The short answer
Most practices can add new patients faster by closing three leaks than by buying anything: calls that go unanswered, booked first visits that never show, and active patients who quietly stop coming. Those three cost nothing to fix and pay off within weeks. Once the pipe holds water, the highest-return spend in 2026 is a steady flow of recent Google reviews, a website that answers real patient questions in plain sentences, and being one of the few practices AI assistants name when someone asks who to see. Paid ads work, but they are the most expensive way to buy a patient, so they belong last, not first.
Fix the bucket before you turn up the tap
When an owner asks how to get more patients, the question usually means "where do I buy more leads." That is almost never the best first move. Three leaks are draining most practices right now, and every one of them is free to close: the phone that rings out, the first visit that never happens, and the patient who stops coming without anyone noticing. Together they typically cost a practice more new-patient value in a month than any campaign of the same month would add.
There is a hard financial reason to go in this order. Marketing pours people into the same pipe you already have. If a quarter of them fall out before the first adjustment, you paid full price for all of them and kept three out of four. Close the holes first and the exact same ad budget buys you more patients, permanently.
Leak one: the calls nobody picks up
New patients call. They call while you are adjusting, at lunch, at 6:40pm, on a Saturday morning when something in the lower back lets go. A person in real pain who reaches voicemail does not leave a message and wait for a callback. They hang up and dial the next clinic. A missed call from a new patient is not a delayed booking. It is a booking that went to a competitor, and you will never see it in any report.
- Pull last month's call log from your phone provider. Count every inbound call that lasted under fifteen seconds — that is your voicemail pile.
- Count how many arrived outside office hours. In most practices this is a third of the total.
- Call your own front desk from a number nobody recognizes, at 12:15pm on a Tuesday, and time how long it rings.
- Check whether missed calls get a text back within a minute. If they do not, the caller has already booked elsewhere.
This is the single highest-return thing most practices can change, because the demand already exists — it is arriving and bouncing. Whether you solve it with a second front-desk person, an answering service, or an AI receptionist that picks up every call and books into your calendar around the clock, solve it before you spend a dollar on getting the phone to ring more.
Leak two: the first visit that never happened
Someone booked. They meant it at the time. Then five days passed, the back loosened up a little, the workday got busy, and nobody heard from your office in between. No-shows for a first appointment are rarely about price or doubt. They are about time and silence.
- Book new patients within 48 hours whenever you can. The further out the slot, the lower the show rate.
- Send a confirmation text the moment the appointment is made, not the night before.
- Send a short reminder the day before and again two hours out, with a one-tap way to move the appointment instead of ghosting.
- Tell them exactly what the first visit involves and how long it takes. Uncertainty is a bigger cause of no-shows than cost.
- When someone does no-show, call the same day. A same-day call rebooks a meaningful share of them; a week later almost none.
Track your show rate as a number every month. Practices that measure it fix it, and practices that do not measure it assume it is fine. It usually is not.
Leak three: the patients who quietly stopped
Open your practice software and list everyone who was seen three or more times and has not been in for six months. In most offices that list is longer than the entire new-patient count for the year. Those people already know you, already trust you, already know where to park. Nobody has spoken to them since they stopped.
They did not leave because they were unhappy. They left because a symptom eased, or an insurance year rolled over, or life happened. A short, human message — no discount, no gimmick, just a check-in on how the neck has been holding up — brings a real share of them back. It is the cheapest new-patient equivalent in the entire practice, and it is sitting in a database you already own.
Where new patients actually come from now
Once the leaks are closed, the question becomes where the next person hears about you. The honest answer in 2026 is that it has split. Word of mouth is still the strongest single source in most practices. Google Maps and Google reviews still do enormous work. And a growing share of people now open ChatGPT and ask it who to see, the same way they used to open a search box — about 45% of patients now ask ChatGPT or a similar AI assistant for recommendations.
That last part matters because the answer format is different. A search results page gives someone twenty options and lets them browse. An AI assistant gives them three names in a paragraph. There is no page two. Either you are one of the names or you do not exist in that conversation.
What that looks like in the toughest markets
We stopped speculating about this and measured it. In each metro we asked ChatGPT the same four patient-style questions and wrote down every practice it named, then checked those names against the full list of real practices in that market. Below are the six metros where it named the smallest share.
Our data
The six hardest metros to get named in
Across 27 metros in Florida and Texas we checked 3,214 chiropractic practices and ChatGPT named 209 of them — 6.5% overall. These six markets are the bottom of the range:
| Metro | Practices we checked | Named by ChatGPT | Share named |
|---|---|---|---|
| Jacksonville, FL | 254 | 2 | 0.8% |
| Sarasota, FL | 173 | 2 | 1.2% |
| Orlando, FL | 305 | 4 | 1.3% |
| Tampa, FL | 515 | 11 | 2.1% |
| Miami, FL | 229 | 5 | 2.2% |
| Port St. Lucie, FL | 86 | 2 | 2.3% |
Jacksonville is the clearest picture of the problem: 254 practices, two names. Even in the hardest market in the study, a handful of practices do get named — so this is not a wall, it is a short list with room on it. Everyone else in those cities is invisible to a patient who asks an AI assistant instead of searching.
Source: the Grow Your Chiropractic Practice AI recommendation benchmark, run 2026-07-03. We asked ChatGPT the same four patient-style questions in each of 27 metros and recorded every practice it named by name. 3,214 practices checked, 209 named.
