Home/Physical Therapy Marketing/Physical Therapy SEO
The guide
Physical therapy SEO is the work of being the clinic that appears when someone nearby searches for physical therapy without a clinic's name in mind. It is decided in three places — the Google map pack, the organic results under it, and increasingly the AI answer above both — and it is won with the same work in all three. This guide covers that work, then gives a separate playbook for a solo practitioner, a small-to-mid-sized practice, and a specialized practice, because the right plan is different for each.
Find My Playbook↓Which practice are you?
Pick one. The page scrolls to your playbook and highlights your row everywhere else — the AI section, the summary table, the asides. Nothing is hidden; your row is just visibly yours.
Part 1 of 12
Key takeaways
Let's start with what it is not.
If a physician hands a patient a slip with your clinic's name on it and that patient types your name into Google, you will be the first result. You did not need an agency for that. You needed a Google Business Profile, which is free, and some reviews, which are also free. Anyone charging you a retainer to rank for your own name is charging you for something that was going to happen anyway.
Physical therapy SEO is about the searches that do not have your name in them. "Physical therapy near me." "Pelvic floor physical therapy [city]." "Does insurance cover physical therapy." "Do I need a referral for PT." Those searches are made by people who have decided to get physical therapy and have not decided where. Whoever appears gets the call.
Three things decide who appears. The map pack — the three listings with the map that sit at the top of most local searches. The organic results — the blue links below it. And, since 2024, the AI answer — Google's AI Overview above the results, and the answers people now get from ChatGPT and its peers before they open Google at all. They are different systems. They reward the same work. Part 6 shows why, in Google's own words.
This guide is about local search and the bottom of the funnel: the patient who is ready to book, within driving distance, this week. It is not a guide to blogging. If you have been told that physical therapy SEO means publishing four articles a month about back stretches, this page will disagree with that, with numbers.
It is written for three kinds of practice, because one playbook does not fit them. A solo practitioner has no admin time and needs the map pack in a two-mile radius. A practice with eight clinicians and a front desk can absorb twenty new evaluations a month and needs a system. A pelvic health or cash-based practice has patients who will drive across the metro and read everything first. Parts 2 through 6 are the foundations every one of them does. Part 7 is the playbook for yours.
Part 2 of 12
Key takeaways
Roughly 90% of physical therapy patients say a physician referred them (APTA, 2025). That number gets used to argue that PT practices should market to doctors, not to Google. Read it more carefully and it argues the opposite.
A referral is an order to get physical therapy. It is not an assignment to a particular clinic, and patients frequently do not know they had a choice. Then the referral leaks. Researchers at University of Utah Health followed 15,877 physical therapy referrals and found that 24% accessed PT inside the system that referred them. A 2024 orthopaedic cohort found 40% never attended a single visit. A urogynaecology study found 45–47% initiated care.
Those patients do not evaporate. A share of them turn up weeks later, still in pain, typing "physical therapy near me" with nobody's name in their head. The referral created the demand. The search decides who collects it.
And a growing number never had a referral. Every US state now permits some form of direct access to physical therapy without a physician's order (APTA, June 2025). Only 41% of the public knows that. The ones who do are searching.
Here is the US search market for physical therapy, without a clinic name in it. Real monthly volume, not an estimate.
| Search | Monthly US volume | Difficulty (0–100) |
|---|---|---|
| physical therapy | 155,000 | 62 |
| physical therapy near me | 137,000 | 40 |
| physical therapist near me | 19,000 | 33 |
| dry needling near me | 6,700 | 0 |
| pt near me | 6,500 | 33 |
| physical therapy clinic near me | 3,400 | 27 |
| physical therapy for sciatica | 1,900 | 12 |
| how much does physical therapy cost | 1,800 | 3 |
| pelvic floor physical therapy near me | 1,400 | 35 |
| physical therapy for knee pain | 1,300 | 5 |
| does insurance cover physical therapy | 900 | 0 |
| sports physical therapy near me | 900 | 30 |
| in home physical therapy near me | 800 | 39 |
| physical therapy after knee replacement | 700 | 4 |
| vestibular therapy near me | 600 | 0 |
| do i need a referral for physical therapy | 500 | 0 |
| best physical therapist near me | 500 | 6 |
| cash based physical therapy | 400 | 0 |
Ahrefs Keywords Explorer, United States, August 2026. National figures; your town is a slice, but the composition of the slice is the same.
"Physical therapy near me" has a difficulty of 40. You are competing with hospital systems and national chains for it, and for most practices that is a multi-year fight that is not worth starting. Now look at the rows in bold. Difficulty zero does not mean easy money. It means nobody has bothered to compete — the clinics who could answer those searches have not written the page.
We audited 100 physical therapy clinics across 12 US metros in September 2026 — the ones Google Maps actually shows for "physical therapy [city]." Only 29% name a single insurance carrier anywhere on their site. Eight of 94 show a price anywhere. Forty-seven percent never tell patients they can come without a referral. Just three of the hundred have all three pages.
Put that next to the table above. "How much does physical therapy cost" is searched 1,800 times a month at difficulty 3. Eighty-five percent of ranking clinics have no cost page. "Does insurance cover physical therapy": 900 a month, difficulty 0, and nearly half of ranking clinics have no insurance page at all. That is demand with no supply. The opening in physical therapy SEO is not more content. It is answering three questions nobody is answering.
Higher-intent patients. Someone who searched "vestibular therapy near me" has self-diagnosed the need and is choosing a clinic. They convert differently from anyone an ad interrupted.
It compounds. A page that ranks keeps producing evaluations without further spend. Ads produce in week one and stop the day the budget stops. That is not an argument against ads — Part 7 says when to use them — but it is the reason a ranking page is an asset and an ad is a rental.
