GEO / patient behaviour

How patients actually use AI to choose a cosmetic dentist

Juan Cruz Chacama

Founder, Worth Finding · 18 July 2026

Patients do not ask an AI engine to book an appointment. They ask it to narrow the field. A patient asks something close to "who does the best veneers in Austin", the engine returns a handful of practice names with a line of justification each, and the patient then verifies those specific names on Google, on review sites and on the practice's own website before calling anyone. The AI answer does not close the patient. It decides which practices get considered at all.

The question is a shortlist request, not a booking request

It is tempting to picture AI search as a replacement for the whole decision — patient asks, engine answers, patient books. That is not what happens with a considered, expensive, semi-permanent purchase like veneers or implants.

What patients delegate to the engine is the tedious part: assembling a candidate list without reading twenty tabs. They are outsourcing the first filter, not the judgement. That distinction matters, because it tells you exactly what your website has to win.

The answer compresses your entire market to a handful of names

A search results page gives a patient ten blue links, plus a map pack, plus ads. It is a menu. A generative answer gives them a few names and a reason for each. It is a recommendation.

This is the structural change worth internalising. On a results page, being eighth still means being visible. In a generated answer there is no eighth position — there is named, or not named. A practice that would have been a respectable middle-of-page result simply does not appear in the conversation.

The competitive question stops being "where do I rank" and becomes "am I in the set".

What happens in the ten minutes after the answer

The AI answer is the start of the process, not the end. Watch what a patient does next and the pattern is consistent: they take the names they were given and check them. They search the practice name directly. They read reviews with a sharper eye, because they are now comparing three specific options rather than browsing. They open the website looking for confirmation — does this practice actually do this procedure, does the work look like what I want, is the price in the range I was told.

So the engine hands you a patient who arrives already warm, already comparing, and already carrying a claim about you that they intend to verify. If your site cannot substantiate the reason the engine gave for recommending you, you lose them at the last step — after doing the hard part.

Why this changes what your website is for

For twenty years the job of a practice website was to be found and to persuade. Those two jobs have now split apart, and they are served by different things.

Being found is increasingly decided before the patient ever reaches you — in the sources the engine reads, the structured data it parses, the corroboration it finds elsewhere. Persuading happens after, on the page, in the ten minutes described above.

Most practice sites are built entirely for the second job. They are polished, warm, image-led, and almost impossible for a machine to quote. That is why a beautiful website and an invisible practice are so often the same practice.

The practical implication

Your site now has two audiences with genuinely different needs, and it has to serve both without compromising either. The engine needs literal, specific, checkable statements: this procedure, this city, this price range, this recovery time. The patient needs reassurance, proof and a reason to trust you.

In practice this is less of a conflict than it sounds. Answer the question plainly at the top of the page, then persuade underneath. Patients who arrive from an AI answer are looking for that confirmation first anyway — the structure that gets you cited is also the structure that converts the patient the citation sent you.