Read it as an opportunity rather than a verdict. Two practices in Jacksonville are being handed patients from a channel their 252 competitors have not noticed yet. The gap between being named and not being named is not clinical skill. It is whether the web describes your practice clearly enough for an assistant to recommend it without hedging.
The ranked list, cheapest first
This is the order we would work through with any practice, and roughly what each step costs in money, work and waiting. The timings are what we typically see, not promises.
| Move | Money it costs | Work it takes | When patients show up |
|---|---|---|---|
| Answer every call, including nights and weekends | Low to moderate | One-time setup | Same week |
| Text back every missed call within 60 seconds | Low | One-time setup | Same week |
| Confirm and remind every booked first visit | Almost none | One-time setup | 1-2 weeks |
| Message patients who stopped coming | Almost none | A few hours | 1-3 weeks |
| Ask every satisfied patient for a Google review, every visit | None | Ongoing habit | 3-8 weeks |
| Put real answers to patient questions on your website | Low | Steady writing | 6-12 weeks |
| Get named by ChatGPT when people ask locally | Moderate | Mostly done for you | 8-16 weeks |
| Build referral relationships with local gyms, trainers and GPs | Low | Ongoing in-person time | 2-4 months |
| Run paid ads to a first-visit offer | High and recurring | Ongoing management | Days, while you pay |
Notice that paid advertising is last and it is the only line that stops producing the moment you stop paying. That is not an argument against it. It is an argument for putting it after the things that keep working when the invoice ends.
The referral engine most practices never actually build
Almost every owner says referrals are their best source, and almost none of them do anything deliberate about it. Referrals are treated as weather rather than a system. They can be a system.
- Ask at the right moment — the visit where the patient tells you they feel better, not the visit where they pay.
- Make it concrete. "Do you know anyone else dealing with the same thing?" beats "tell your friends about us."
- Build two or three real relationships with people who see back pain all day: personal trainers, running-club organizers, physical therapists, massage therapists, a couple of family doctors.
- Give before you ask. Send them a patient first, and be the person who answers their text.
- Thank every referral by name, fast. People refer again when they know it landed.
What we would not spend money on yet
- A logo refresh or a redesign that only changes how the site looks. If the current site loads fast and answers questions, patients do not care about the typeface.
- Deep discounting your first visit. It fills the schedule with people shopping on price and trains your market to wait for the next offer.
- Printed mailers to a whole zip code. Expensive per response, and impossible to tell what worked.
- A social media posting service that publishes daily filler nobody reads. Posting volume is not distribution.
- Any package sold with a promise about a specific position in search results. Nobody can guarantee that, and the ones who claim it are guessing.
What to do in the next 30 days
- Week one: get your missed-call number. Fix the after-hours gap so nothing rings out.
- Week one: turn on confirmations and reminders for every new-patient appointment.
- Week two: pull the list of patients not seen in six months and message them personally.
- Week two: start asking for a Google review at the visit where the patient says they feel better.
- Week three: write plain answers to your ten most-asked patient questions and put them on your site as text.
- Week four: measure. New patients, show rate, review count, calls answered. Compare to the month before, then repeat.
If you would rather see the whole picture before you start, our free audit checks your practice on all of it — how your calls are handled, how your reviews look, and whether AI assistants can find and describe you at all.
Common questions
How many new patients should I expect from fixing missed calls alone?
It depends entirely on how many calls you are currently missing, which is why the first step is pulling the log rather than guessing. Practices that were sending a third of their inbound calls to voicemail typically see the largest single jump of anything on this list, because the demand already existed and was simply bouncing off a closed door.
Is it worth running ads if my schedule is already half full?
Usually not yet. A half-full schedule with unanswered calls, no-shows and a stale patient list means the practice is losing people it already reached. Ads add cost on top of that problem instead of solving it. Once the show rate is healthy and every call is answered, ads become a reasonable way to buy volume on demand.
How many Google reviews do I need?
There is no threshold number that flips a switch. What matters far more is recency and steadiness — a practice adding a few reviews every month reads as busy and current, while a practice with a large pile of reviews that all stopped two years ago reads as coasting or closed. Aim for a habit, not a target.
Does being named by ChatGPT actually bring patients through the door?
It brings the specific patient who asked an assistant instead of searching, and that group is growing. In our benchmark of 3,214 practices across 27 metros, only 209 were named at all, so the practices that get in are sharing a very short list rather than competing with the whole market.
What does it cost to get set up properly?
Our AI Website is $297 a month with everything included — the practice website plus Claire, the AI receptionist who answers calls 24/7, books appointments, handles website chat and collects Google reviews. Getting named by AI assistants is a separate service at $97 a month while we do the work, moving to $495 a month only once ChatGPT is actually naming your practice.
I have tried marketing before and it did not work. Why would this be different?
Most marketing that fails was pointed at a practice that could not receive the patients it produced. Calls went to voicemail, first visits were booked ten days out, and nobody followed up. Working the list in the order above means the traffic you eventually pay for lands somewhere that converts it.
How long before I can tell if any of this is working?
The call and no-show fixes show up in the same month, because they change what happens to demand that is already arriving. Reviews and website content usually take one to three months to move anything. AI visibility is the slowest of the group and should be judged over a quarter, not over two weeks.
See exactly where your practice is leaking patients
The free audit checks your calls, your reviews, your website and whether AI assistants can find you — then tells you the one thing to fix first.
Start my free audit →Going deeper on this topic: More on building a steady new-patient flow
Read next