AI Dental Clinic Recommendations Study: Gemini Favors Independent Clinics, Claude Leans Toward Chains

AI Is Becoming the New Gateway for Patients Seeking Dentists

Seventy percent of patients check online reviews before choosing a dentist. In the past, that meant Google and Doctoralia — but today, more and more people are simply asking AI. Unlike search engines that return ten results, AI only offers three or four names — and who makes that shortlist is becoming a new battleground for clinic patient acquisition.

GenScore conducted an empirical study in Madrid and Valencia, Spain, asking ChatGPT, Gemini, and Claude the same patient-perspective questions and recording which dental clinics were mentioned. The results reveal a fragmented and highly divergent AI recommendation ecosystem.

Three Key Numbers: The Brutal Truth About AI Recommendations

77% of independent clinics never appeared in any AI response. Starting from independent clinics with strong online reputations, more than three-quarters were not mentioned across all three engines.

Only 7% of recommended clinics appeared in responses from more than one engine. In other words, AI recommendations have virtually no cross-platform "consensus list."

Two-thirds of Claude's responses mentioned dental chains, while Gemini did not recommend a single chain across tests in both cities.

These numbers show that AI recommendations are not a unified ranking, but rather independent cognitive worlds constructed by each model.

City-Level Differences: Madrid vs. Valencia

Madrid: Chains Dominate Claude, Independent Clinics Hold Ground on Gemini

In Madrid, Sanitas Dental (including Milenium clinics) ranked first with 19% overall appearance rate, but all of its exposure came from Claude (58%), with zero mentions from Gemini and ChatGPT. Ferrus & Bratos ranked second at 17%, driven primarily by Gemini (42%), with no mentions from Claude. Vitaldent ranked third at 14%, again with Claude contributing 33% of mentions.

Citadental's case was even more extreme: it appeared in 33% of Gemini's Madrid responses, but was completely absent from ChatGPT and Claude.

Valencia: Independent Clinics Have the Edge, Engine Divergence Equally Pronounced

Valencia's recommendation landscape was somewhat different. Clínica Llobell and Asensio Odontología Avanzada tied for first place (22% each), with nearly all their recommendations coming from Gemini (50% each). Vitaldent ranked third at 17%, but its recommendations came entirely from Claude (50%).

Clínica Badal made the list at 11%, the only clinic ChatGPT specifically recommended in Valencia (33%), with no mentions from Gemini or Claude.

Four Notable Phenomena

Each Engine Has Its Own "Preference List"

A cross-engine overlap rate of just 7% means that being recommended on one AI almost guarantees nothing about being seen on another. Monitoring only one engine produces a severely distorted picture.

Claude Systematically Favors Chains

Two-thirds of Claude's responses mentioned chains, primarily Vitaldent and Sanitas Dental. Gemini recommended no chains in either Madrid or Valencia, while ChatGPT mentioned them occasionally but far less frequently than Claude.

Good Reviews Don't Equal AI Recommendations

The study started from independent clinics with excellent reputations, yet more than three-quarters were never mentioned by any AI. Even the most-recommended independent clinic appeared in fewer than one-quarter of responses. The market is extremely fragmented — no independent clinic can consistently secure an AI recommendation spot.

Directory Sites Are a Key Information Source for AI

When AI cited sources, Top Doctors, Doctoralia, and Doctory were referenced in 40% of Valencia responses and 25% of Madrid responses. Clinic websites and some media outlets were also cited.

AI citations and traditional search engine rankings are two different mechanisms. Search engines return a list of links that users must click and evaluate themselves; AI, when generating responses, "reads" several web pages and integrates their content into a natural language answer — the pages it reads are the citations. For clinics, this means appearing on directories like Top Doctors, Doctoralia, and Doctory isn't just about being seen directly by patients — it's about being read as trusted source material by AI. Directory platforms typically have structured formats, complete fields, and real patient reviews, making them easier for AI to crawl and trust. Therefore, a clinic's information completeness on directories, name consistency, and review volume may all indirectly affect its probability of being mentioned by AI — not just its traditional search ranking.

The 7% cross-engine overlap rate essentially reflects the divergence among the three models in training data, retrieval mechanisms, and brand priors. ChatGPT, Gemini, and Claude are trained by different teams with different corpus sources, update frequencies, and web-access capabilities, so the "cognitive worlds" they construct are not aligned. The implication for clinics is clear: AI recommendations are not a unified ranking that can be optimized once, but a result that must be managed separately across multiple platforms. This contrasts sharply with the traditional search engine logic of "optimize once, benefit everywhere," suggesting that visibility management in the AI era is closer to multi-platform operations than a single-point breakthrough.