A patient is an episode, not a visit. Published claims research puts a typical outpatient PT episode at six to ten visits: 7.0 visits for physician-referred and 5.9 for self-referred episodes across 62,707 commercial episodes; 6.8 across 1,840 Medicare episodes; 10.2 in the FOTO registry. Musculoskeletal problems recur. One booked evaluation is worth an episode, and sometimes a second one four years later. Almost no clinic values a patient that way when deciding what to spend on acquisition.
It fills whatever gap you have. A new clinician, a new specialty, Saturday hours — search demand can be pointed at any of them.
It is defensible. The map pack has three slots. Once you hold one in your area, a competitor has to displace you.
It gets you into AI answers. Google's own documentation says there are no additional requirements to appear in AI Overviews and that the best practices for SEO remain relevant. The same work, a second surface. Part 6.
Search is the most expensive of the three ways to grow a practice. There are only three: treat more patients, make each episode worth more, or bring patients back more often. SEO addresses the first, and the first is the hardest.
So if you have two thousand names on a dormant patient list and nobody has contacted them in a year, that list will produce booked visits faster and more cheaply than any amount of search work. Do that first. → How to reactivate a dormant patient list
SEO also cannot fix a five-week wait, a phone that goes to voicemail at lunch, or a clinic that is already full. Physical therapy patients waited a median of 25 days from referral to evaluation in a 2024 JOSPT Open analysis, and patients booked more than 30 days out no-show at roughly twice the rate of those booked inside 30 days. In a 2022 survey of 1,005 patients, 61% had skipped care in the previous year because scheduling was too difficult, and 41% had switched providers over a poor digital experience. If someone searches, finds you, and does not become a patient, SEO was not the problem and buying more of it will not help. → Stop leaking patients at the phone
It takes months, not weeks. Anyone promising otherwise is describing advertising.
Part 3 of 12
Key takeaways
"Bottom of the funnel" means one thing here: the patient has decided to get physical therapy and is choosing where. Those are the searches this guide is built around. There are seven families of them.
1. Generic near-me. Physical therapy near me, physical therapist near me, PT near me, physical therapy clinic near me. The biggest pool, the hardest to win, and decided almost entirely by the map pack.
2. City or neighborhood plus service. Physical therapy [city], physical therapist in [neighborhood]. The same intent, typed by someone who does not trust "near me."
3. Specialty plus near-me or city. Pelvic floor, vestibular, sports, pediatric, dry needling, hand therapy, aquatic, post-surgical. Smaller pools, far less competition, and the searcher has already narrowed to what you do.
4. Condition plus near-me or city. Sciatica, knee pain, rotator cuff, ACL rehab, vertigo, low back pain. The patient names the problem, not the service. These are the searches condition pages exist for.
5. Modifiers. Best, top-rated, same-day, open Saturday, no referral, in-home, cash pay. Each one is a patient telling you what they care about.
6. Payer and access terms. Physical therapy that takes [insurer]. Accepts Medicare. Does insurance cover physical therapy. Do I need a referral for physical therapy. These read as questions and behave as bookings — the person asking them is one answer away from calling. They are informational in form and commercial in intent, and they are the most under-served searches in the category.
7. Competitor brand searches. People searching another clinic's name. Know they exist. Do not chase them; you will look like what you are doing.
List every service you actually deliver. List every condition you treat and want more of. List every payer you accept, by the name a patient uses. List your geography — your city, your neighborhoods, and the towns inside your real drive radius, not the ones you would like to serve. Combine them: service × place, condition × place, specialty × place, payer × service.
Then open Search Console and your Business Profile's performance report and look at what you already appear for. You will find searches you never planned for. Some of them are your best opportunities.
Score what is left by three things: intent (is this person choosing a clinic, or reading?), volume (is anyone searching it?), and honesty (can you treat it, and can you convert it?). A pelvic health term you cannot staff is not an opportunity. It is a booking you will have to turn away.
Map each primary search to exactly one page. Two pages chasing the same term do not double your chances; they split your authority and Google picks neither. This is the single most common structural problem on physical therapy websites, and Part 5 is about fixing it.
A physical therapy site has to serve three different states of mind, and they cannot live on the same page.
Commercial — "I need a PT, which one." Service pages and location pages.
Problem — "What is this pain and what do I do about it." Condition pages and resource pages.
Question — "Does insurance cover this, do I need a referral, how much is it." Bridge pages.
Each intent gets its own page type. Part 5 builds them. The reason to name them now is that the question intent is where the audit found the gap: 1,800 searches a month for the cost of physical therapy, and 85% of ranking clinics with no page that answers it.
Part 4 of 12
Key takeaways
In local search, the Business Profile is the listing. The website supports it. Get the profile wrong and nothing downstream saves you.
Own it. The profile lives in the practice owner's Google account. Not the agency's, not the web designer's, not the front-desk manager who left. This is the most common way a clinic discovers the real cost of leaving a vendor, and it is avoidable in one afternoon.
Primary category. In Whitespark's Local Search Ranking Factors 2026, the primary category ranked as the single most influential map-pack factor of the 187 studied — and an incorrect primary category ranked as the second most damaging negative factor. That is a survey of 47 practitioners scoring by judgment, not a controlled measurement, and it is also the most consistently corroborated dataset the field has. For most outpatient clinics the answer is "Physical Therapist" or "Physical therapy clinic." Where it gets expensive is guessing. A sports clinic that lists "Sports Medicine Clinic" as primary may be trading every "physical therapy near me" search in its postcode for a much smaller pool. Secondary categories are where the specialties go.
The name is the name on the sign. Adding "Physical Therapy | Sports Rehab | Dry Needling" to your business name violates Google's guidelines and is one of the reliable ways to get a listing suspended. It also makes you look like what you are doing.
Phone and address. A local number, identical everywhere it appears. In our September 2026 audit, 11% of ranking clinics showed a different phone on their website from the one on their Google listing — usually a corporate or routing number in the site header with the clinic's actual line buried in a footer list. Google reads that inconsistency as uncertainty. Address for a clinic; service area for in-home and mobile practices, with no address shown (7.3 covers this).
