How a Palma-based clinic could implement follow-up automation
This is a walkthrough, not a case study. No clinic named here is a real client. It's how we'd actually approach follow-up automation for a small private clinic in Palma, using the same pattern we build for real businesses, so you can see what the work involves before you hire anyone.
The problem, before any tooling
A small clinic (physiotherapy, dental, aesthetic, general private practice) loses patients at three predictable points: the enquiry that never got a reply because reception was with a patient, the appointment that got no-showed because the reminder never went out, and the patient who had a good first visit but never got asked to book a follow-up or leave a review. None of these losses show up as a single dramatic failure. They show up as a slowly declining book of patients that nobody can quite explain.
Follow-up automation, done properly, closes all three gaps without adding headcount.
Where this differs from speed-to-lead
We've written before about speed-to-lead, which is about the first reply to a brand-new enquiry. Follow-up automation picks up after that first contact: everything that happens between "someone showed interest" and "someone becomes a returning patient." For a clinic specifically, that's appointment reminders, no-show recovery, and post-visit follow-up, three distinct sequences that each need their own logic.
Sequence 1: appointment reminders that actually reduce no-shows
A reminder 24 hours before the appointment, and a second one 2 hours before, sent via WhatsApp (the channel most patients in Spain actually read) with a one-tap confirm or reschedule link. If someone doesn't confirm within a set window, the system flags it to reception rather than assuming silence means they're coming. This alone is usually the single most valuable piece of the whole system, because a no-show is a fully wasted appointment slot with real opportunity cost.
Sequence 2: no-show recovery
When someone does no-show, most clinics do nothing, or a receptionist calls when they remember. An automated sequence instead sends a same-day message acknowledging the miss (no guilt-tripping, just a genuine "we noticed, here's how to rebook") with a direct link to reschedule. If there's no response in 48 hours, a second, softer message goes out. If that gets no response either, the patient goes into a longer-term reactivation list rather than being lost entirely.
Sequence 3: post-visit follow-up
A day or two after a visit, an automated check-in asks how the patient is doing (genuinely useful for a physio or dental practice tracking recovery) and, where relevant, prompts booking the next session. For patients where the visit went well and no urgent follow-up is medically needed, the same sequence asks for a review at a natural, non-pushy moment, which is also how the clinic accumulates the social proof that brings in new patients.
The stack we'd use
- n8n as the workflow engine coordinating all three sequences, self-hosted on European servers so patient contact data stays in the EU (this matters more for a clinic than most businesses, given the sensitivity of health-adjacent data).
- The clinic's existing booking system or calendar as the source of truth for appointment times, so reminders trigger automatically off real bookings rather than a duplicated calendar someone has to maintain by hand.
- WhatsApp Business API as the primary channel, since it's where patients in Mallorca actually respond, with email as the fallback for patients without WhatsApp.
- Claude to draft the tone-sensitive messages (no-show recovery, post-visit check-ins) so they read as genuinely considerate rather than templated, while appointment reminders themselves stay simple templated automation since there's no judgment call involved.
Notice that only two of the three sequences actually need AI. Appointment reminders are plain automation: a fixed rule, a fixed time, a fixed message. That's deliberate. Paying for AI-level complexity on a job automation already solves is the most common way this kind of project gets over-scoped and overpriced.
What this looks like in the clinic's first month
- Reminders start firing automatically off the existing calendar. No-shows should visibly drop within the first two to three weeks.
- No-show recovery messages start going out same-day instead of not at all. Reception sees a live list of who's been recovered versus who's gone quiet.
- Post-visit follow-ups start generating a steady trickle of reviews and rebookings that previously depended entirely on the patient remembering to come back.
Nobody on the clinic's team writes a single message by hand for the routine cases. They see the exceptions the system flags, and they still make every judgment call that actually needs a human.
What NOT to do
Don't automate the clinical conversation. Anything touching symptoms, diagnoses, or treatment decisions stays entirely with the clinical staff. Follow-up automation handles logistics and light-touch check-ins, never medical judgment.
Don't skip the opt-out. Every automated sequence needs a clear, one-word way to stop it. Patients who've had a bad experience and want to be left alone should never receive another automated message.
Don't launch all three sequences at once without review. Start with reminders (the simplest, most measurable win), confirm it's working, then layer in no-show recovery and post-visit follow-up once the first sequence is trusted.
The honest summary
Follow-up automation for a clinic isn't exotic. It's three well-scoped sequences, mostly plain automation with AI used only where a genuine judgment call is involved, built on infrastructure that keeps patient data in the EU. The return is straightforward to measure: fewer no-shows, faster no-show recovery, and a steadier stream of reviews and rebookings, all without adding a person to reception.