Patient communication for dental clinics: recalls, reminders and the front desk
Where a dental practice's admin time actually goes, which patient messages repay automating, and how to handle clinical and privacy boundaries properly.
Rebecca Pearson11 min read
On this page
A dental practice loses money in two specific places: appointments that do not happen, and patients who quietly stop coming. Both are communication problems, both are largely predictable, and both are handled in most practices by whoever is on the front desk remembering.
That is the case for automating patient communication in a clinic, and it is narrower and more reliable than the general pitch. This page covers which messages repay building, and where the clinical and privacy boundaries sit.
What we'll cover
The two numbers that matter
Failed appointments. A chair sitting empty is revenue that cannot be recovered, because the hour cannot be sold twice. Most practices know their failure rate roughly and few track it by appointment type, time of day or how long ago it was booked, which is where the actionable pattern lives.
Recall attendance. The proportion of patients due for a check-up who actually book. This is usually the larger number and the less examined one, because a patient who drifts away does not generate a gap in the diary that anyone notices. They simply stop appearing.
Worth measuring both before building anything, because the practice-management system probably already reports them and the answer tells you which of the two sections below to start with.
Recalls, the larger opportunity
A recall is a patient who was due six months ago and has not booked. In most practices this is handled by a list somebody works through when there is time, which means it happens in quiet weeks and not in busy ones — the opposite of when you need it.
Make the recall automatic and staged. A first message when due, a second after a fortnight, a third after a month, each different from the last. A single message sent once and never followed up is most of the recall effort wasted.
Make booking possible from the message. A recall telling somebody to ring during working hours, when they are also at work, converts far worse than one with a link. This single change usually matters more than anything about the message text.
Vary by interval. A patient on a six-month recall and one on a two-year interval are different conversations, and the system already knows which is which.
Stop when they book. Obvious, and the most common complaint about automated recalls is receiving a reminder after booking, which makes the practice look disorganised.
Report what happened. How many recalls went, how many booked, over what period. Without that, nobody can tell whether the change worked and the effort gets quietly abandoned.
Reminders and no-shows
The better-understood problem, and there are still details that decide how well it works.
Two reminders beats one. Typically at booking and the day before, with a longer-notice message for appointments booked weeks ahead.
Make rescheduling easy. A patient who cannot attend and cannot easily say so becomes a no-show rather than a cancellation. A cancellation with notice is a slot you can refill; a no-show is not. Making the cancel route as easy as the confirm route feels counterintuitive and converts empty chairs into filled ones.
Match the channel to the patient. A text or WhatsApp message is read; an email frequently is not. Recording each patient's preference and using it is a small change with a measurable effect.
Build the short-notice list. When somebody cancels, the slot is worth offering to patients who wanted an earlier appointment. Maintaining that list by hand is exactly the job that does not happen on a busy morning.
Track which appointments fail. By type, by time, by how far ahead they were booked. Most practices find a specific pattern — a particular slot, or appointments booked more than eight weeks out — and that pattern is actionable in a way the overall rate is not.
Treatment plan follow-up
The most commonly neglected process in a practice, and often the most valuable.
A patient is given a treatment plan, leaves to think about it, and in many practices nothing further happens. Nobody has decided not to follow up; there is simply no mechanism.
Follow up on a schedule you decide in advance. A message a week later, another after a month, phrased as a genuine offer to answer questions rather than as a sales sequence.
Send what they need to decide. The plan, the cost, the options, the finance arrangements if you offer them. Patients decline treatment for reasons that are frequently practical rather than clinical.
Get the unanswered ones in front of a person. After two attempts, a plan that has had no response is worth a phone call from someone who can answer clinical questions, which an automated message cannot.
Never automate the clinical content. The plan itself is a clinical document. The automation should send what the clinician produced and prompt at the right times, and nothing more.
What the front desk actually spends time on
Worth listing, because these are the automatable minutes.
Answering the same questions. Opening hours, parking, whether you take a particular insurance, what to do before an appointment, cost of a check-up. All answerable from a fixed set of information.
Chasing forms. Medical history, consent, updated details. Sending these ahead of the appointment, with a reminder if not completed, removes a queue at the desk and a delay in the surgery.
Relaying messages to clinicians. A patient asks something the receptionist cannot answer, and it becomes a note, a conversation and a callback. Getting that into a routed, tracked form removes the bit that gets forgotten.
Post-treatment check-ins. After an extraction or a longer procedure, a message asking how the patient is doing is clinically useful and reassuring, and it is consistently skipped when the day runs late.