Hours, including holiday hours. A listing that says you are open on a day you are closed produces a one-star review from someone standing at a locked door.
Services. The services list is where your keyword list from Part 3 goes — each service you actually deliver, named the way a patient names it. The description is not a ranking field. It is read by a human deciding whether to call you. Write it for them.
Photos. Real ones: the entrance, the treatment floor, the team, the parking. Patients use them for wayfinding and for deciding whether the place looks like somewhere they want to be. Do not geotag them. Joy Hawkins at Sterling Sky tested it and found Google strips the metadata on upload with no measurable ranking change; a 27-location controlled study published in Search Engine Land in March 2025 reached the same conclusion on six of seven metrics; Whitespark's 2026 survey ranks geotagged photos the single lowest-impact factor of the 187 it scores.
Q&A and attributes. Seed your own Q&A with the questions the front desk hears — parking, referrals, insurance, first visit. Set the attributes: accessibility, appointment required, what payment you take.
Booking link. If you have online scheduling, put the link on the profile. Tag it so you can see in analytics what the profile produces.
Posts are worth little for ranking and something for looking alive. Once a month is enough.
Practitioner listings. Google allows individual clinicians to have their own profiles. Left unmanaged, a clinic can end up with four listings at one address competing with each other and splitting reviews. Decide which listing is the one you want to rank and keep the others accurate but secondary.
Report the spam. Competitors with keywords stuffed into their business name are violating the guidelines you are following. Google acts on reports. It takes two minutes.
One more reason to get all of this right: Google names accurate Business Profile data as one of its explicit recommendations for appearing in its AI features. The profile is no longer only a map-pack task.
How the map pack actually works, pillar by pillar→Reviews do two jobs. They move map-pack rank, and they decide who gets chosen once the searcher is looking at three listings. For healthcare, the second job is the bigger one. NRC Health's 2023 study of 155,603 healthcare consumers found 60.5% had looked at online ratings for a doctor, and 85% of the reviews they saw were on Google. A randomised experiment with 949 participants found that moving a physician from two stars to four measurably changed which one people picked.
They are not a gate. In our audit, 23% of the clinics in the map pack had fewer than 100 reviews. Ten percent had fewer than 50. The range ran from 2 to 3,316, and one clinic ranked with none. The median was 160. Get reviews because they convert. Do not wait until you have a number before doing the rest of this guide.
Ask at discharge. It is the one moment in the plan of care when the patient feels the best they have felt since the injury, and it is a date your practice management software already knows. Ask in person, then follow with a link the same day. A card with a QR code at the front desk; one shared review link, not a different one per therapist.
Velocity beats total. Ten reviews a month for a year beats a hundred reviews from 2022. Recency is what the searcher sees and what Google appears to weight.
Respond to every one, within 48 hours. A visible reply tells the next reader someone is home.
Build compliance in on day one, because the reply is the risk. You may ask any patient for a review. What you may never do is confirm, in public, that they were one. "Thank you for choosing us for your knee rehab" is a disclosure. "Thank you for taking the time to leave feedback — we're glad to hear it" is not. Never name the condition, the treatment, or the fact of treatment in a reply, even when the patient did. Texting the request has its own rules under the TCPA — the healthcare exemption is narrower than most owners assume, and one sentence can turn a reminder into a marketing message. And never gate reviews — asking only happy patients, or routing unhappy ones to a private form — which is an FTC problem on top of a Google one.
We are not lawyers and this is not legal advice. Your counsel decides.
Two replies that work.
Positive: "Thank you for taking the time to write this. It means a lot to the whole team. We're here whenever you need us."
Negative: "Thank you for the feedback. We take it seriously and would like to understand what happened — please call [front desk name] at [number] so we can talk it through directly."
Neither confirms the person was a patient. Both sound like a human.
A citation is your name, address, and phone number on someone else's site. Ten years ago they were a major ranking factor. Today they are a verification factor — and, in healthcare specifically, they have become something else again, which Part 6 explains.
The core set. Google, Apple Maps, Bing Places, Yelp, Facebook. The healthcare directories: Healthgrades, Zocdoc, Vitals, WebMD, and APTA's Find a PT. The insurer provider directories for every plan you accept — patients use these to find in-network care, and they are wrong more often than you would think. Your state licensing record.
Then the aggregators, which seed the long tail of directories you will never visit.
Identical everywhere. The same name, the same address format, the same phone. "Suite 200" on one and "#200" on another is a mismatch. Find and remove duplicates — old addresses, a previous practice name, a listing the last agency created.
Why this matters more than it did. Across 6.8 million citations in AI answers, 52.6% of healthcare citations came from third-party listing sites — the heaviest listings dependence of any industry measured (Yext, 2025). When ChatGPT names a physical therapy clinic, the odds are better than even that it learned about the clinic from a directory. Being present, accurate, and complete on the directories your specialty actually uses is not a tidiness exercise. In healthcare, it is most of the AI surface area.
Name, address, and phone in the footer of every page. Our audit found 45% of ranking clinics do not have this, and 17% have neither a footer NAP nor a phone number in the header. It is the cheapest thing on this page, and most skip it.
A location page with an embedded map, written directions from the landmarks people actually use, parking, and transit. One per physical clinic. Never several addresses on one page.
Schema markup for what the clinic is and who works there: MedicalBusiness for the practice, Person or Physician for each clinician, FAQ markup on service pages. This helps search engines understand the page. It does not, on the evidence, help AI citation — Part 6 — so the facts go in the visible copy, and schema describes them.
Local context in the copy. The neighborhoods you serve, the teams you sponsor, the running club that meets outside your door. Written for a patient, not for a crawler.
The physician who refers to you has a website. So does the gym you rent from, the running club, the high school team you cover, the chamber of commerce, the local paper that wrote about the 5K you sponsored. Each of those is a link or a mention from a site that is local, relevant, and impossible for a national chain to replicate.
