When a Patient Asks AI “Which Dentist Should I Choose,” What Does It Answer?

We asked exactly the same question to three AI platforms — ChatGPT, Perplexity, and Google AI Overview — and documented every response.

The result: three different answers, three distinct recommendation logics, and concrete lessons for any dental practice in Paris that wants to appear in these results.

This article shows you exactly what your patients see when they ask AI for a dentist recommendation. And why what appears (or doesn’t appear) is no accident.

The Test Protocol

We tested 5 queries worded as a real patient would phrase them, on each of the three platforms. Here are the queries used:

Query 1: “What is the best dentist in the 15th arrondissement of Paris?” Query 2: “I’m looking for a dentist for a dental implant in Paris 15th, which one do you recommend?” Query 3: “Emergency dentist Paris 11th, who do you recommend?” Query 4: “Which orthodontist do you recommend in Paris, near Bastille?” Query 5: “Comparison of the best dentists in Paris 17th for teeth whitening”

For each response, we noted: the practices cited, the order of mention, the sources used by the AI, the information provided (address, specialties, reviews), and any errors or inaccuracies.

Tests were conducted in March 2026 on accounts with no search history to avoid personalization bias.

What ChatGPT Answers

ChatGPT adopts a personal recommendation tone. It doesn’t simply list names — it formulates a conversational response that feels like advice from a well-informed friend.

For the query “best dentist 15th arrondissement,” ChatGPT typically proposes 3 to 5 practices, with a short description of each one’s strengths. It often mentions the Google rating, approximate number of reviews, and one or two specialties. It typically ends with a phrase like: “I recommend checking availability on Doctolib to book an appointment.”

What ChatGPT cites as sources: ChatGPT doesn’t always show its sources explicitly, but analysis of its responses reveals a strong influence from Google Reviews (descriptions match review content), Doctolib (specialties and availability), and practice websites when they exist.

ChatGPT’s strengths for dental queries: The conversational format is reassuring for patients. Recommendations often come with useful context (proximity to metro stations, Saturday availability, languages spoken). ChatGPT consistently encourages booking, making it a genuine conversion channel.

Observed weaknesses: Information accuracy isn’t always reliable. In our tests, some addresses were incomplete, hours sometimes incorrect, and in one case, a cited practice had changed its name. A recent study of 350,000 businesses measured an accuracy rate of about 68% for business information cited by ChatGPT — meaning a third of the information can be inaccurate or outdated.

The most important observation: ChatGPT doesn’t recommend the same practices from one test to the next for the same query. A study presented at SMX Paris 2026 showed that ChatGPT explores an average of 132 unique URLs when answering the same question asked 100 times. This means your practice can appear once and disappear the next time. To be cited consistently, you need to be present on as many sources as ChatGPT is likely to consult.

What Perplexity Answers

Perplexity takes a radically different approach: it systematically cites its sources, with numbered clickable links next to each claim. It’s the most transparent of the three.

For the same query “best dentist 15th arrondissement,” Perplexity provides a structured response with thematic sub-sections. It may separate recommendations by specialty (general dentist, implantologist, orthodontist) and cites between 5 and 10 sources per response.

What Perplexity cites as sources: Doctolib is the most frequently cited source for dental recommendations in Paris (present in roughly 68% of responses during our tests). Followed by Google Maps and Google reviews, practice websites, PagesJaunes, and sometimes specialized health directories.

Perplexity’s strengths: Source transparency is an advantage for patients who want to verify information. Responses are more stable than ChatGPT’s — the same SMX Paris study shows that Perplexity uses only 40 unique URLs across 100 submissions of the same question, compared to 132 for ChatGPT. This means if you appear on Perplexity, you’ll likely stay there consistently.

Observed weaknesses: Perplexity is more factual and less “recommendation-oriented.” It provides information, but the tone is less engaging than ChatGPT’s. There’s less “I recommend” and more “here are the practices matching your search.” For a patient seeking advice, the difference is noticeable.

The most important observation: The importance of Doctolib for Perplexity is striking. A dental practice with a complete Doctolib profile — photos, detailed description, and good reviews — has a considerable advantage on this platform. If your Doctolib profile is minimal, Perplexity will overlook you in favor of a better-documented colleague.

What Google AI Overview Answers

Google AI Overview is the AI-generated summary that appears above traditional search results on Google. For local dental queries, it doesn’t appear systematically — data shows that AI Overviews trigger on roughly 40% of local queries and about 7% of queries with strong local intent.

When it does appear, Google AI Overview presents a short summary (2 to 4 sentences) synthesizing information from search results, often accompanied by the usual Map Pack (the top 3 Google Maps listings).

What Google AI Overview cites as sources: Unsurprisingly, Google draws from its own ecosystem. Google Business Profile listings are the primary source, followed by websites with structured data (Schema.org) and quality content indexed in its organic results.

Google AI Overview’s strengths: It’s the most reliable platform in terms of local data accuracy. The SOCi 2026 study measured a 100% accuracy rate for business information cited by Gemini (Google’s AI model), compared to 68% for ChatGPT and Perplexity. The Google Maps anchor guarantees accurate addresses, hours, and contact details.

Observed weaknesses: Google AI Overview is conservative in its recommendations. It synthesizes rather than recommends. The format is shorter and less engaging than ChatGPT or Perplexity. And importantly, it doesn’t appear for every dental query — some searches only show the traditional Map Pack without an AI layer.

The most important observation: Google AI Overview is the only channel where your Google Business Profile has a direct, measurable impact. Optimizing your GBP profile simultaneously improves your visibility in the classic Map Pack AND in Google’s AI responses. It’s the lever with the best return on investment.

