Chiropractic Reactivation: AI-Drafted Messages to Lapsed Patients

A chiropractic clinic can use AI to draft a short sequence of two or three messages to lapsed patients: a warm check-in, a gentle second reminder and a final note, each with one easy way to book. Build the list from your own records, contact only patients you are allowed to, and send nothing that claims a clinical need.

Most reactivation fails for two reasons, and neither is the wording. The list is wrong (it includes patients who were discharged, moved away, complained or asked to be left alone), or the message sounds like a sales push to someone who left on good terms. AI is good at the second problem and no help at all with the first, so the order below puts the list before the drafting.

Follow me on Instagram@sagnikteaches

Who counts as lapsed, and who you must leave out

Start with a definition you can run as a filter in your practice-management system. A workable default: the last attended appointment was more than six months ago and nothing is booked. Then set the exclusions, which matter more than the inclusion rule.

Connect on LinkedInSagnik Bhattacharya

• Discharged or completed care. A patient you discharged because they met their goals isn't lapsed; they're done. Contact them with a different message (a general check-in, not a nudge to return) or not at all.
• Referred elsewhere. If you referred a patient for imaging or to another practitioner, someone else is in charge of the next step.
• Complaint, dispute or unpaid account. Route to the clinic owner, never an automated sequence.
• Asked not to be contacted, or opted out. Check the record and your marketing list, and honour the opt-out everywhere.
• Safeguarding flags, children and young people. Contact the parent or carer through the details held for that purpose, and consider whether an automated sequence is right at all.
• No usable contact detail or an out-of-date one. A wrong number sends a message about a patient to a stranger.

Consent to contact: the check before the draft

Electronic-marketing rules treat a message that promotes your services as marketing, even when it goes to an existing patient, and the rules differ by country and by channel. Under the stricter regimes you may only text or email an individual with their specific consent, or under a narrow existing-customer exception: they bought, or asked about buying, a similar service from you, and you gave them a simple way to opt out both when you collected their details and in every message since. That exception never covers new contacts or bought-in lists. Other regimes allow marketing email without prior consent but still require an accurate sender, a working opt-out and prompt removal (one well-known rule sets ten business days). Check the rules where your patients live, and your professional regulator's advertising guidance, which applies on top. Because the line between a care follow-up and promotion can be fine, treat anything encouraging a booking as marketing unless you've taken advice that says otherwise.

Subscribe on YouTube@sagnikteaches

A practical way to do this without a lawyer on call:

1. Add a field or tag to the patient record for "agreed to receive reminders and updates by text/email", with the date and how it was given.
2. Filter your lapsed list down to patients with that tag. The patients without it are not lost; they just get no automated message.
3. Put an opt-out line in every message ("Reply STOP" or an unsubscribe link) and make it work.
4. Write the decision in your data protection records so you can explain it later.

If you aren't sure about your consent records, a confidentiality checklist for small practices helps you see what you hold and what a message may reveal.

From 1,200 records to 190 messages: the list in numbers

Say a clinic with two chiropractors has 1,200 patient records. These figures are illustrative, to show how the list shrinks and what a sequence might return.

The numbers show something owners often miss: the list falls from 420 to 190 before you write a word. If your first attempt contacts all 420, you've messaged around 230 people you shouldn't have. A modest result from 190 messages, say 20 to 30 rebookings, is a realistic win, and the real gain is that nobody was irritated or put at risk. Count it your own way, though: bookings, not replies, because a polite "no thanks" is also a response.

The three-message sequence

Space messages so the sequence takes about four weeks: day 0, day 10 and day 28. Each message does one job, and none mentions a condition, a diagnosis or a treatment.

Message 1: a check-in, not a pitch

Hi [first name], it's [your name] at [clinic]. It's been a while since we saw you and I wanted to check in. If you'd like an appointment, you can book here: [link]. If now isn't the right time, no problem at all. Reply STOP to stop messages.