Being mentioned by name, without a link, also counts — more than most people assume. Across 75,000 brands, unlinked brand mentions correlated with AI visibility at 0.66; backlinks managed 0.22 (Ahrefs, 2025). Correlation, not proof of cause. But it points somewhere useful for a clinic: local press, a physician's blog, a patient's YouTube video, a coach's newsletter. None of them need to link to you to count.
Part 5 of 12
Key takeaways
Most physical therapy websites are not under-optimised. They are structurally broken. One "Services" page that lists everything the clinic does. A blog with nine posts from 2022. A homepage trying to rank for every term at once. No amount of writing fixes that, because the problem is not what the pages say; it is that there is no page for the thing the patient searched.
In our audit of 100 ranking clinics, one in five still ran a single Services page instead of a page per service. Thirty-eight percent had no condition pages at all. These are the clinics that rank. The ones that don't are worse.
The job of architecture is three things: give every search intent its own home; make it obvious to Google which page answers which query; and pass authority from the pages people read to the pages that book evaluations.
The six page families. Keep them separate.
Service pages. Commercial intent — "I need pelvic floor PT, who does it near me." One per service line. These are the pages that convert. Everything else on the site exists to feed them.
Location pages. Commercial intent with a place attached. One per physical clinic: hours, the clinicians who work there, the services offered at that location, the payers accepted there, directions, parking. Never several addresses on one page. Never a page for a town you are not in.
Condition pages. Problem intent, close to commercial — "sciatica physical therapy," "physical therapy for vertigo." One per condition you treat and want more of. Written to the patient's problem. Each links to the service that treats it.
Bridge pages. Question intent, one step from booking: cost; insurance and payment; direct access and whether a referral is needed; what to expect at the first visit. Three or four pages every practice needs and almost none has — three of the hundred clinics we audited had all of them.
Long-form resource pages. Informational intent, evergreen, meant to be the best page on the internet for one patient problem. "Recovering from a total knee replacement: what the first twelve weeks look like." "Pelvic floor problems after childbirth: what is normal and when to get help." Not dated, not chatty, updated when the evidence changes. One per major problem your practice wants to be known for.
Blog posts. Informational, timely, narrower. A specific question a patient asked this month. Each one lives under the resource page it belongs to and points at it.
On that last family: 91% of the clinics in our audit have a blog. Nineteen percent have not posted in a year. Several stopped in 2020. And the dead-blog clinics rank alongside the daily-blog clinics, in the same map packs, for the same searches. The blog is not what ranks the clinic. The listing, the reviews, and the service pages are.
The rule that keeps the families apart. A service page never turns into an explainer. A resource page never lists what is included in a plan of care or pitches a package. When a page tries to do both, it ranks for neither, because Google cannot tell which searcher it is for. The intent decides the page type, not the keyword.
Why this matters for physical therapy in particular. The patient's journey runs across all six families in one week. Tuesday, 11pm: "why does my knee still hurt six months after surgery" — a resource page. Wednesday: "physical therapy after knee replacement" — a condition page. Thursday: "physical therapy near me that takes Aetna" — a bridge page, a service page, and a location page. If your site has one page for all three moments, the patient meets it at the wrong moment every time.
URLs that mirror the families. /services/pelvic-floor-physical-therapy/, /conditions/sciatica/, /locations/[city]/, /resources/knee-replacement-recovery/, /blog/…. Flat. Everything reachable in two clicks from the homepage.
Patients do not search for "manual therapy" or "therapeutic exercise." They search for the thing that hurts, the thing they can't do, and the thing they are afraid of. Every page family opens on that.
A service page: "You've been told you need physical therapy after your ACL repair, and you want someone who has done this a hundred times." Then the service.
A condition page: "The pain runs from your lower back down one leg, and sitting makes it worse." Then what it usually is, what physical therapy does about it, and what to expect.
A bridge page: "You want to know what this will cost before you call." Then the answer, in numbers or in the honest range.
A resource page: the whole arc of the problem — what is normal, what is not, when to worry, what you can do today, when to see someone.
The patient's real questions. Will it hurt? How many visits? Will my insurance cover it? Do I need a referral? Can I keep running? How long until I'm back to normal? Do you treat people like me? Each one gets a home on the site, and the page that answers it links to the page that books it.
Proof belongs where the concern is. A review that mentions knee surgery sits on the knee page, not on a testimonials page nobody visits. The clinician's pelvic health certification sits on the pelvic health page.
Tone. Plain, direct, second person. A worried patient at 11pm should be able to read it on a phone and know what to do next.
The pattern is hub-and-spoke, and the spokes point inward.
Blog post → resource page → condition page → service page → location page and booking. Every informational page links down the chain toward a booking. A blog post that does not lead to a service you deliver is costing you money to host.
Resource and blog pages link sideways — to related posts, to the parent resource, to the next question a patient would ask. It keeps readers on the site and tells Google which pages belong together.
Service pages link to the conditions they treat; condition pages link back to the service. Reciprocal, always.
Location pages link to every service offered there; service pages link to every location that offers them.
Bridge pages are linked from everywhere — footer, service pages, condition pages — because the cost, insurance, and referral questions come up at every stage.
Anchor text says what the page is. "Physical therapy for sciatica," not "click here."
No orphans. Every page reachable from at least two others. A page nobody links to is a page Google assumes does not matter.
