180 AI answers across 10 US cities show what ChatGPT and Google tell patients looking for a chiropractor.
We asked ChatGPT and Google’s AI 180 questions about finding a chiropractor in 10 US cities. When patients asked for a chiropractor, both named some. When they described the problem instead, the two gave very different answers.
The Chiro Site Club · November 9, 2026 · Chris Serra
- When patients ask for a chiropractor, AI names one. “Who is a good chiropractor?” and “Best chiropractor for lower back pain?” got chiropractors named every time, 60 of 60 answers on ChatGPT and 60 of 60 on Google.
- Google still shows chiropractors when patients describe the pain. Asked “I woke up and can barely turn my neck. Who should I see?”, Google’s AI named at least one chiropractor in 16 of 30 answers and never warned against chiropractic.
- The gap is ChatGPT on symptoms. On that same neck question, ChatGPT named a chiropractor in 0 of 30 answers, and in 29 of 30 said to see someone else first, mostly orthopedic and spine care or urgent care.
- Being named once means little. Ask the same question three times and you often get a different list.
Why we ran it
A patient can now ask an AI tool who to see instead of scrolling a page of search results. We wanted to see what they actually get back when they ask about a chiropractor, in a way anyone can repeat.
What we did
Cities (10): New York, Los Angeles, Chicago, Houston, Atlanta, Sacramento, Boise, Omaha, Tulsa, Knoxville.
Questions (3), written the way a patient types them:
- “Who is a good chiropractor in [city]?”
- “Best chiropractor for lower back pain in [city]?”
- “I woke up and can barely turn my neck. Who should I see in [city]?”
AI tools (2): ChatGPT and Google’s AI Mode, both used logged out, in a fresh chat for every question.
Repeats: every question asked 3 separate times. 10 cities x 3 questions x 3 asks x 2 tools = 180 answers.
When: November 9, 2026, from Sacramento, California.
How we counted: an answer “named a chiropractor” when it recommended a specific chiropractor or chiropractic practice by name. A general line like “you could also see a chiropractor” did not count. Every answer to the neck question was read and scored by hand, by a reviewer who had not seen our early counts.
We planned to include Perplexity too. It asked us to sign up after two questions, so we left it out rather than mix logged-in and logged-out results.
Ask for a chiropractor, get one
When the patient asked for a chiropractor by name, “Who is a good chiropractor in [city]?” or “Best chiropractor for lower back pain in [city]?”, both tools named chiropractors in every answer. That was 60 of 60 on ChatGPT and 60 of 60 on Google, in big cities and mid-size cities alike.
Describe the pain, and ChatGPT steers you away
When the patient described a stiff neck instead of asking for a chiropractor, the picture split.
Named a chiropractor
Warned against chiropractic, or said see someone else first
| ChatGPT | Google AI Mode | |
|---|---|---|
| Named a chiropractor | 0 | 16 |
| Warned against chiropractic or neck manipulation, or said to see someone else first | 29 | 0 |
| Where it pointed people | Orthopedic or spine care (16), urgent care (12), a mix (2) | A mix of urgent care, physical therapy and chiropractic (27), urgent care (3) |
Of ChatGPT’s 29, seven were softer, suggesting another provider first rather than warning against chiropractic outright.
Typical ChatGPT wording ran from “avoid chiropractic neck manipulation until you’ve been evaluated” to “rather than immediately booking a chiropractor.” Google more often listed urgent care, physical therapy and chiropractic side by side and let the patient choose.
We are not judging the medical advice. A sudden, severe stiff neck can have causes that need a medical exam first, and the tools are being careful. The point for practices is simpler: a patient who describes their problem in plain words, instead of typing “chiropractor”, often never hears a chiropractor’s name.
Ask twice, get a different list
We asked every question three separate times. On a typical pair of asks, the practices named overlapped only partly, roughly a third of the time on ChatGPT and a little under half on Google (rough, name matching was automated). In 21 of 30 city-and-question sets on ChatGPT, and 26 of 30 on Google, at least one practice did show up all three times.
One screenshot of an AI answer, good or bad, says little. What matters is how often a practice comes up across many asks.
The two tools rarely agree
For the same city and question, only about a third of the practices ChatGPT named also appeared anywhere in Google’s answers. Google named about 6 practices per answer, ChatGPT about 3 to 4. Big cities and mid-size cities looked about the same.
They lean on different sources
Google’s answers were built around business listings, with ratings and links to each practice’s own website. ChatGPT more often pointed to hospital systems, doctor directories and review or “best of” sites.
One more thing we noticed
Used logged out, ChatGPT showed an ad under 37 of 90 answers: mattresses, desks, supplements, a massage studio, a law firm. The ads often had nothing to do with the city asked about. None appeared on the neck question.
What this means for a practice
- ChatGPT steered away. Google still showed a chiropractor. On the neck question, Google’s AI named at least one chiropractor in 16 of 30 answers and never warned against chiropractic. Its answers were built around business listings, with star ratings and links to practices’ own websites. This study cannot say what earns a place in those answers. It does show where Google was looking, and a practice’s listing, ratings and website are all things it can keep accurate and complete.
- Answer the plain-words question on your own site. Patients describe symptoms, not specialties, and in this study that is where chiropractors dropped out of ChatGPT’s answers. ChatGPT was careful about red flags, and a careful patient is too. Say plainly what you treat, who it suits and when someone should see a physician first. To a nervous patient reading at 9pm, that is what a responsible practice looks like, whatever any AI does with it.
- Track it over months, not one screenshot. Ask the same question twice and you often get a different list. One answer is weather. The number that means something is how often a practice comes up across many asks, on more than one tool, over time.
We cannot promise any tool will name a practice. No one can.
Run it yourself
Open ChatGPT and Google logged out, ask the three questions above for your city three times each, and write down who gets named.
If you’d like someone to look at your own site and how clearly it explains what you treat, the free check is here.
Limits
- One day, two tools, 10 cities, logged out, from one location. AI answers change often and can differ by location, account and time.
- This shows what came back on that day. It does not show why, and it does not show that any website change causes a practice to be named.
- No practice is named in this study. Results are counts only.
Raw answers and the full method are kept on file and available on request.