Message 2: a useful nudge with a practical reason

Hi [first name], [your name] from [clinic] again. We've got appointments free on [days]. If you'd like one, book here: [link] or call [number]. Reply STOP to stop messages.

Message 3: a graceful close

Hi [first name], last message from me. If you'd like to come back to us in future, you're always welcome: [link]. We won't keep messaging. Reply STOP to be taken off our list.

Notice the third message promises the sequence ends. That promise is the point: it tells the patient no pressure is coming, gives anyone who does want to return a natural moment to reply, and spares the clinic the irritation a stronger fourth message would cause.

Using AI to draft in your clinic's voice

AI works best as a drafting aid once you give it your voice, the constraints and the rules above. It should never see patient names or notes. It only needs the template and the style.

Draft three short text messages (under 320 characters each) from a chiropractic clinic to patients who haven't visited for six months. Plain British English, warm and practical, no pressure.
Rules: do not mention any health condition, symptom, treatment, outcome or price. Do not say or imply the patient needs care. Include one booking link placeholder [link] and an opt-out line "Reply STOP to stop messages". Use [first name] and [your name] as placeholders. Message 1 is a check-in; message 2 mentions available appointments; message 3 closes the sequence and says no more messages will follow.
Give me two versions of each, one warmer and one more concise.

Illustrative output usually needs three fixes. First, models like to add claims ("relieve your back pain", "get back to feeling your best"); delete these, because they imply outcomes and health claims of that kind are exactly what advertising rules for health professions restrict. Second, they sometimes add urgency ("limited slots"), which is pressure you don't want. Third, they may drop the opt-out line from one version. Read each message against the rules before anything goes to a patient.

A claims check before anything is approved

Health marketing is held to a stricter standard than most sectors. Have one person check every message, and every later edit, against these questions:

1. Does it name or imply a condition, symptom or result?
2. Does it say or suggest the patient needs treatment?
3. Does it include testimonials, before-and-after claims or numbers about success?
4. Does it use urgency, scarcity or a discount to push a decision?
5. Is the sender clearly identified, and is there a working opt-out?

Any yes means rewrite. If you aren't sure how the rules apply, read your professional regulator's current advertising guidance and the advertising standards body's guidance on health claims, rather than asking an AI model to interpret them for you.

Tone without pressure: before and after

Here are two illustrative rewrites. The first group is the kind of message AI will happily produce if you let it.

• Before: "Don't let your back problems come back! Book now to get on top of your pain before it gets worse." After: "Hi [first name], it's been a while since we saw you. If you'd like an appointment, you can book here: [link]."
• Before: "We've noticed you've missed your last review. This is important for your health." After: "Hi [first name], we've got appointments free next week if you'd like to come in: [link]."

The "before" versions imply a clinical need and a consequence for staying away. Only a clinician who knows the patient should say anything of that kind, in a conversation, not an automated text.

Segment by how long they've been away

One message for everyone ignores the fact that a patient who missed one review is not the same as one who left two years ago. A simple three-way split costs you a few minutes in the filter and noticeably improves the fit of the wording. These groups are illustrative; set your own boundaries from your appointment history.

For the long-gone group, one message is usually plenty, and you should check the contact details very carefully. After a year, phone numbers change hands, and a message about someone's appointments reaching a stranger is a confidentiality problem on its own, even with no health detail in it. A good safeguard is to leave the clinic's name out of a text to a number you haven't verified in a year, or to skip texts for that group and use email only where you have a recent confirmed address.

An email version, with a sample AI draft and what to fix

Email gives you room for a little more warmth and a clear opt-out link. Use the same rules: no condition, no outcome, no urgency. Here is a prompt and an illustrative first draft of the kind a model returns.

Write a short email from a chiropractic clinic to a patient who hasn't visited for over six months. British English, under 90 words, no health claims, no urgency, no discount. Include a subject line, a booking link placeholder [link], a sign-off from [your name], and a one-line footer with an unsubscribe instruction. Offer three subject lines.