What not to do. A "related posts" widget that links to random articles. A blog that links only to itself. Service pages with no links out because someone worried about "leaking" visitors.
| Page | Targets | One per |
|---|---|---|
| Homepage | Practice name + primary city + primary service | site |
| Service page | [Service] in [City] | service line |
| Condition page | [Condition] physical therapy [City] | condition you treat and want |
| Location page | Physical therapy [neighborhood / city] | physical clinic |
| Insurance & payment | Physical therapy that takes [insurer]; does insurance cover PT | site |
| Direct access | Do I need a referral for physical therapy [state] | site |
| Cost | How much does physical therapy cost | site |
| What to expect | First physical therapy appointment | site |
| Resource page | The patient's problem, in their words | major problem you want to own |
| Blog post | A specific question, this month | question |
| Team page | [Clinician name] PT DPT | clinician |
| About | Why the practice exists, for a patient | site |
The insurance page names every plan by the name a patient uses — including Medicare and Medicaid if you take them, and workers' compensation and auto injury, which are among the most urgent, highest-intent searches in the category. The direct access page says what is true in your state for your payers: state law permits it everywhere, but APTA found nearly two-thirds of physical therapists still work in settings that require a referral anyway, most citing reimbursement. A page that simply announces "no referral needed" is wrong for a lot of practices. A page that explains the real answer is the most useful page on the site.
Title tag and H1: [Service] in [City] | [Practice]. Plain. The keyword is the service and the place.
Above the fold: the patient's concern in one sentence, what you treat, where you are, a phone number, and a book button. Nothing else.
Proof, immediately: reviews for this service, the clinician's credentials for this service, real outcomes if you have consent to show them.
Conditions treated, each linked to its condition page.
What a visit looks like, and how many visits to expect. Patients ask this and almost no page answers it.
Insurance accepted for this service — because pelvic health and dry needling are not covered the way orthopedic PT is, and the patient knows it.
The clinician's bio, on the page, not a link away.
FAQ, with schema, built from the real questions in 5.2.
Booking, again, at the end.
Links: to related services, to the resource page on this problem, to the location page.
Length: as long as it takes to answer every question above and no longer. A 300-word service page is thin. A 3,000-word one is an explainer wearing a service page's URL.
The title answers the problem, not the service. "Recovering From a Total Knee Replacement: Weeks 1–12." Not "Post-Surgical Physical Therapy Services."
It opens on what the patient is feeling right now. Swelling at week three. Fear of the stairs. The question they are afraid to ask the surgeon.
It is structured by the patient's timeline or decision points, not by the clinic's treatment categories.
What is normal. What is not. When to call someone. In that order.
What the evidence says, with the sources named in visible text. Not in a footnote, not in schema — in the sentence. This is also what gets a page cited by an AI answer (Part 6).
What you can do today, before you see anyone.
Where physical therapy fits and when — then, and only then, the links to the condition page and the service page.
Related questions — the blog posts that live under this resource.
A visible "reviewed by" line with a named clinician, and an updated date.
No package, no pricing, no "call now" above the fold. The call to action is at the end, once, after the page has earned it. The moment a resource page starts selling, it stops being the page people trust, and it stops ranking for the problem.
Mobile-first. Fast — Core Web Vitals in the green. HTTPS. Crawlable: robots.txt permits Googlebot, which Google lists first among its recommendations for AI features. XML sitemap submitted in Search Console. No duplicate pages. Schema validated. Indexing checked — a page that is not indexed is not in the game, and it happens more than you would think.
Click-to-call on every page. Online scheduling that finishes online — 70% of patients in the 2022 survey started booking online and were forced to a phone call to finish. Forms answered the same day. A call answer rate you actually measure. An AI receptionist for the hours nobody is at the desk.
Call tracking, set up so it does not break your NAP: the tracking number goes on the site with dynamic swapping, and the real number stays in the schema and on the listing.
And wait time. If the next evaluation is five weeks out, fix that before spending anything on search. Ranking sends people to a full calendar, and a full calendar sends them to the next clinic.
Part 6 of 12
Key takeaways
Google now answers a large share of searches above the results, with an AI Overview. AI Mode turns the results page into a conversation. The map pack and the blue links are still there; the answer sits on top of them.
And a real share of health questions have moved into chat entirely. When OpenAI launched ChatGPT Health on 7 January 2026, it said that over 230 million people globally ask health and wellness questions on ChatGPT every week — understanding test results, preparing for appointments, evaluating insurance options. Health is one of the most common things the product is used for.
For physical therapy that looks like: "Is this sciatica or something else?" "Should I see a physical therapist or a chiropractor for this?" "Does insurance cover physical therapy in Virginia?" "Do I need a referral?" "Who is a good pelvic floor PT in Arlington?" Questions that used to be typed into Google one at a time are now asked as a conversation. The answer names clinics.
From Google Search Central's documentation, AI features and your website, updated 10 December 2025:
"There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary."
"The best practices for SEO remain relevant for AI features in Google Search."
"You don't need to create new machine readable files, AI text files, or markup to appear in these features. There's also no special schema.org structured data that you need to add."
Google's own list of what to do: make sure crawling is permitted; build internal linking; deliver a good page experience; put content in text form; support it with quality images and video; keep Business Profile data accurate.
Every item on that list is in Parts 4 and 5 of this guide. That is the point of this section. There is no separate AI playbook. A practice that does the local SEO work is doing the AI work.
Google's document tells you what is required. The measurements tell you where the weight is.
Healthcare is the most listings-dependent industry in AI answers. Yext analysed 6.8 million citations across the major AI engines in July and August 2025 and found 52.6% of healthcare citations came from third-party listing sites — WebMD, Vitals, Zocdoc, state licensing databases. The heaviest listings dependence of any industry measured. For a clinic, the citation work in 4.3 is most of the AI surface area.
Ranking helps, but it is not the same job. Only about a third of AI Overview citations come from the organic top ten (seoClarity, 362,000 keywords). For local healthcare queries, one academic study found roughly 12% overlap between the domains AI cited and the domains Google ranked. Directories, reviews, and mentions fill the gap.
Being mentioned beats being linked. Unlinked brand mentions correlated with AI visibility at 0.66 across 75,000 brands; backlinks at 0.22 (Ahrefs, 2025). The local mentions in 4.5 — press, referral partners, the running club — count without a link.
Pages with sources get cited. The one controlled study on this — Princeton and IIT Delhi, published at KDD 2024 across roughly 10,000 queries — found that adding statistics, cited sources, and quotations to a page raised its citation rate in generative engines by 30 to 40%. This is why the resource pages in 5.6 name their sources in the visible text. It is not a stylistic preference; it is the only thing anyone has demonstrated works.
