The owner of a med spa sat across from me and said something I've heard a dozen times: "We're great at getting people in the door. We're terrible at getting them to come back."
She wasn't wrong. Her clinic did Botox, fillers, laser treatments, body contouring — services where repeat visits are the entire business model. A single Botox patient who comes back every three months is worth $2,400 a year. But her rebooking rate was hovering around 30%. That meant 70% of her patients were one-and-done, not because they were unhappy, but because nobody followed up.
Her staff was already stretched thin. Two front desk people handling check-in, check-out, scheduling, phone calls, and insurance — they didn't have bandwidth to run a follow-up program. She'd tried automated email blasts through her CRM. Open rates were around 12%. Basically useless.
So we built something better.
The Problem in Detail
Med spas have a unique follow-up challenge. Different treatments require different follow-up timelines. Botox patients should rebook at 10-12 weeks. Filler patients at 6-9 months. Laser patients need a series of sessions spaced 4-6 weeks apart. Body contouring might need 2-3 sessions over 8 weeks, then maintenance every 3-6 months.
A generic "Hey, time for your next visit!" email doesn't cut it. Patients want to feel like their provider remembers them, knows their treatment plan, and is reaching out with relevant information — not spam.
The old system was: front desk person checks the schedule from 3 months ago, finds patients due for follow-up, tries to call them between handling walk-ins. On a good day, they'd make maybe 10 calls. Most went to voicemail.
How We Built It
The system runs on three components:
1. n8n workflow automation — This is the brain. n8n monitors the clinic's scheduling system and triggers follow-up sequences based on treatment type and date of last visit.
2. Claude API — This generates personalized follow-up messages. Not templates — actual personalized messages based on the patient's treatment history, preferences, and timing.
3. Twilio for SMS and email delivery — Because texts get opened. Email open rates were 12%. Text open rates hit 94% in the first month.
Here's how the workflow runs:
Every morning at 7 AM, n8n queries the scheduling system for patients who are approaching their follow-up window. It pulls their treatment history, any notes from their last visit, and their communication preferences.
Then it passes that context to Claude with a prompt like: "Generate a friendly, professional follow-up message for a patient named Sarah who had Botox in her forehead and crow's feet 11 weeks ago. She mentioned at her last visit that she was going to a wedding in April. Keep it under 160 characters for SMS."
Claude generates something like: "Hi Sarah! Your Botox from January is probably starting to fade — want to freshen up before that April wedding? We have openings this week 💫"
That message goes out via Twilio. If the patient responds, their reply goes into the clinic's existing communication system where staff can book the appointment. We also built a quick-book link that lets patients confirm a suggested time slot with one tap.
The entire build took about six hours. Two hours designing the workflow logic and treatment-specific timing rules. Two hours building the Claude prompts and testing message quality. One hour on Twilio integration and delivery. One hour on testing and edge cases.
The Results
First month numbers:
- Follow-up messages sent: 340 - Response rate: 47% (compared to 12% email open rate previously) - Rebooked appointments from follow-up: 89 - Revenue recovered: Approximately $22,000 - Staff time saved on follow-up: ~20 hours per month - Rebooking rate: Jumped from 30% to 52%
By month three, the rebooking rate stabilized at 58%. The system was generating roughly $25,000 per month in rebookings that would have otherwise been lost.
What Didn't Work (At First)
The first version of the messages was too clinical. Claude defaulted to a professional medical tone — "Dear Patient, your treatment interval suggests scheduling a follow-up..." We had to explicitly prompt for warmth and personality. Med spa patients expect a vibe closer to a high-end salon than a hospital.
We also learned that timing matters enormously. Messages sent before 9 AM or after 7 PM got significantly lower response rates. We narrowed the delivery window to 10 AM - 5 PM and saw a 15% bump.
One tricky edge case: patients who had complications or weren't happy with results. The system needed to know not to cheerfully suggest a rebook if a patient had filed a complaint or requested a refund. We added a flag in the scheduling system that excludes certain patients from automated follow-up.
Lessons Learned
1. SMS beats email for patient communication by a mile. It's not even close. If you're still relying on email for patient follow-up, you're leaving money on the table. 2. Personalization isn't optional. Generic messages get ignored. A message that references the specific treatment, timing, and something personal from the last visit gets a response. 3. AI-generated doesn't mean AI-sounding. The best messages from Claude sound like they were written by a friendly front desk person. If a patient can tell it's automated, you've failed. 4. Exclusion logic is as important as inclusion logic. Knowing who NOT to message is just as critical as knowing who to reach out to. 5. Start with the highest-value treatments. We launched with Botox follow-up first because the volume was highest and the rebooking window was most predictable. Prove it works on one treatment, then expand.
FAQ
Do patients know they're talking to an AI? The messages come from the clinic's name and phone number. They read like a message from the front desk. If a patient replies with a question, it goes to a human staff member. We don't try to maintain the conversation via AI — the goal is to initiate contact, not replace the relationship.
How do you handle patients who opt out? Standard opt-out via "STOP" reply, which Twilio handles automatically. We also honor do-not-contact preferences in the scheduling system. Opt-out rate was under 3%, which is well below industry average for healthcare SMS.
What about appointment reminders — is this the same thing? No. Appointment reminders are for already-booked visits. This system targets patients who don't have anything booked and are approaching their ideal follow-up window. They're complementary — most clinics need both.
Can this work for other healthcare verticals? Absolutely. We've since adapted the same system for dental practices, dermatology clinics, and physical therapy offices. Any practice where repeat visits drive revenue can benefit from intelligent automated follow-up.
What does it cost to run monthly? About $80-120 per month in API and Twilio costs for a clinic doing 300-400 follow-up messages. The ROI is absurd — $100 in costs generating $25,000 in recovered revenue.
Stop losing patients after their first visit.
If your rebooking rate is below 50%, you don't have a marketing problem — you have a follow-up problem. We build AI-powered follow-up systems that bring patients back automatically, without adding work to your staff's plate.
Start with a Strategy Audit and we'll show you exactly how much revenue you're leaving on the table.