Illustrative draft: "Subject: Thinking of you, [first name]. Hi [first name], it's been a while, and we hope you've been keeping well. If you'd like to come back for a check-up and get ahead of any niggles, we have appointments this month: [link]. Warm wishes, [your name]. If you'd prefer not to hear from us, unsubscribe here."

What to fix in that draft, line by line:

• "Come back for a check-up" suggests a clinical need. Change to "come back and see us".
• "Get ahead of any niggles" implies a health benefit and a risk if they don't come. Delete the phrase.
• "We have appointments this month" is fine if true; check it against the diary and name the days.
• The subject line "Thinking of you" can look odd or intrusive to some patients. A plainer subject, such as "A note from [clinic]", is safer and says honestly who the message is from.

After those four edits the email is accurate, calm and complies with the rules you set. Notice that the useful work was the human review, which is why the approval step in the routine above has a named owner.

Capturing consent properly in the first place

The best fix for a thin consent list is to improve how you ask new patients, so that next year's lapsed list is bigger and cleaner. Add one optional question to your intake form, separate from the treatment consent and not pre-ticked: "Would you like us to send you appointment reminders and occasional updates by text or email? You can change your mind at any time." Record the answer with the date, and put the opt-out in every message. Do the same for returning patients when they next attend, at the front desk, in a sentence: "Is it all right if we message you about appointments and check-ins?" It takes ten seconds and the answer goes straight onto the record.

An AI tool can help you word that form question in plain English, because it contains no patient data. Ask for three versions at a reading age of about 12, then pick the one that sounds like your front desk.

What the work costs in staff time

A quick sum, illustrative, for the 190-patient example above. Building the list and applying the exclusions: about 90 minutes the first time, 30 minutes thereafter once the filter is saved. Drafting and approving the three messages: 30 minutes. Sending in two batches of 95 and handling replies over four weeks: roughly five minutes a day. Over four weeks that is around two and a half hours of front-desk time, plus the first-time setup. If those hours produce even a handful of rebookings, the sum works, and the main ongoing cost is attention, not software.

Your practice-management system may already have a recall or messaging feature that does the sending, with the list filter built in. Use it, since it will also log opt-outs automatically. Keep AI for the writing step and the quarterly refresh of wording.

The "personalised for each patient" shortcut that backfires

Picture a clinic, for illustration, that exports its whole patient list into a spreadsheet and asks a chatbot to "write personalised messages for each patient based on their history". The tool produces fluent messages that mention conditions, and one goes to a number that has since been reassigned. Three things go wrong at once: health data has been pasted into an AI tool, the messages name conditions, and the contact details were never verified. The correct process is the one in this tutorial in reverse: filter the list in your own system, send the same non-clinical message to everyone in a segment, and let the template, not the patient record, carry the personalisation (a first name and the sender's name are plenty).

Scripts for the phone when a patient calls back

Some patients will ring rather than click. Give whoever answers a short, honest script so the call stays friendly and doesn't turn into a clinical conversation that belongs with the chiropractor. An illustrative version: "Thanks for getting back to us, [first name]. I can book you in with [practitioner] on [day] or [day]; which suits you? They'll go through what's been happening when you come in." That sentence leaves the clinical discussion where it belongs, and keeps the front desk from giving health advice. If the patient describes symptoms on the phone, the script is to offer the earliest appointment and, where there is any sign of something urgent, to pass the call to a clinician at once.

Where replies go

Decide this before the first message goes out. Replies fall into four groups, and each needs an owner.

• Wants to book. The front desk books and confirms. If you have an AI assistant for booking, see how a clinic can use AI at the front desk safely.
• Stop or unsubscribe. Remove the patient from the marketing list the same day, in every system.
• Clinical question. A clinician replies, by phone or in a secure channel.
• Complaint or upset. The clinic owner replies personally and takes the patient out of any automation.

Counting rebookings and opt-outs after eight weeks

Use only counts you already have in your system. After eight weeks, record: messages sent, opt-outs, appointments booked from the link or reply, and how many of those patients attended. Compare opt-outs with bookings. A high opt-out rate (more than a handful per hundred is worth looking into) suggests the list, the timing or the tone is off, not that the idea is wrong. Run the next batch with a smaller group and change one thing: the gap since last visit, the wording or the send time.