Schema does not lift AI citation. Google says none is needed. A controlled test of 1,885 pages that added JSON-LD found no measurable citation gain (Ahrefs, 2026). Keep schema for search understanding. Put the facts in the copy.
AI engines deprioritise brand-owned and social content. The same academic study found social profiles near zero percent of citations in several categories. Keep your profiles accurate because patients read them. Do not buy "social profile optimisation for AI."
And the facts have to be on the page to be read. An AI answer cannot tell a patient which insurance you take if you never published it — and in our audit, 71% of ranking clinics never did.
| The question | The page that answers it |
|---|---|
| "Is this sciatica?" | Resource page |
| "PT or chiropractor for this?" | Resource or condition page |
| "Does insurance cover PT? Do I need a referral in [state]?" | Bridge pages |
| "How much does physical therapy cost?" | Cost page |
| "Who is a good pelvic floor PT in [city]?" | Business Profile + directories + reviews + location page |
The same architecture as Part 5. The AI assembles its answer from the same pages a searcher would have landed on.
Within what Google has published and what has been measured: it reads the listings — name, category, services, hours; the reviews — what patients say you treat; and the site — what it says plainly, in text, about what you treat and where. The clinic that is consistent across all three, with real reviews and a page that says "we treat X in [city]" in plain words, is the one that gets named. Beyond that, nobody outside Google knows the mechanism, and anyone who tells you they do is selling something.
| The SEO task | What it does in Google | What it does in AI answers |
|---|---|---|
| Accurate, complete Business Profile | Map-pack ranking | On Google's own list for AI features; the primary local data source |
| Reviews, asked and answered | Ranking and conversion | What the model reads to learn what you treat |
| Citations and consistent NAP | Trust, verification | 52.6% of healthcare AI citations |
| Service pages in plain text | Organic ranking | "Content in textual form" — Google's requirement |
| Resource pages with visible sources | Long-tail organic | +30–40% citation rate |
| Internal linking | Authority flow | On Google's list |
| Local mentions | Local authority | 0.66 correlation with AI visibility |
"GEO packages" sold as separate from SEO. An llms.txt file as a ranking lever. AI-specific schema. Social profile optimisation for AI. Anyone promising to "get you into ChatGPT" without doing the listings, the reviews, and the pages first. Every one of these is being sold right now, and none of them has evidence behind it.
Ask ChatGPT, Gemini, and Perplexity the ten questions your patients ask, with your city attached. Note which clinics are named and which sources are cited. Repeat quarterly. The sources named tell you which directories and pages to fix.
Solo: your Business Profile, the core listings, and one resource page with real sources. That is the entire AI strategy. There is nothing to add.
Small-to-mid: three to five resource pages with visible sources, and a program for getting mentioned locally.
Specialized: resource depth is the moat. Pelvic floor after childbirth, vertigo, return to sport after ACL — these are exactly the questions people ask a chatbot before they call anyone. The practice that wrote the best page on the problem is the one that gets named.
How we do SEO and AI visibility for clinics→Part 7 of 12
Key takeaways
Everything above is the work. This is the order, the scope, and the budget for each kind of practice — and what to skip, which matters as much.
Your situation. You are the clinician. Treating hours are revenue. You have no admin time, a room in a gym or a small suite or a car, and one or two things you are known for.
The honest picture. In our audit, 70% of the clinics in the map pack were multi-location operators. Independents were 30% — outnumbered, but there. And the ones ranking with under 100 reviews, some with under 50, show what it takes: a correct listing, a small site that answers the three questions, and reviews asked for on purpose. You do not need to out-build a chain. You need to be exactly right about a small number of things.
Your goal. A map-pack slot in your immediate radius for your two or three core searches. Two to four new evaluations a month from search. That is a real number for a solo practice and it changes a year.
Your keywords. [Core service] near me and [core service] [city]. Your specialty term. One or two conditions you want more of. "No referral" or "cash pay" if that is your model.
Business Profile. Everything in 4.1, done once, properly. Correct primary category. Twenty reviews in the first ninety days, asked for at discharge. A photo a week — it takes thirty seconds.
Website: eight to ten pages. Home. About, written for a patient. Two or three service pages. Two or three condition pages. The cost-and-insurance bridge page. The referral bridge page. Contact and location — or a service-area page if you are mobile. One resource page, on the one problem you most want to own, written once and kept current. No blog until all of that exists.
Citations. The core ten to fifteen. Not two hundred.
Reviews. You ask, in person, at discharge. A card with a QR code. One link.
Local links. Three to five relationships — the gym, a physician, the running club. Not a campaign.
Budget and time. Doing this yourself is realistic: one afternoon to set up, one or two hours a month after. If you would rather not, the lowest retainer band — Ahrefs' poll of 439 providers found $501–$1,000 a month the most common — buys this scope. Above that, for a solo practice, you are paying for things you should skip.
Skip. The blog. Multiple location pages. Paid directories. "AI optimisation" of your social profiles. The fight for "physical therapy near me" against the hospital system.
Ninety days. Weeks one and two: Business Profile and citations. Weeks three to six: the pages. Weeks seven to twelve: reviews and the three local relationships.
The mistakes that are specific to you. A home address on the Business Profile. Sharing a listing with the gym you rent from. Letting reviews stall at eight. Letting a vendor own the profile. A single Services page that lists everything you do.
The clinic owner's guide to doing your own local SEO→ If the constraint is time, not knowledge→
Two to ten clinicians, one or two locations, a front desk.
Your situation. You can absorb new evaluations, which is the thing that makes search worth paying for. You have several service lines, a front desk that can run a process, and you are competing with hospital outpatient departments and national chains that have more of everything except attention to your market.
Your goal. Top three in the map pack across a wider grid, not just at your front door. Organic top five for service-plus-city terms. Ten to twenty new evaluations a month from search.
