In Beverly Hills, AI Names Four Dentists and One Practice

We asked ChatGPT, Claude and Gemini the same dental question 15 times in every incorporated city in Los Angeles County. One result has been bothering me since I first read it, because it quietly breaks the way almost everyone is measuring this.

Here is the top of the Beverly Hills page.

The Beverly Hills top five

Twelve of the fifteen runs came back with a real answer. Across those twelve, 35 different names got recommended. The ones that came up more than once:

1. Dr. Arthur Glosman, DDS, named in 8 of 12
2. Dr. Kourosh Maddahi, DDS, 6 of 12
3. Dr. Bill Dorfman, DDS, 5 of 12
4. Dr. Kevin Sands, DDS, 5 of 12
5. Bedford Dental Group, 4 of 12

Four people and one business. The AI was asked for a dentist and it answered with dentists, in the literal sense.

That sounds obvious until you look at what it does to measurement. Dr. Bill Dorfman's practice is called Century City Aesthetic Dentistry. If you were tracking AI visibility for Century City Aesthetic Dentistry, watching for that string to appear in answers, you would have recorded zero mentions out of twelve. The truthful number is five. Nothing about the practice changed. The string you chose to watch did.

Long Beach does the same thing. Temple City does not.

The obvious next question is whether Beverly Hills is just Beverly Hills. It is a small, celebrity-adjacent cosmetic market and you could talk yourself into any theory about it. So we read the other city pages.

Long Beach, 15 of 15 answered, 29 companies named:

1. David Younan, DDS (Long Beach Dental Wellness), 9 of 15
2. Dr. Stephen Coates, DDS, 9 of 15
3. Dr. John H. Cho, DDS, 7 of 15
4. Dentist of Bixby Knolls, 4 of 15
5. Dr. Will Grella, DDS, 4 of 15

Same shape. Four people, one business. Look at the first row: the page carries the practice in parentheses because the answers named the person and the practice is a separate string sitting behind him.

Temple City, 15 of 15 answered, 17 companies named:

1. Beautiful Dentistry, 15 of 15
2. Camellia Dentistry, 12 of 15
3. Temple City Dental Care, 11 of 15
4. Parklane Dental, 7 of 15
5. Lad Family Dentistry, 6 of 15

Not one person. Five business names.

Calabasas is the same, five business names in the top five, led by Calabasas Dental Institute at 13 of 15. West Hollywood is the same, led by three practices tied at 7 of 13. Santa Clarita is the same, led by Aesthetic Dental and Specialty Center at 6 of 15. Glendale has exactly one person in its top five, Dr. Lucio H. Kim, DDS at 6 of 9.

So in two of the seven LA markets in this piece, the majority of the leaderboard is people. In the other five, it is almost entirely businesses. Same question, same three engines, same fifteen runs, same week.

We do not have a clean explanation, and that is worth saying

I would like to tell you this tracks something tidy. It does not.

It is not city size. Long Beach has roughly 450,000 residents and Beverly Hills has about 32,000, and they behave identically here. It is not how crowded the market is. Temple City is the most concentrated market in the county we have looked at, one practice named in all 15 answers, and it is all business names. Glendale is one of the most fragmented, 49 names across 9 answers, and it is also almost all business names.

It is not even the source mix in the way you would expect. Santa Clarita's most cited sources are Healthgrades and WebMD Doctor, both of which are built around individual practitioners, and Santa Clarita's top five contains no practitioners at all.

What we can say is the narrow, checkable thing: the form of the answer changes by market, it changes by a lot, and you cannot predict which form your market returns without asking.

What this costs you if you get it wrong

Two ways, and they point in opposite directions, which is what makes this expensive.

You conclude you are invisible when you are not. You run a scan, or your agency runs one, it watches for your practice name, and it reports zero. In a person-name market that number is fiction. You then spend real money solving a problem you do not have, and worse, you stop trusting the measurement entirely the first time a patient says "ChatGPT sent me."

You conclude you are fine when you are not. The reverse case is quieter and costs more. The doctor gets named, the practice does not, and nothing on the practice website connects the two clearly enough for the next engine to make the link. The recommendation happens and it does not route to your front desk. Someone hears a name, searches it, lands on a directory profile with an old phone number, and books somewhere else. You never see that call, so you never count it.

Either way the actual patient volume is unaffected by what you believed. That is the part worth sitting with.

What to do this week

Check both strings. Whatever tool you use, run your practice name and every practitioner's full name as separate queries. If your tool cannot accept an alias list, that is a real limitation and you should ask the vendor about it before you trust another report from it.

Ask for the denominator. Any AI visibility number without a denominator is not a number. "Named in 5 of 12 answers" means something. "Low visibility" does not. Run the question enough times to have one, because these models give a different answer every time you ask.

Connect the person to the practice in public. If the engines are naming your doctor and not your business, the fix is not to fight it. It is to make the two inseparable everywhere the engines read: the site, the directory profiles, the review platforms, the schema on your own pages. You want the model that knows the name to also know the address and the phone number attached to it.

Every figure in this piece is on a public page you can open right now. Beverly Hills is at hollar.ai/ai-business-index/beverly-hills-dental and the rest of the county is linked from the LA dental index, 89 locations and 1,335 AI answers as of today. Go check your own city before you take my word for any of it.

If you want to know which form your market returns, and whether the engines are naming you, your practice, or neither, run a scan at hollar.ai/scan.