Do not give an AI tool patient-level results to analyse. Aggregate counts are enough, and they keep the exercise free of health data.

A second example: patients on a maintenance plan

Take a patient who used to come every six weeks and stopped after a house move. The message that suits them differs from the one for a patient who finished a short course. Illustratively: "Hi [first name], we haven't seen you for a while and wondered how you were getting on. If you'd like to book a visit, here's the link: [link]." It mentions nothing clinical, but it acknowledges the relationship. Keep these patients as a separate segment so you can test whether the shorter, warmer message performs better than the standard one.

Learning from other clinics' recall work

The mechanics are similar in other practices with regular patients. Winning back lapsed dental patients covers a similar sequence in a dental setting, and cutting missed appointments with AI deals with the next problem you'll meet once patients rebook. For a different trade with the same principle, see winning back lapsed clients as a barber.

A short routine you can keep

1. First week of each month: run the lapsed filter and the exclusion list.
2. Check consent tags and opt-outs; remove anyone without a valid basis.
3. Paste only the template (never patient data) into your AI tool and refresh the wording once a quarter.
4. A named person approves the messages against the claims check.
5. Send in small batches so the front desk can handle the replies the same day.
6. Review counts after eight weeks and write down one change for next time.

Reactivation questions clinic owners ask

How long since the last visit makes a patient lapsed?

There's no standard. Many clinics pick a gap tied to the care plan, such as no visit and nothing booked for six months, then adjust by patient type. A patient finishing a short course of care is different from one on regular maintenance, so segment before you set one number.

Can I text patients who haven't visited for a year?

Only if you have a valid basis for that channel. Messages that promote your services count as marketing under electronic-marketing rules, and those usually need consent or a narrow exemption for existing customers with an opt-out. Check your own consent records and your regulator's current guidance first.

Should the message mention their condition?

No. A text or email can be seen by someone else, and a condition is health data. Keep the message to a general check-in and an easy way to book. Clinical conversation belongs in the appointment or a secure channel.

What if a patient replies with a complaint or a clinical question?

Take it out of the automated sequence at once. Someone on your team should answer personally, and a clinician should handle anything clinical. Never let an AI tool reply to a health question or a complaint.

Further reads

• How Dental Practices Use AI for Recalls and Appointment Reminders — How recalls and reminders work in another clinic setting.
• Telling Patients You Use AI Note-Taking: Consent Wording That Works — Consent wording patterns that translate to patient messaging.
• AI Tools for Physiotherapy Clinics: What Each One Actually Does — What AI tools actually do in a neighbouring treatment clinic.
• 10 Admin Tasks a Small Clinic Can Hand to AI This Month — Other clinic admin jobs that suit AI drafting.
• Marketing a Physio Clinic With AI: Posts, Tips, and Newsletters — Marketing a treatment clinic without overclaiming.
• Patient Reviews and AI Replies: Staying Within Confidentiality — Replying to patients in public without breaching confidence.
• Chiropractic Software With AI: Which Features Are Genuine? — ChiroTouch Rheo, Jane's AI Scribe and ChiroHD checked against their own pages, plus a ten-minute test for separating real AI from marketing.
• AI Marketing for Chiropractors: Claim Limits and Review Replies — Use AI for chiropractic posts and review replies without outcome promises: ASA and FTC limits, a claims register, safe reply wording and prompts.
• Chiropractic AI for Clinic Owners: Which Admin Jobs to Start With — A clinic owner's order of play for AI: bookings and recalls first, notes with review, marketing within claim limits, no AI diagnosis.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: a data protection regulator's published guidance on electronic mail marketing (specific consent, the existing-customer exception and opt-outs in every message) and a consumer protection regulator's commercial email compliance guide (opt-out model, sender identification), both read 4 November 2026; your professional regulator's advertising guidance and your practice-management system's own documentation for filters and messaging.