Your keywords. The full service-by-city matrix. Your top ten conditions. Every payer term. "Best physical therapy in [city]."
Business Profile. One per location. Practitioner listings managed so they do not compete with the clinic's. Services fully mapped. Q&A seeded. A post a month.
Website: thirty to fifty pages. Every service page. The top ten condition pages. All four bridge pages. Team pages with credentials. A what-to-expect page. One location page per clinic. Three to five resource pages on the problems you want to own. A blog that publishes only under those resources, on questions the front desk actually hears — and if the front desk is not hearing questions, do not publish. Internal linking audited once a quarter against 5.3.
Reviews. A front-desk workflow, not a hope: a compliant automated request at discharge, a velocity target, a reply within 48 hours, the two templates from 4.2.
Citations. The core set plus the healthcare directories, the insurer directories for every plan you take, and the aggregators. Duplicate cleanup once.
Local links. A physician outreach program with a name and a calendar. Sponsorships you would have done anyway, with a link attached. Local press when there is a reason.
Conversion. Online scheduling that finishes online. A measured call answer rate. Call tracking. Front-desk scripts. And wait time, fixed first.
Reporting. Grid rank tracking, calls, forms, booked evaluations, cost per evaluation. Part 8.
Budget: do the arithmetic before the call. Take your collected revenue per visit — collected, not billed. Multiply by your average visits per completed episode. That is your episode value. Divide by your evaluation-to-plan-of-care rate. That is what a booked evaluation is worth to you.
For the sake of an example only — these are not our figures and not industry data — suppose an episode is worth $900 and 60% of evaluations become a course of care. An evaluation is worth $540. At a $1,500 retainer, search has to produce fewer than three additional evaluations a month to break even.
What is a booked evaluation worth to you?
Runs in your browser. Nothing is sent anywhere. Use collected revenue, not billed.
Run it with your own numbers. Not because it tells you whether to hire anyone — it doesn't — but because a clinic that knows an evaluation is worth $540 asks completely different questions from one that only knows the retainer is $1,500.
A second location gets its own profile, its own page, its own review stream. Never merge them.
Skip. Pages for towns you are not in. Blogging for its own sake.
The mistakes at this size. One service page for everything. An insurance page that was accurate two years ago. A blog that links only to itself. Review replies that confirm the reviewer was a patient. Texting patients without the consent the TCPA requires.
How we do SEO for clinics→ Stop leaking patients at the phone→ What we charge, and why→
Pelvic health, vestibular, sports, pediatric, hand therapy, cash-based, out-of-network, in-home.
Your situation. Your patient is choosing you rather than picking the nearest covered option, and is prepared to travel and to read before they call. Your keyword pool is smaller and much more specific. Proximity works less in your favour than it does for a general clinic; education and credentials work more.
Your goal. Rank for the specialty term across the metro, not the neighborhood. Convert readers who need convincing. Be the clinic an AI names when someone asks the specialty question.
Your keywords. [Specialty] near me, [specialty] [city], [specialty] [metro]. The condition terms tied to the specialty — diastasis recti, vertigo, ACL rehab, torticollis. "Cash based physical therapy." "Does insurance cover [specialty] physical therapy."
Look at the table in Part 2 again. Pelvic floor physical therapy near me: 1,400 searches a month, difficulty 35. Vestibular therapy near me: 600, difficulty 0. Dry needling near me: 6,700, difficulty 0. Cash based physical therapy: 400, difficulty 0. These are your searches, and almost nobody is competing for them.
Business Profile. The primary-category decision is the big one, and it is made on search volume, not on pride. "Physical Therapist" as primary with your specialty as secondary usually wins more than the reverse, because the specialty category cuts you out of the general pool. Test it if you can. Services and photos that show the specialty.
Website: the resource layer is where you win. A specialty hub page. Condition pages — for you, these are the bottom-of-funnel pages, because your patient searches the condition. A cost page, because your patient is often paying. An insurance and superbill explanation, because they will ask. A what-to-expect page, because specialty patients are anxious and it is the first thing they read. And the deepest resource pages on the site — the specialty patient reads everything before they call, and the practice that wrote the best page on their problem is the one they trust. Credentials and certifications on every page, not on a team page.
Reviews. Your patients will name the condition themselves. You never do, in a reply. For a pelvic health practice this is not a technicality.
Citations. The general set, plus the specialty directories: APTA's specialty sections, the pelvic health provider directories, the vestibular association's provider list, the sports medicine networks. These are where your referral sources and your patients look.
Local links. Referral partners specific to the specialty — OB/GYNs, midwives, ENTs, orthopedic surgeons, coaches, athletic trainers.
Three sub-cases.
Cash-based and out-of-network. Proximity-weighted local search is calibrated for the in-network patient choosing from a covered list. Your patient is choosing you, and will travel to do it. Your content has to do the convincing that insurance normally does: the about page, the condition pages, and the price on the page. Hiding the price is the objection most cash practices handle worst.
In-home and mobile. You are a service-area business on Google: no address shown, a service area instead, and a listing that is harder for Google to verify. Service pages per area instead of one address page. A worked example is live → Case study: launching a new in-home PT practice
Multi-location specialty. One profile and one page per location, as in 7.2.
Skip. The fight for generic "physical therapy near me." Condition content outside your specialty.
The mistakes. The wrong primary category. Hiding the price. A pelvic health page with no pelvic health therapist behind it. Resource pages that pitch instead of explain.
Part 8 of 12
Key takeaways
Five layers, in order. Each one should move before the next.
Visibility. Map-pack position on a grid — not a single point at your front door, but a grid of points across the area you want to win. Organic position for the searches on your list.
AI visibility. Which of your patients' ten questions name you, checked quarterly (6.8).
Traffic. Business Profile views and actions. Site sessions to the pages that matter.
Actions. Calls, direction requests, bookings, form fills.
Outcomes. Booked evaluations. Plan-of-care conversion. Revenue.
