AI Marketing for Chiropractors: Claim Limits and Review Replies

Give AI an approved list of what you may say, ban outcome words such as cure, fix and guaranteed, and let it draft posts, page copy and review replies from that list only. Then check every line against the advertising guidance and your evidence before publishing. Review replies never confirm a patient, a condition or a treatment.

The risk with AI is not that it writes badly. It is that it writes fluently and confidently, and fluent copy about back pain drifts straight into claims you cannot support: "relieves sciatica in weeks", "safe for babies", "natural alternative to medication". One page full of those can draw a complaint, so the guard rails come first and the content second.

Follow me on Instagram@sagnikteaches

What the advertising rules actually limit

Two regulators publish guidance that makes a useful worked example: the ASA and CAP (which write and enforce the advertising codes in their market) and the FTC. Which one applies to you depends on where your customers are, so read the one that governs your advertising, and ask your professional body or an adviser what your own regulator adds. What follows is the practical content of their published guidance, not legal advice, and it changes: open the live pages before you rely on any line.

Connect on LinkedInSagnik Bhattacharya

ASA and CAP guidance on chiropractic claims

The ASA chiropractic advice page (dated June 2023; check it for later changes) lists the conditions that, on the evidence submitted and reviewed so far, the ASA and CAP accept chiropractors may claim to treat. The list includes general backache and back pain not arising from injury or accident, mechanical neck pain (as opposed to neck pain after an injury such as whiplash), headache arising from the neck, migraine prevention, joint pains, muscle spasms, cramp, minor sports injuries, short-term management of ankle sprain and of plantar fasciitis, some shoulder complaints, and sciatica in general references only. Anything else needs strong clinical evidence behind it. The page points to a separate ASA guidance note, written after a 2017 evidence review, on sciatica, whiplash, sports injuries and treating babies, children and pregnant women. That note and the wider CAP Code add these points:

Subscribe on YouTube@sagnikteaches

• Saying chiropractic can help with sciatica is acceptable. Saying or implying it will eliminate sciatic pain is likely to mislead, and the note urges caution about claims on reducing its intensity.
• Claims that chiropractic helps with whiplash, or with pain from other neck injuries, are likely to mislead, because the ASA and CAP have not seen sufficient evidence. The note says this applies to client testimonials used in marketing too.
• You may mention treating babies, children and pregnant women, but claims about specific conditions such as colic or morning sickness are unlikely to be acceptable without a substantial body of evidence, and the ASA describes the current evidence for those as limited or negative.
• CAP's advice on testimonials says the code applies to claims in a testimonial exactly as it does to claims anywhere else in an ad, and that testimonials alone do not constitute substantiation.
• The CAP Code's health section (rule 12.2) says marketers must not discourage essential treatment for conditions that need medical supervision.

FTC principles for health claims and testimonials

The FTC's Health Products Compliance Guidance says health benefit claims need competent and reliable scientific evidence, which for benefit claims generally means randomised, controlled human trials. Its endorsement guidance says an endorsement cannot be used to make a claim the marketer could not legally make itself. So an honest patient saying "it fixed my sciatica" does not make the claim safe. If you cannot show that the endorser's experience is what people will generally achieve, the ad has to make the generally expected results clear, and the health-products guidance adds that a "results not typical" disclaimer does not cure a testimonial that overstates results. That guidance is written for health products rather than services, so read it as the clearest statement of the principle, not as a rulebook for clinics: no outcome claim without evidence, and no testimonial that carries one.

What this means for your copy

Whichever guidance applies, the working rules are the same. Describe what you do, who you see and how an appointment works. Refer to the conditions you are allowed to name, in general terms. Avoid outcome promises, timescales, comparisons with medication or surgery, and claims about conditions you cannot back up. Do not suggest a patient should stop or avoid medical care.

Build a claims register before you prompt anything

A claims register is a one-page list that every piece of AI-drafted marketing is checked against. It takes an hour and turns checking from opinion into a lookup. Make it a spreadsheet with four columns: the claim, status (Approved, Approved with wording, Not allowed), the evidence or guidance behind it and the date you last checked.

Notice that the safer wording says "we see people with" rather than "we treat" or "we fix". That small shift describes the service rather than promising an effect. Add your own rows as drafts throw up new phrases. After a month the register is the best prompt you own.

The prompts: draft inside a fence, then audit with a second pass

Run two prompts, never one. The first drafts within your rules. The second acts as a sceptical reviewer, and a person reads its flags.

