AI Visibility of Dentists, City by City: How Often ChatGPT, Gemini and Perplexity Recommend Local Practices

Methodology

In each city we wrote 30 prompts that sound like real patients, not marketers. They covered four kinds of questions: "best of" requests, comparisons between practices, specific needs, and urgent situations. We ran them across three AI engines — ChatGPT, Gemini, and Perplexity, and asked the highest-intent questions more than once, since answers can shift between runs. That adds up to 138 valid answers per city.

From those answers we counted every practice the AI named. Each recommendation is an exact quote from the answer text — no guessing, no interpretation. We store the raw answers in full, so every number on this page can be rechecked against the original data.

We measure frequency, not rank. AI assistants don't have a fixed ranking the way Google search results do — the same question can get a different answer tomorrow. So the honest metric is how often a practice gets recommended across many realistic questions.

A note for practices that already invest in SEO: this is a different measurement. SEO is about where you rank on Google. This benchmark is about whether AI assistants recommend you when a patient asks. A practice can do well on one and be invisible on the other.

In each city the tracked group covers the metro area — the city plus its suburbs — and includes every practice we could confirm as a real, operating business, including every practice the AI engines themselves named. That is on purpose: AI answers a question about "Dallas" with names from the whole metro.

To compare the cities fairly, we used the same prompt structure and the same three engines in each one. What changes from city to city is the local set of practices and how the AI talks about them — not our method.

Results: city by city

The clearest pattern is how far the "never mentioned" share moves between cities: 17% of tracked practices in Phoenix, 26% in Miami, and 68% in Dallas. Dallas stands out sharply — two-thirds of its tracked practices are invisible, while in Phoenix and Miami the invisible share is far smaller (though the typical practice there still scores under 4 out of 100). This is an observed comparison across cities, not a controlled experiment (see the limits below).

One thing holds in all three cities: there is room on the shortlist. An average answer named between about 5.6 and 6.4 practices. The AI is willing to recommend several names at once — most tracked practices simply weren't among them.

The same pattern underneath: an open shelf

Even the most-visible practice in each city is far from dominant in absolute terms. The best visibility score we recorded was 27.0 in Dallas, 26.3 in Miami, and 17.5 in Phoenix — all far below 100. No practice in any of the three cities has come close to owning AI recommendations.

The field is wide open, too. In every city, the five most-mentioned practices captured only 12–14% of all recommendations — attention is spread across hundreds of names, not locked up by a handful. For a practice starting from near-zero, that combination — barely-visible leaders and an open field — is the opportunity.

Where AI gets its information

In every city, AI leaned on the same short list of third-party platforms — plus the practices’ own websites, which made up the large majority of sources. Here is where each city’s recommendations came from.

Individual practices are shown only as a combined total — we never publish specific clinic names.

Limitations

• These are snapshots from single measurement days — Dallas on July 7, 2026, Miami and Phoenix on July 12, 2026. AI answers change over time, and the same question can get different answers on different runs.
• The three cities were measured on two different dates, so some of the difference between them could reflect timing, not only location.
• We tracked a different number of practices in each metro — 262 in Dallas, 89 in Miami, 87 in Phoenix. The percentages describe each tracked group, not every dentist in that city.
• We used 30 prompts per city. They cover common patient questions, but different wording could surface different practices.
• This is an observational comparison across cities, not a controlled experiment — it shows differences, not what causes them.
• Visibility measures how often a practice is recommended by AI — it says nothing about the quality of care a practice provides.
• Nothing here is a guarantee. No one can promise that an AI assistant will recommend a specific business — anyone who promises that isn't being honest. Closing specific gaps raises the probability; it does not guarantee the outcome.