A well-reviewed cosmetic dentist in Philadelphia — and AI keeps sending patients elsewhere.

Your next veneers patient is asking a robot for a referral

The way people pick a cosmetic dentist changed in the last two years. A lot of them no longer scroll through Google and compare ten websites. They ask an AI assistant and trust the short list it gives back.

When a patient types "best cosmetic dentist in Philadelphia for veneers" into ChatGPT, they don't get ten blue links to sort through. They get a written answer that names two or three practices, with a sentence on why each one is good. For that patient, those two or three names are the choices. Everyone else may as well not exist.

This isn't a someday thing. In our own work, roughly a quarter to a third of new inquiries already come from people who found a business through an AI assistant. It is one of the busiest doors into the business, and for high-value cosmetic work it's growing fast.

We measured it — we didn't guess

Every number in this report comes from a live test on May 31, 2026. No opinions, no estimates.

• We asked the four assistants real patient questions.  ChatGPT, Perplexity, Claude and Gemini — the 10 questions a Philadelphia patient actually types when shopping for veneers, a smile makeover, Invisalign, or a full-mouth restoration. That's 40 recorded answers.
• We checked two things in each answer.  Did the assistant pull the practice's site in as a source, and did it name the practice in the words the patient reads?
• We noted who won instead.  Every time the practice wasn't named, we recorded which practices were.
• We looked under the hood of the site.  To see why the assistants reach for other names — covered in section 04.

We scoped every question to Philadelphia, and several to a specific Center City neighborhood — because "near me" and "in Philadelphia" is exactly how patients phrase these to an assistant.

Ask the question a patient asks — and listen to the answer

Here is a real, word-for-word answer from the test. We asked ChatGPT the most valuable question there is: who is the best cosmetic dentist in Philadelphia for a full smile makeover.

Read that again. ChatGPT named three other practices and the patient never heard about the one we reviewed — a practice with a long-running local "Top Dentist" recognition behind it. This is the whole problem in one sentence: the reputation that should win you the patient is working for someone else.

Now here is every question we asked, and who the assistants named in each.

The one place you reliably win is the question that names the neighborhood out loud — the assistant can't miss the address. But on the five veneers questions — the exact work this practice is most awarded for — it is named zero times. The patient who's ready to spend on veneers hears three other names.

And it's worst where it counts most. Here's the same finding broken out by assistant — including ChatGPT, which more patients use than the other three combined:

*Gemini routes its sources through a redirect we can't follow, so for Gemini we counted only what it said out loud, not what it linked.

It's a plumbing problem, not a reputation problem

This is the good news. You didn't lose these patients because the competition is better than you — they aren't. You lost them because the assistants can't read what makes you better. And that's fixable on the site you already own.

Put simply: the practice that should win these answers on reputation is losing them on plumbing. The names getting picked aren't more qualified — their information is just easier for a machine to read and repeat.

A 60-day pilot to get you named

Not a new website. We make the reputation you already earned something the assistants can read and repeat — and we measure the before and after on the same test you just saw.

Why this should move fast. The assistants build these answers by searching the live web in the moment — so we're not waiting months for anything to "relearn" you. And because the site is an established, trusted domain that Google and Bing already crawl constantly, new pages and labels get picked up in weeks, not months. The slow part that holds back a brand-new website doesn't apply.

Honest about the clock: by day 30, expect the first new citations to surface — fastest on Perplexity, with ChatGPT starting to break off zero. By day 60, steady gains on the winnable questions: the cost, comparison, financing and neighborhood queries where the assistants are guessing with generic national pages today. The single most competitive question — "best cosmetic dentist in Philadelphia" — sits on a longer 3-to-6-month track, because the practices holding it have years of reviews and write-ups behind them. The 60 days proves the approach works and banks the early wins; it's the start of the climb, not the finish.

The same questions, 90 days from now

Here is the result we're aiming for if the plan runs correctly — the same 10 questions, re-scored. This is a target we hold ourselves to, not a guarantee, and the next section is exactly how we check it.

Measuring the patients, not just the mentions

Getting named is the point, but you should see it in the schedule — not only in a report. Here is how we connect AI visibility to real new patients, and why one of these four matters most for a practice like yours.

• The leading indicator — this exact test.  We re-run the same 10 questions across all four assistants at day 30 and day 60. Citation share is provable and repeatable, measured against the baseline you're reading now.
• The click path — an "AI assistants" report in Google Analytics.  We tag every visit that arrives from ChatGPT, Perplexity, Gemini and Copilot, and watch what those visitors do: booking requests, contact forms, click-to-call.
• The path that matters most for you — "How did you hear about us?"  One question on the booking form and at the front desk, with an "An AI assistant (ChatGPT, etc.)" option. Most of these patients arrive by phone or by name, not a tracked click — so this is where the real signal lives.
• The dollars.  AI-attributed new patients × your average case value. Because veneers and smile makeovers are four- and five-figure cases, even a handful of attributed patients is real money on the books.

Honest caveat: no tool catches everything. Analytics reliably sees only about a third of AI-driven visits — the rest show up later as a "direct" visit or a plain Google search for your name. That's exactly why the front-desk question carries the most weight, and why we report a defensible range rather than a made-up exact number.