Drafting prompt for a Google Business Profile post

You write marketing text for a chiropractic clinic. Use ONLY the approved
facts below. Do not add any claim about results, speed, cure, safety, or
comparison with medication, surgery or other clinics. Do not mention babies
or children's conditions. Do not use testimonials.

Approved facts:
- Registered chiropractors, [names removed]
- First appointment 45 minutes, includes history and examination
- We see people with back pain, neck pain linked to posture or desk work
  (not pain after an injury or accident), joint pain and headaches that
  start in the neck
- Open Mon-Sat, [hours]; booking at [link] or [phone]

Task: write a 90-word post inviting people to book a first appointment.
Plain, calm tone. No exclamation marks. End with the booking link.

Auditing prompt

Review the draft below as an advertising compliance checker for health
services. List every sentence that could be read as: an outcome promise, a
speed or duration claim, a cure or elimination claim, a safety claim, a
comparison with medication or another provider, a claim about a specific
condition not on the approved list, or advice discouraging medical care.
Quote the sentence and say which type it is. If none, say "No flags".
Do not rewrite it.

An illustrative first draft from a model told only "write a post about back pain" might read: "Say goodbye to back pain with gentle, natural chiropractic care that gets you moving again in weeks." The auditing prompt flags three phrases: "Say goodbye to" (elimination), "gentle, natural" (safety and comparison) and "in weeks" (timescale). The fenced draft instead produces: "Back pain, neck pain from desk work or headaches that start in the neck getting in the way? Our registered chiropractors see people with these problems every week. Your first appointment lasts 45 minutes and includes a full history and examination. Book online at [link] or call [phone]." You still read it, but there is nothing to remove. Notice how narrowly the neck pain is described: the register ties it to everyday mechanical causes, because pain after an injury or accident is the claim the guidance rules out, and a draft that says only "neck pain" next to a photo of a car can imply exactly that.

A 30-day posting plan, drafted by AI and checked by you

Google Business Profile posts are a steady source of fresh content that needs no ad spend. Posts can be scheduled natively in the profile, and a post takes a description, a photo, a video and a button. Events need a start and end date, and Google archives posts older than six months unless they carry a date range, so keep the evergreen ones short and the dated ones clearly dated. Say the clinic posts twice a week, eight posts a month:

1. Week 1: a "what to expect at a first appointment" post and a "meet the team" post using a real photo.
2. Week 2: an opening-hours post for a public holiday and a post explaining the 45-minute first visit.
3. Week 3: an educational post about desk set-up, written as general information with no promised effect, plus a post about how to prepare (clothing, forms).
4. Week 4: a post on how to reach the clinic and parking, and a post inviting existing patients to leave feedback, using neutral wording with no reward.

Use real photographs of your premises and staff. Google's photo guidelines say an image should represent reality, and generated treatment images or stock photos of smiling patients can suggest results you do not own. If you want more help with the local side, see AI for chiropractors' local pages and reviews.

Time sum, illustrative: eight posts at 10 minutes to draft with AI, 5 minutes to check against the register and 3 to upload and schedule is about two and a half hours a month. Without the fence you might draft faster, but one removal request or complaint costs more than that.

Replying to reviews without patient detail

Reviews are where AI is most useful and most dangerous. The useful part is tone: it softens a defensive reply and gives a consistent voice. The dangerous part is that it follows the reviewer's lead, repeating what they said, and that repetition is where you disclose. Rules for every reply:

• Do not confirm the person was a patient, even if they say so.
• Do not name a condition, treatment, date, clinician or fee.
• Do not repeat an outcome claim. If a review says "cured my sciatica", do not reply "so glad we cured it".
• Do not argue facts in public. Offer an offline route to the practice manager.
• Do not offer anything in return for a review, or for changing or removing one. Google's policy bans incentives such as discounts or free services in exchange for reviews.

Three before-and-after replies

Positive review that contains a claim. Review: "Dr [name] fixed my back in three visits, I'm pain free!" First AI draft: "Thank you, we are delighted that three visits fixed your back." That repeats the claim. Safer: "Thank you for taking the time to leave kind feedback. We are glad you had a good experience at the clinic."

Negative review about service. Review: "Rude receptionist and my neck is worse." Safer: "Thank you for letting us know. We take feedback seriously and would like to look into it. Please contact our practice manager on [number]." No mention of the neck, the receptionist or any treatment.

Review that looks fake. Review: "Terrible, never been here but heard bad things." Do not accuse. Flag the review to Google through the profile and post a short, neutral reply: "We cannot find a record of this feedback and would welcome a conversation. Please contact the practice." Even "we cannot find a record" implies you checked patient files, so many clinics prefer only "We would welcome a conversation if you would like to contact the practice." Pick one approach and use it consistently.

