We asked AI for the best
med spa in Las Vegas
fifteen times.

What we did

Three assistants: ChatGPT, Gemini and Perplexity. Five questions, phrased the way a customer asks rather than the way a marketer does.

1. What's the best med spa in Las Vegas?
2. I'm looking for a med spa in Las Vegas for Botox. Who should I go to?
3. Recommend a medical spa near me in Las Vegas
4. Top rated med spas in Las Vegas
5. Who does good filler in Las Vegas?

Every question was asked in a fresh private window, signed out, one question per window, with no follow-up and no refining. Fifteen answers, recorded verbatim on 19 October 2026, including the sources each assistant cited.

Each assistant was used exactly as a signed out visitor gets it. No model was chosen by hand and no setting was changed. The questions were asked over a residential connection in Las Vegas, so each assistant inferred location from the IP address rather than from anything we set. Two things we did not record and therefore cannot claim to have controlled: which model version each assistant served that day, and whether any of them searched the web while answering. Anyone repeating this should expect both to differ, and the results with them.

Perplexity also returned a map style panel alongside its written answer. Three further businesses appeared only in that panel and never in a written recommendation, which takes the total number of distinct businesses surfaced to twenty seven. The figures below count written recommendations unless stated otherwise.

Finding one: there is no consensus

Fifteen answers produced twenty four different businesses. Only two of them, Advanced Aesthetics and The Look Med Spa, were named by all three assistants. Fifteen of the twenty four appeared on a single assistant and were absent from the other two entirely.

Six different businesses took first place across the fifteen answers.

That is a very different shape from a Google map pack, where the same three businesses tend to hold the top slots for weeks at a time. An AI answer is not a ranking that moves slowly. It is a different answer each time, drawn from a much wider pool.

Finding two: some businesses are invisible to half the market

This is the part that should concern anyone running a clinic.

Number of answers in which each business was named, out of five per assistant.

Each of these is prominent on one assistant and does not appear at all on another. Not lower down the list. Absent.

A clinic that checks its visibility by asking one assistant will conclude either that it is doing well or that it does not exist, and both conclusions will be wrong about two thirds of the market.

Finding three: Perplexity recommends a different set of businesses entirely

Perplexity returned two layers in most answers: a map style panel of nearby clinics, and a written recommendation. Six businesses appeared only in Perplexity's answers and were never named by ChatGPT or Gemini in any of the fifteen runs.

They share a pattern. They are small and very highly rated, with few reviews behind the rating.

Ratings and review counts as the assistant reported them.

A clinic with four reviews was included in an answer to "top rated med spas in Las Vegas". Another with two reviews appeared in a Botox recommendation. To Perplexity's credit, it was the only assistant that said out loud that a small review base is weak evidence.

Finding four: one answer invented a doctor

Asked about filler, Gemini named the surgeon at a well known Las Vegas cosmetic surgery practice as Dr Todd Richards.

The surgeon at that practice is Dr Bryson G. Richards. That is confirmed by three independent sources, and Gemini itself had given the correct name three questions earlier in the same session.

It is worth being precise about what this is. It is not two sources disagreeing. It is a confident, specific, wrong fact about a named medical professional, produced by the same assistant that had it right minutes before.

Two smaller faults appeared alongside it. One answer attached a thumbnail and caption belonging to a completely different business. Another broke off mid-sentence while listing a surgeon's name.

Every other named clinician we checked was real. Dr Mehdi Sina and Dr Harry H. Ching are both practising facial plastic surgeons in Las Vegas, and both were named correctly.

Finding five: the sources are not the ones anybody optimises for

Each assistant cited where it got its information. The pattern was consistent and surprising.

Sources as cited in the answers. Some answers cited a business's own website.

Seven of the eight are niche directories. Most of them are free to join and almost nobody in this market is paying attention to them. What to do about that is written up separately. ChatGPT cited Yelp on two questions and then cited three med spa directories and no Yelp at all on the next.

What a local business should actually do with this

1. Check all three, not one. Your visibility is not a single number. Ask each assistant the questions your customers would ask and record what comes back.
2. Look hard at the directories that are feeding the answers. This is the most promising thing on the list and also the least proven. We recorded which sources the assistants cited. We did not test whether adding a business to one of them changes whether it gets named, so treat this as a hypothesis worth testing rather than a lever known to work. Several of the ones above are free, which makes it cheap to find out.
3. Make the basics unambiguous. Consistent name, address, phone and service list everywhere they appear. When an assistant is assembling an answer from seven directories, inconsistency between them is how you end up described incorrectly or not at all.
4. Volume of reviews still matters, but differently. Two of the three assistants mostly named established practices with large review bases. The third surfaced tiny clinics with near perfect scores. Both patterns are in the data. Neither is explained by it.
5. Check what is being said about you. If an assistant can invent a doctor's name at a real practice, it can get your pricing, your services or your hours wrong too.

If you want the fifteen questions run against your own business rather than someone else's category, and the source list turned into something you can act on, that is what I do as an AI marketing consultant in Las Vegas.

What this study does not show

Fifteen answers is a snapshot, not a trend. We asked each question once rather than repeatedly, so the figures describe one day and not an average. Claude was not included. The ratings and review counts are as the assistants reported them, and where two assistants disagreed we have usually been unable to establish which platform each was quoting.

It shows no cause anywhere. We recorded which sources the assistants cited and which businesses they named. We did not add a business to a directory and watch what happened, so nothing here establishes that a listing produces a mention, or that any of the other patterns above are the reason a business did or did not appear. Every recommendation in the section above is an inference from what the answers looked like on one day.

And nothing here says anything about the quality of any business named. Appearing often in AI answers means a clinic is well represented in the sources these systems read. It is not a measure of clinical skill, and it should not be read as one.

The full list

Number of answers naming each business, out of five per assistant. Medicis Medical Spa, Inspire Med Spa and iGlow Aesthetics are not in this table because they appeared only in Perplexity's map panel, never in a written recommendation. They are listed in the Perplexity section above.

If you run a Las Vegas clinic and want to know where you sit in this, the questions and the method are above. It takes about twenty minutes to run for yourself.