The 5 Key Differences Between Platforms

After analyzing dozens of responses, here are the five most consequential differences for a dental practice.

Difference 1: Response format. ChatGPT gives conversational advice (“I recommend…”). Perplexity provides a documented synthesis (“according to reviews and sources consulted…”). Google AI Overview summarizes existing results. The patient receives a different experience depending on the platform they use.

Difference 2: Preferred sources. ChatGPT explores broadly (Google reviews, Doctolib, websites, directories, editorial content — up to 132 unique sources per query). Perplexity focuses on fewer sources but more consistently (40 unique sources, with strong Doctolib dominance). Google AI Overview relies on its own index, with GBP listings first.

Difference 3: Recommendation stability. Perplexity is the most stable: ask the same question 10 times and you’ll get very similar answers. ChatGPT is the most variable: recommended practices can change from one session to the next. Google AI Overview falls in between, anchored by the Map Pack which changes little.

Difference 4: Information accuracy. Google is the most accurate (100% reliability on local data). Perplexity and ChatGPT have an error rate of about 32% on business information. If your information is inconsistent online, ChatGPT and Perplexity may cite incorrect data about you — which is worse than not being cited at all.

Difference 5: Linguistic behavior. A recent discovery: ChatGPT automatically generates queries in two languages when searching for information. If a patient asks their question in French, ChatGPT searches in French AND in English. This means if your practice has English-language content (even basic) on its website, it increases its chances of being found by ChatGPT. Perplexity, on the other hand, only searches in the language of the question.

What This Means for Your Practice

These platform differences have very concrete implications for your visibility strategy.

To appear on ChatGPT, you need to be present on as many online platforms as possible. ChatGPT explores broadly and changes its sources frequently. The more places you’re listed (Google, Doctolib, PagesJaunes, health directories, your own site, editorial content), the more likely you are to be picked up by its exploratory queries.

To appear on Perplexity, focus on Doctolib and Google Reviews. These are the two dominant sources for dental queries. A complete Doctolib profile with a good description and positive reviews is your best investment for Perplexity.

To appear on Google AI Overview, optimize your Google Business Profile. It’s Google AI’s primary and direct source. A complete listing with recent reviews, listed services, and structured data on your site positions you for both the Map Pack AND AI responses.

To appear on all three simultaneously, the common strategy is clear: consistent information everywhere, recent and specific Google reviews, a comprehensive Doctolib profile, and a website with structured data. That’s exactly what a complete GEO strategy covers.

Do the Test Yourself in 5 Minutes

Want to see exactly what AI says about your practice? Here’s how to test in 5 minutes.

Step 1: Go to chat.openai.com. Type: “Which dentist do you recommend in the [your arrondissement] of Paris?” Note whether you appear and who appears in your place.

Step 2: Go to perplexity.ai. Ask the same question. Look at the sources cited at the bottom of the response — those are the platforms the AI draws its information from.

Step 3: Search on Google for “best dentist [your arrondissement] Paris.” Check if an AI Overview box appears above the results. Are you mentioned?

Step 4: Test with a specialty query: “dental implant Paris [arrondissement] recommendation” or “orthodontist Paris [arrondissement] reviews.”

Step 5: Type your practice’s exact name into ChatGPT: “What do you know about [Practice Name] in Paris?” This query reveals what AI knows (or doesn’t know) about you specifically.

If you don’t appear anywhere, you now know why — and the levers to fix it. If you appear on one platform but not others, you know which one to optimize first.

The Reality These Tests Reveal

The most striking finding from this comparison is that the same question asked to three AI platforms produces three different answers, with different practices. A patient who consults ChatGPT won’t see the same recommendations as one using Perplexity or Google.

This means being visible on just one platform isn’t enough. A practice optimized only for Google loses patients who come through ChatGPT and Perplexity — and vice versa.

That’s the entire philosophy of GEO (Generative Engine Optimization): optimizing your presence to be recommended by all AI platforms, not just Google. Practices that understand this now and act accordingly are capturing patients their colleagues don’t even see.

---

Curious to see exactly what each AI says about your practice? Eddie Miller Agency conducts a complete GEO audit that tests your visibility on ChatGPT, Perplexity, and Google AI, with a detailed report and action plan. [Request my free GEO audit →]

---

FAQ — AI Recommendations and Dental Practices

Why doesn’t ChatGPT always give the same answer?

ChatGPT operates probabilistically: it generates responses by predicting the most likely continuation, which introduces natural variability. Additionally, it explores a large number of different sources with each query (up to 132 unique URLs according to studies). Both factors mean that recommended practices can change from one session to the next, even for the same question.

Is Perplexity reliable for choosing a dentist?

Perplexity is the most transparent of the three because it systematically cites its sources. This allows patients to verify the information. However, like ChatGPT, it can present inaccurate business information (address, hours) in roughly a third of cases. It’s recommended to always verify information directly on the practice’s website or on Doctolib.

My practice appears on Google but not on ChatGPT. Is that normal?

Yes, it’s very common. Only 14% of URLs cited by AI appear in Google’s top 10 for the same query. Ranking well on Google doesn’t guarantee being recommended by ChatGPT or Perplexity. AI has its own selection criteria, and a dedicated GEO strategy is needed to cover all platforms.

Which AI platform is most used by patients in Paris?

ChatGPT dominates with roughly 80% market share among AI chatbots. However, Google AI Overview potentially reaches more patients because it appears directly in Google search results, without the patient needing to go to another platform. Perplexity is used by a smaller but highly qualified audience (30% of its users are senior executives). Ideally, you should be visible on all three.