The prompt:

Draft a public reply to the review below for a chiropractic clinic.
Rules: under 50 words; do not confirm the reviewer is a patient; do not
mention any condition, treatment, clinician name, date or fee; do not repeat
any claim about results; no offers or incentives; invite contact with the
practice manager on [number] if the review is negative. Warm, plain tone.
Review: [paste review text]

A person reads the draft and the review together before posting. The tutorials on patient reviews and confidentiality and on whether AI should auto-post review replies argue the same point: drafts are fine, automatic publishing is not.

What patient information may go into the AI tool

Marketing needs no patient data at all, and that is the best protection. Do not paste a review that names a patient into a tool without checking your agreement and settings, and do not paste notes or case details to "make a story". If you want a case-based post, write a composite described as general information, or skip it. The tutorial on using ChatGPT safely without a signed agreement lists what is safe to enter and what is not. Business plans of the main assistants do not train on your content by default, but check the setting on your own account.

Before-and-after for a landing page section

Illustrative page intro generated without a fence: "Struggling with back pain? Our world-class chiropractors will fix the root cause so you never have to suffer again. 98 per cent of patients recommend us." Problems: elimination claim ("fix the root cause", "never suffer again"), unsupported superlative ("world-class"), and a statistic with no source. A fenced version: "If back or neck pain is making daily life harder, you can book a first appointment with one of our registered chiropractors. The visit lasts 45 minutes and includes a history and examination, and the chiropractor will explain what they find and the options before any treatment begins. Call [phone] or book online." Check what remains: every statement can be shown to be true from practice facts, nothing promises an effect.

Statistics deserve their own rule. If a sentence has a number, put the source in the register or delete it. AI invents plausible figures, and an invented percentage on a clinic website is among the easiest claims to challenge.

Social captions, newsletters and chatbot answers need the same fence

The claim problem does not stop at the website. A Facebook caption, an Instagram carousel, a monthly newsletter and a chatbot on your booking page are all advertising when they promote your service, and AI drafts all of them in the same overconfident voice. Reuse the same approved-facts block in every prompt and the same auditing pass on every output.

A newsletter example, before and after

Illustrative first draft for a monthly email: "This month we are talking about posture. Poor posture causes back pain and our adjustments will correct it. Book now before slots fill." Flags: "causes back pain" (causal claim), "will correct it" (outcome promise) and "before slots fill" (pressure that may sit badly with a health service). A checked rewrite: "This month's note is about desk set-up. Here are three small changes people often find comfortable: raise the screen to eye level, stand up every 30 to 45 minutes and keep your feet flat on the floor. If you are having problems with pain, you can book a first appointment at [link]." The tips are general and carry no promised effect. A person should still confirm none of the three tips contradicts what your clinicians actually advise.

A chatbot or enquiry reply

If you use AI on your booking page or in messages, it must not give clinical advice or answer "Can you fix my sciatica?" with a yes. Set the rule in the bot's instructions: answer only from an approved fact sheet (hours, prices, location, how bookings work), and for anything about a symptom reply "A chiropractor can discuss that at an appointment. If you have severe pain, numbness, weakness or you are worried, please contact a doctor or emergency services." That last line matters, because the guidance warns against discouraging essential medical care. Test the bot with ten awkward questions before launch and keep the transcripts of the tests.

The colic claim a skim-read let through

Imagine a clinic that asks AI for a "services" page and publishes it after a skim. Three weeks later a competitor reports the page. The draft had said "suitable for infants with colic" and "clinically proven". The practice had no evidence for either. The clinic removes the page, spends a few days dealing with the complaint and pulls its other recent posts while it checks everything else. What would have caught it: the register had no line for colic, the auditing prompt had never been run and no one had read the page against the guidance. The fix is not to avoid AI, it is to never publish an AI draft that has not passed the register and the audit prompt.

Photos, consent and before-and-after material

Patient photos and posture or X-ray images are health information. Do not use them in marketing without written, specific consent, and consider whether to use them at all, because a before-and-after posture picture implies a result. If you use a model or stock image to illustrate a page, caption it neutrally and do not place it beside a claim about results. Keep the written consent on file with the date and the exact use allowed, and remove the image when consent is withdrawn. AI image tools can make attractive but misleading spine graphics, so check that any diagram you publish shows anatomy accurately and that a clinician has approved it.