Why Gemini Is More Willing to Recommend Independent Clinics

This is the most striking engine difference in the study. Gemini recommending independent clinics while Claude recommends chains may have several explanations.

Gemini relies on Google Search. Before answering, it queries Google results, and "dentist in [city]" is inherently a local search where clinic business information, reviews, and location carry high weight. Under this logic, a well-reviewed community clinic can compete on equal footing with chain brands.

Gemini reads specific pages, not brands. A page explaining dental implant treatment in a city — who does it, how it's done — is direct source material for AI responses. Chain brands typically only have generic city pages lacking the depth of specific clinic information.

Gemini trusts directories with patient reviews. Platforms like Top Doctors and Doctoralia list specific professionals and clinics, not big brands. Having solid information on these directories is equivalent to being on Gemini's reading list.

Claude relies more on training memory. Chain brands have more advertising, more media coverage, and more accumulated online presence — exactly the kind of content that parameter-memory-dependent models are more likely to output.

It's worth noting that these explanations are inferences by the researchers based on data and engine mechanisms, not causal conclusions directly measured by the study.

The distinction between RAG and parameter memory is the key technical context for understanding this phenomenon. Gemini retrieves external web pages in real time before answering (i.e., Retrieval-Augmented Generation), so its recommendations are based on currently crawled specific content — an independent clinic's website page, patient reviews on directories, Google business information. Claude, in these types of tests, relies more on knowledge solidified in model parameters during training, where information distribution naturally favors heavily-exposed, frequently-mentioned brands. This means that even a clinic with an excellent reputation may struggle to enter the shortlist of parameter-memory-dependent models if its information exists only on a few specific pages without large-scale online presence. Conversely, if a clinic's page information is well-structured and indexed by authoritative directories, it's more likely to be captured by real-time retrieval engines. This also explains why the same clinic can show vastly different visibility across different engines.

Actions Clinics Can Take

For dental clinics hoping to gain visibility in AI recommendations, the study offers four suggestions:

• Maintain complete and updated clinic information on directory platforms that AI frequently cites, ensuring names match the official website
• Create clear, standalone pages for each treatment type and city, leading directly with answers to typical patient questions
• Keep a consistent clinic name across the web to avoid AI failing to correctly attribute information due to name inconsistencies
• Monitor all three engines — ChatGPT, Gemini, and Claude — simultaneously, since results on one engine don't automatically carry over to another

Methodology Notes

The study used an algorithmic system to generate the most relevant search questions in the dental field, conducting multiple deterministic analyses on ChatGPT, Gemini, and Claude in Madrid and Valencia.

Test questions simulated real patient searches, including "What is the best dental clinic in Valencia," "Where can I get dental implants in Madrid," and "Recommend an invisible orthodontics clinic" — without mentioning any specific clinic names. Only clinics whose names explicitly appeared in AI responses were counted.

Data was collected on October 10, 2026. The researchers emphasize this is a snapshot at a specific point in time, that results may vary on different dates, and that the study does not judge the service quality of any clinic.

Additional Context

Local Search is an important concept for understanding Gemini's behavior. When users query "dentist in [city]" on Google, the search engine prioritizes business information with geographic location, operating hours, and user ratings — these results are known as the local search pack. When Gemini answers similar questions by calling on Google's search results, it effectively inherits this local search ranking logic — distance, reviews, and information completeness all affect ranking, while brand size is not a decisive factor. This is exactly why community independent clinics can compete alongside chain brands: as long as local information and patient reviews are solid enough, there's a chance of being included in the answer. In contrast, models without web retrieval capabilities can only rely on brand names memorized during training, making it naturally difficult for small local businesses to enter their field of view.

One-line Summary

AI is becoming the new gateway for patients choosing dentists, but recommendations across the three major engines are highly divergent — only 7% of clinics appear across engines.

Paragraph Summary

Research shows AI is becoming a new gateway for patients seeking dentists, but its recommendation ecosystem is highly fragmented. GenScore tested ChatGPT, Gemini, and Claude with identical questions in Madrid and Valencia, finding that 77% of well-reviewed independent clinics were never mentioned by any engine, and only 7% of recommended clinics appeared on more than one engine. Two-thirds of Claude's responses recommended chains, while Gemini recommended no chains in either city, favoring independent clinics instead. Researchers attribute this to Gemini's reliance on real-time Google retrieval and Claude's reliance on training memory. Directory sites are important citation sources for AI. The study recommends clinics maintain directory information, create standalone pages for treatments and cities, unify their names across the web, and monitor multiple engines simultaneously.