The tools. Business Profile performance. Search Console. GA4. Call tracking. A grid rank tracker. And the "source" field in your EMR, which is the only tool on this list that tells you what an evaluation was worth.
Timeline. What we see, not what any industry report says: the listing can move inside a month if it was genuinely wrong before. Service-plus-city terms take three to six months. Competitive condition and specialty terms are six to twelve. AI answers follow the directories and the resource pages, on their own schedule.
Attribution. Ask every new patient how they heard about you and write it down. "ChatGPT" is now an answer you will hear.
Get a free grid audit of your current visibility→Part 9 of 12
Key takeaways
The wrong primary category. Keywords in the business name. A vendor that owns the Business Profile. A home address listed. Duplicate listings. One page for all services. Several addresses on one page. Service pages that turn into explainers. Resource pages that turn into pitches. A blog that links only to itself. An insurance page that is out of date. No direct access page. Not asking for reviews. Review replies that confirm the patient. Texting without consent. Geotagged photos. Buying "AI optimisation" as a separate service. A solo practice chasing the head term. No online booking. No call tracking. A slow site.
Every one of these is on a checklist somewhere right now, being done or being sold. That is why they persist: a checklist item that does nothing looks exactly like one that works, unless somebody tests it.
Part 10 of 12
Key takeaways
What you can realistically do yourself. At solo scale, all of it — 7.1 is an afternoon and an hour a month. At small-to-mid scale, the Business Profile, the reviews process, and the citations; the site architecture and thirty pages are where most owners stall. At specialty scale, the resource pages need a clinician's time either way, and that clinician is usually you.
When hiring makes sense. You have no time and know it. You are opening a second location. You are pushing into a specialty and need the resource layer built properly. A competitor has moved in and the map pack has changed.
Eight questions to ask any agency, and listen for specifics rather than reassurance:
Ask all eight of three agencies. The differences will tell you more than the proposals.
What we promise, and what we don't.
We do not guarantee rankings, and you should be wary of anyone who does. Not because guarantees are bad — we put a hard one on our front-end work, with a number and a deadline attached. But a ranking guarantee is a promise about a system nobody selling it controls. Google ships updates nobody is warned about. The results page keeps changing shape. AI Overviews now sit above results that used to be the whole game.
What we do instead: we keep working until your rankings move. At the start of an engagement we agree, in writing, which searches and which grid positions count as "moved," and what we need from you to get there — access to the profile and the site, approvals inside a week, and review requests actually being made at discharge. Hold up your side and we hold up ours, for as long as it takes.
What we guarantee, and what we don't→Part 11 of 12
Key takeaways
How long does physical therapy SEO take? The Business Profile can move inside a couple of months if it was wrong before. Service-plus-city terms, three to six months. Competitive condition and specialty terms, six to twelve. Any shorter answer is describing advertising.
How much does it cost? Ahrefs polled 439 SEO providers: those serving local markets charged $1,557 a month on average; the most common band was $501–$1,000; roughly 69% charged $2,000 or less. That polls providers, not clinics, and the underlying data is a few years old, so treat it as a floor. The number that matters more is what an evaluation is worth to you — 7.2 shows the arithmetic.
Do I need a blog? Not the way it is usually sold. In our audit, 91% of ranking clinics have one, 19% have not posted in a year, and both kinds rank side by side. Build the bridge pages and the condition pages first. A blog comes after, under a resource page, on questions patients actually ask — or not at all.
Should I build a page for every nearby town? Not until the town you are in is producing. Practice-type pages convert better than location pages for towns you are not in, because a page with a city in the heading and nothing else reads like a mail merge.
Can I rank without a physical address? Yes, as a service-area business. It is harder to verify, and the playbook is different — 7.3.
Do reviews affect ranking, or only conversion? Both, and conversion is the bigger one. And they are not a gate: 23% of the clinics in the map pack have fewer than 100.
Should the Business Profile be in my name or the practice's? The practice's, owned by the practice owner's Google account. Practitioner profiles for individual clinicians exist and should be managed, not left to compete.
Does any of this help with ChatGPT and AI Overviews? Yes. Google's own documentation says there are no additional requirements and the same SEO best practices apply. Part 6.
Do I need an llms.txt file or AI schema? No. Google says so, in the same document.
Is SEO worth it if my schedule is already full? Not yet. Fix capacity first, or search sends people to a waiting list and they book elsewhere. If you are full because of one referral source, though, read Part 2 again — that is the situation search is for.
Part 12 of 12
Key takeaways
| Solo | Small-to-mid | Specialized | |
|---|---|---|---|
| Do first | Business Profile, correct category, 20 reviews | One profile per location, review workflow, bridge pages | Primary-category decision, condition pages, price on the page |
| Website | 8–10 pages, one resource | 30–50 pages, 3–5 resources | Hub + conditions + deep resources |
| Skip | Blog, town pages, the head term | Town pages, blogging for its own sake | Generic near-me, content outside the specialty |
| Budget band | DIY, or $500–$1,000 | The arithmetic in 7.2 | Clinician time on resources; retainer for the rest |
| 90-day target | Map pack in your radius; 2–4 evals/month | Top 3 on a grid; 10–20 evals/month | Specialty term across the metro |
Nothing on this page is a secret. Most of it is available to anyone willing to read a few studies and do four minutes of arithmetic on their own practice data. The reason so few clinics do it is not that it is hard. It is that the referrals keep arriving, and a pipeline that keeps arriving never feels like it needs attention — even while a share of it is leaking to a clinic whose only advantage is being easier to find.
Get a free grid audit of where you stand→ Or book a call and we'll walk through your practice type and market→
Sources
Search volumes are from Ahrefs, August 2026. Audit figures are from our own audit of 100 physical therapy clinics across 12 US metros, September 2026 — the full data and method publish as their own page. Everything else is peer-reviewed research, an industry measurement study, or Google's own documentation.
We are not lawyers and this is not legal advice. Your counsel decides.