Building these means connecting the practice-management system to a messaging channel with rules about who gets what and when. On CodeWords you describe that in plain language — which patients, which messages, what should stop them, what needs a person — and Cody, the automation builder, builds it, connects it to the systems you already use, and deploys it. Automations connect to more than 3,000 integrations. The free plan covers light use, with Pro at $39 per month and Business at $100 per month as usage grows; details are on the pricing page.
Privacy and clinical boundaries
Dental practices hold health data, which changes what is appropriate. Three distinctions worth keeping firm.
Contact instructions are not clinical advice. Telling a patient when their appointment is, or what to bring, is administrative. Telling them what to do about pain is clinical and belongs with a clinician.
Notifying the practice is not the same as replying to the patient. An automation can route an incoming question to the right person without answering it, which is frequently the correct design for anything clinical.
Passing someone to a person means a real person, available. If a patient's message needs a clinician, the automation should get it to one rather than producing something that reads like an answer.
Beyond that: keep clinical detail out of message content where a generic prompt to contact the practice will do, record consent for the channel you are using, apply your retention policy to message histories as well as to records, and check what your practice-management supplier's terms permit before connecting anything to it.
What not to automate in a clinic
Anything about symptoms or pain. A patient in discomfort needs a person, quickly.
Cancellations initiated by the practice. When you have to move a patient, that is a phone call. An automated message moving somebody's appointment reads badly and generates more work than it saves.
Anything involving a complaint. Route it to a person immediately, with no automated acknowledgement that could read as dismissive.
Clinical results. Whatever the delivery mechanism, a person decides what goes and when.
The first contact after a difficult appointment. Automated check-ins are valuable in general and wrong immediately after something went badly. A suppression rule for those cases is worth building.
Frequently asked questions
What should a dental practice automate first?
Recalls, if the recall attendance rate is below where you want it, because the revenue is larger and the process is usually the least systematic. Reminders first if the failure rate is the more pressing number. Both are worth measuring from the practice-management system before choosing.
Will patients find automated messages impersonal?
Less than practices expect, provided the messages are useful and stop when the patient acts. What patients notice and dislike is a reminder arriving after they have already booked, or a message that clearly ignores something they have told you.
Is WhatsApp appropriate for patient communication?
Many patients prefer it and read it far more reliably than email. The considerations are consent for the channel, keeping clinical detail out of message content, and applying your retention policy to the conversation. Use it for administrative contact and keep clinical discussion in the practice.
How do we handle patients who do not use messaging?
Record the preference and use the channel they want, including post and telephone. An automation can prepare a call list as easily as it sends a message, which keeps the process consistent without forcing a channel on anyone.
Does this replace the receptionist?
It removes the repetitive part of the role — sending reminders, chasing forms, answering the same five questions — and leaves the part patients actually value, which is talking to someone who can help. Practices that automate this generally report the desk feeling calmer rather than smaller.
How do we measure whether it worked?
Failed appointment rate and recall attendance, before and after, from the practice-management system. Both are already reported and both should move within a quarter. If neither does, the messages or the timing need changing rather than more automation.
What about patients who reply to an automated message?
Decide this before you switch anything on, because they will. Replies need to reach a person promptly, and a channel nobody monitors is worse than not offering the channel. Routing replies to whoever is on the desk, with an alert, is the minimum.
Should we offer online booking as well as messaging?
Yes, and they solve different problems. Online booking serves the patient who has decided and wants a slot now. Messaging serves the patient who has a question first, which is most of the ones who have drifted away. A recall that offers only a booking link converts worse than one a patient can simply reply to.
How far ahead should recalls go out?
Starting at the due date rather than before it, in most practices, then staged over the following weeks. Sending ahead of the due date sounds proactive and tends to produce bookings the patient then cancels, because they were not actually ready.
Can this work alongside our practice-management system?
Usually, and how depends on what the supplier exposes. Some have proper APIs, some will produce scheduled exports, and a few offer neither. Where direct integration is unavailable, working from a scheduled export and sending through a separate channel still covers recalls and reminders perfectly well, and those are the two processes carrying nearly all of the value here. Check the supplier's terms before connecting anything to it.
How many messages is too many?
Patients tolerate more than practices fear and less than marketing tools encourage. A booking confirmation, one reminder, and a recall sequence that stops on booking is comfortably within what people accept. The complaints arrive when messages keep coming after the patient has acted, or when a practice starts sending things that are plainly promotional rather than about their care.