evidence looks like:
Without that, you’re a dentist in the right city. With it, you’re the dentist who fits the question.
A lot of AI visibility advice is confident guesswork. Here’s how the current claims line up against what Google, OpenAI and Perplexity have actually published:
|
|
||
|---|---|---|---|
|
|
||
|
|
||
|
|
||
|
|
||
|
|
Google has been unusually direct on the right-hand column. Its AI optimization guide, last updated in July 2026, says:
Schema still helps Google read your address, hours and services cleanly. It’s worth having. It just won’t jump you ahead of a practice with clearer, more useful pages.
One more caution: industry reports list anywhere from a handful to a dozen “directories that control dental AI citations.” Directories clearly matter as corroboration. No platform has published a list saying which ones decide recommendations, so treat any fixed list as a hypothesis to check against your own test results.
Use a practical order: facts first, clarity second, corroboration third, then measure.
Audit and align these everywhere they appear: your website, Google Business Profile, Bing Places, Yelp, Healthgrades, Zocdoc, insurer directories and the ADA Find-a-Dentist listing where relevant.
Check:
Bing Places deserves attention here. OpenAI lists Microsoft among ChatGPT search’s third-party providers, and plenty of practices have never touched their Bing listing.
This is where experience beats generic copy. Google’s own guide favors unique expert or experienced takes over common knowledge—a point covered in Why Experience-Led Content Is Surviving the Shift to AI Search.
For more on how independent signals build trust, see The New Trust Signals Google and AI Systems Use.
Keep everything fixed so the scores reflect your changes instead of random variation:
Some fixes show up within days. Others take weeks while crawlers and directories catch up. A day-30 retest shows direction; the full effect can take longer.
One screenshot of ChatGPT naming your practice feels great. It proves very little, because the next person asking might get a different answer.
Track a fixed set of measures over time:
|
Measure |
How to track it |
||
|---|---|---|---|
|
|
||
|
|
||
|
|
||
|
|
||
|
|
||
|
|
||
|
|
||
|
|
||
|
|
That last row matters most. Visibility that doesn’t turn into calls and booked appointments isn’t finished work.
A profile can look perfect and still sit unseen, which we covered in A Complete Google Profile Can Still Be Invisible.
No. OpenAI doesn’t guarantee placement, and answers change by prompt, location and date. Anyone promising a guaranteed AI recommendation is selling something they can’t control.
You can remove the blocks, improve clarity and measure progress.
They overlap heavily. Google says optimizing for its AI features is still SEO.
The difference is mainly in measurement. AI answers often present a shorter set of options, may draw on different sources and require a prompt-based test alongside traditional rank tracking.
Start with the foundations in Local SEO for Independent Service Businesses.
Not on its own. Google states that structured data isn’t required for AI search and that there’s no special markup for it.
Schema helps search engines read your details accurately, so keep it—but pair it with clear pages and consistent listings.
llms.txt file?Not for Google, which says such files neither help nor harm visibility. Other platforms haven’t said they use it for recommendations.
Spend that time improving your treatment pages instead.
Monthly is a good rhythm for most practices, plus a retest after any major change such as a new location, new service or website rebuild.
Use the same prompts each time.
They draw on different sources. ChatGPT search uses third-party search providers and its own crawler.
Blocked crawler access, an incomplete Bing presence or conflicting directory data may contribute to your practice appearing in Maps but not ChatGPT.
Yes. Old hours, a former dentist or a dropped insurance plan can linger on directories and get repeated to patients.
Week 1 of the plan above targets exactly this.
You can run the test above manually in about 15 minutes. For a broader automated check, run the free AI Visibility Preview to see how your business appears across more buyer questions and AI platforms.
The results provide a point-in-time automated assessment and may vary by platform, prompt, location and date.
Looking for a broader assessment of Google, Bing, local search, important website pages and AI Overviews? Request a Search Visibility Snapshot.
For a broader self-check, use The 10-Part Search Visibility Checklist.