Who signs off and how long it takes

Name one person, ideally a clinician or the practice owner, as the sign-off for anything that makes a statement about care. Ask them to apply a three-question test to every draft: can I prove this sentence, could a reasonable patient read it as a promise, and would I be comfortable if the regulator read it? A typical check of a 90-word post takes about two minutes once the register exists. A longer page might take ten. That is the real cost of AI marketing in a regulated profession, and it is worth paying every time.

A monthly check routine

1. Open the register and read the "Not allowed" rows aloud to anyone who writes copy.
2. Re-read the live guidance pages once a quarter, and note any change and its date in the register.
3. Sample five published posts or web pages and run the auditing prompt on them. Fix anything flagged.
4. Check review replies from the last month for any condition, name or date.
5. Remove anything old that no longer meets the register, including pages, posts and image captions.

Take the register to your professional body if it publishes its own advertising guidance and compare. The regulator and the body can both be stricter than the examples here. A clinic that treats the register as a working document does fewer rewrites, answers a complaint faster, and has a record showing the steps taken.

If you would like help turning this into a routine that suits your clinic's software and staff, my AI implementation consultation covers it; you can get in touch here.

Chiropractic marketing and AI: common questions

Can AI write my website copy about what chiropractic treats?

It can draft it, but it will often overclaim. Give it your approved list of conditions and ban outcome words, then check every sentence against the advertising guidance for your profession and the evidence you hold. If a claim is not on your list or backed by evidence, it comes out.

Can I share a patient's glowing review on my website?

Treat a review that says you fixed a condition as a claim, not decoration. Advertising regulators generally treat testimonials as implied efficacy claims, and they do not replace evidence. Quoting one on your own site is riskier than leaving it on Google, so get advice before you reuse any.

How do I reply to a review without naming the patient or condition?

Thank the person, keep it general, and invite them to contact the practice manager offline. Do not confirm they were a patient, name a condition, mention a treatment or defend your care in detail. Have a person read every AI-drafted reply before it goes live.

Should AI-generated images be used on Google Business Profile?

Use real photos of your premises and team wherever you can. Google's photo guidelines say an image should represent reality, so a stock or generated picture of a treatment can mislead. Never use generated images that suggest a patient or a result.

Further reads

• Chiropractic Reactivation: AI-Drafted Messages to Lapsed Patients — Win-back messages that avoid pressure and promises.
• Chiropractic AI for Clinic Owners: Which Admin Jobs to Start With — Admin jobs where AI is safer than marketing copy.
• AI Tools for Chiropractors: Report of Findings, Care Plans and Handouts — Explaining findings to patients clearly with AI help.
• AI Notes for Chiropractors: Consent and Data Checks Before Recording — Consent and data checks before AI touches clinical notes.
• AI Marketing Automation for Doctors: Recall, Reminder and Review Flows — Recall, reminder and review flows for a practice.
• AI Receptionist for a Chiropractic Office: Cost, Recurring Visits — What an AI receptionist costs a chiropractic office.
• Health Food Store Marketing Ideas: AI Posts Without Health Claims — A fact-sheet brief, a five-band claim check and eight post types that need no health claim, so a health food shop can use AI without risky wording.
• Mobile Notary Marketing With AI: Profile, Posts and Review Replies — Profile text, a four-week post calendar and review-reply prompts for mobile notaries, with the rules that keep reviews honest.
• AI Software for Chiropractors: A Data Export Checklist Before You Switch — A chiropractic clinic's checklist for leaving or joining a practice system: export formats, notes ownership, exit terms, a migration test and a cut-over plan.
• AI Marketing for Medical Practices Without Making Health Claims — Service pages, review replies, referrer updates and local search for a medical practice, drafted with AI and kept free of health claims.
• AI for Naturopaths: Intake Summaries and Follow-Up Emails — A naturopath's guide to AI intake summaries and follow-up emails: what to redact, what to keep out, and prompts to copy.
• Jane App for Chiropractors: The AI Scribe and Online Booking — Step-by-step set-up for Jane's AI Scribe (five free notes a month, unlimited add-on) and online booking in a chiropractic clinic.
• AI Contract Review Best Practices for a Private Practice Owner — Use AI as a first read of your lease, supplier and associate agreements: what to redact, which clauses to check, and when a solicitor takes over.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: ASA chiropractic advice page and its 2017 guidance note on sciatica, whiplash and treating babies, children and pregnant women, CAP advice on claims in testimonials and CAP Code rule 12.2 (asa.org.uk), FTC Health Products Compliance Guidance and endorsement guidance (ftc.gov), Google Maps user-generated content policy and Business Profile help on posts and photos. Check the live pages for changes.