A woman in Tulsa wakes up at 7am, rolls over, and her lower back locks. She doesn't call a friend. She doesn't open a map. She picks up her phone, half-awake, and types something like "woke up with sharp lower back pain can barely stand what should I do." What comes back isn't ten blue links. It's a calm paragraph about common causes, a note that she should get it evaluated, and then — this is the part that matters — the names of a few clinics near her that could see her.
One of those names gets a new patient this week. The others get nothing. They don't know the conversation happened.
The question people actually ask
The old assumption was that someone needing a chiropractor types "chiropractor near me." Some still do. But the more common entry point now is a symptom, described in a person's own words, at the moment it's hurting. "Neck stiff after sleeping wrong." "Shooting pain down my leg when I sit." "Been hunched at a desk and my upper back is killing me." These are not category searches. They're pain, narrated.
An assistant handles that narration in two moves. First it addresses the symptom in general, careful terms, usually with some version of "this could be a few things, and you should have it looked at." Then it does the thing a search engine never did so smoothly: it hands the person a specific place to go. A named clinic. Sometimes with a line about what that clinic is known for.
That handoff is the whole game. The person was in pain and uncertain, and the assistant resolved the uncertainty by naming you. Or by naming the practice three blocks from you. The symptom-first query is how someone who never searched for "chiropractor" at all still ends up walking into a chiropractor's office — just not necessarily yours.
Why "short list" changes everything
For twenty years the fear in local marketing was ranking on page two. Page two was bad, but page two still existed. A determined person scrolled. Your clinic was findable even if it wasn't first.
An assistant answer has no page two. When someone asks about their back at 7am, the response is a short, named list — a handful of clinics, often with a reason attached to each. Nothing resembling a scrollable page of ten results. If your clinic isn't in that handful, you weren't ranked lower. You were absent from the only list the person will ever see. There's no second screen for them to reach. The answer felt complete, so they acted on it.
That's the shift that should make any clinic owner sit up. Being left out of an AI answer isn't a worse position than before. It's a near-total miss on that patient, in that moment, when they were most ready to book.
Why your name shows up — or doesn't
Here's the part worth slowing down on, because it explains almost everything about who gets named.
AI models don't take a business's word for itself. If your website says you're the best chiropractor in town and open Saturdays, that's a claim — and a claim made in one place, by the party who benefits from it, is weak evidence to a model. What these systems actually trust is agreement. They look for the same facts about you showing up, the same way, across independent sources that have no reason to coordinate. When your hours, your location, your focus, and the basic fact that you exist and take new patients all line up across many places a model can check, that consistency reads as true. Confidence climbs. Your name becomes safe to say out loud to a person in pain.
When those facts don't line up — when one source has an old address, another a dead phone number, a third lists a service you dropped years ago — the model notices the contradiction. And a model that spots contradictions hedges. It reaches instead for the clinic it can describe cleanly, without tripping over conflicting information. Hedging models recommend someone else. That's the quiet mechanism behind a lot of good clinics never getting named: not that they're bad, but that the story about them is inconsistent, and inconsistency makes a model nervous.
Think about what that means for a practice that's been great for fifteen years. All that goodwill lives in your patients' memories and in word of mouth. A model can't read a memory. It can only read confirmable text, and if that text is thin, stale, or self-contradictory, your reputation might as well not exist as far as the 7am question is concerned.
Take Healthgrades, for instance
To make this concrete, look at Healthgrades. Most clinic owners think of it, if they think of it at all, as "that site with the star ratings." That framing badly undersells what it does inside an AI answer.
Healthgrades is a place where facts about a provider are stated by a third party and structured in a way machines read easily — who you are, where you practice, what you focus on, what patients said after visits. It's independent of your own website. So when a model is trying to decide whether to trust what your site claims, a profile like this becomes a corroborating witness. The model sees your own site say one thing, then sees an outside source say the same thing, and the agreement between them is what turns a claim into something the model will repeat with confidence.
The flip side is where it gets dangerous. A Healthgrades profile that's half-finished, lists a provider who left your practice, or shows an address that doesn't match your site becomes a contradicting witness. Now the model has two sources disagreeing about a basic fact, and it does the cautious thing — it hedges, or it skips you for a clinic whose story holds together. The profile that could have been your strongest corroboration quietly becomes the reason you got passed over.
That's why "I have a Healthgrades page" and "my Healthgrades page is pulling its weight" are completely different statements. One is a box checked. The other is a witness who backs up your story instead of undermining it. Worth understanding deeply, because it's a clean picture of the whole trust-through-agreement idea playing out in a single place.
This is measurable, not vibes
None of this is theoretical hand-waving. RankCommander maintains an AI Visibility Index built on tens of thousands of real AI answers evaluated across the seven assistants people actually use — ChatGPT, Claude, Gemini, Perplexity, Google AI Overviews, Grok, and Copilot — and still counting. The pattern holds across verticals: in any given local market, a handful of providers get named again and again while the rest, including plenty of excellent ones, almost never surface. You can see the benchmark data at /ai-visibility-index.
What the Index makes undeniable is that this isn't random and it isn't a fluke of one platform. The same clinic that gets named by Gemini tends to get named by Perplexity, and the same clinic that's invisible on one is usually invisible across the board. The person in pain gets a consistent experience of you existing — or not existing — no matter which assistant they happened to open.
The competitor who's already there
Picture the practice across town. Maybe newer than you. Maybe not as good, if you're honest about the clinical side. But their story is clean and consistent everywhere a model looks, so when the woman in Tulsa asks about her back at 7am, their name is the one that comes back. They're getting the patient you spent fifteen years earning the right to.
That's the loss that stings, and it's happening now, not someday. Every morning someone wakes up in pain within a few miles of your clinic and asks an assistant what to do. Every one of those is a real person who could have been your patient — a person who would've become a regular, told their family, filled a slot on your schedule for years. When the assistant names a competitor instead, that person is simply gone. You never see the appointment that didn't get booked. You just feel the schedule staying a little lighter than it should, and you blame the economy, or the season, or anything other than the fact that the question moved to a place where your name doesn't come up.
The clinics winning this aren't the ones with the flashiest sites. What separates them is that the story a machine can verify about them actually holds together, so a model feels safe saying their name to someone who's hurting. Where you start closing that gap depends entirely on where your own gaps are — and that's a different picture for a solo practice than for a three-location group, which is exactly why a spot-check in your own browser only tells you so much.
See what the assistants say about you
You've built something real. Don't let it go quiet in the one place patients now go first. RankCommander scans how all seven assistants answer the symptom-first questions your future patients are asking right now — which clinics get named in your market, how consistent your story looks to a model, and where a competitor is being handed the patient who should have been yours. It's the difference between guessing and knowing, and between losing ground you can't see and defending it while there's still time. Run your scan at /medical and find out whose name comes up at 7am. Make sure it's yours.