Playbooks · 5 min read
Healthcare on WhatsApp: The Reminder, and Nothing Else
A clinic's messaging number receives, in one afternoon, a request to reschedule Thursday, a question about parking, a photograph of a blood report, and a message that begins "since morning chest mein" and ends with a question mark.
Three of those are administrative. The fourth is why this post is mostly a list of refusals.
Nothing here is medical advice, and nothing here tells you what any regulator, council or authority currently permits — those rules differ by country and by state, they change, and the only correct source is the one that governs your own practice, read with your own professional and legal advice. What follows is about a messaging channel.
The four refusals
Symptoms. Any message describing what somebody is feeling stops the automation. It does not get triaged, ranked, softened or partially answered. An automated system produces fluent, confident sentences, and it has no way to represent "the consequence of being wrong here is severe". Somebody describing symptoms needs a clinician, and the reply's only job is to say how to reach one and what to do if it is urgent.
Diagnosis. Including the softened forms — "it sounds like", "this is usually", "probably nothing to worry about". Especially those, because they are believed.
Medication. Doses, timings, whether to continue, whether two drugs can be taken together, what to do about a missed one, substitutes. All of it, without exception, from a machine.
Price. Consultation fees, procedure costs, whether something is covered by an insurer. These change, they vary by doctor and by case, and insurance answers depend on a policy nobody in the thread has read. A quoted figure is a figure a patient plans around.
The report arrives as a photograph
This is the healthcare-specific problem, and it is bigger than it first appears.
Patients send images constantly: a lab report, a prescription written by hand, a photo of a rash, a discharge summary, an X-ray held up to a window. It is the natural way to use the channel, and it is how they have always shared these things with family.
Two things follow.
First, nothing in this class of tool reads images. To an automation that message is empty — nothing to match on, nothing to route by. Whatever the picture contains, urgent or routine, the system sees a blank. Any clinic building automation must therefore treat media as an unconditional handoff to a person, not as an unmatched message that falls through to a generic reply.
Second, and more important: that photograph is now in a chat, on a phone, in a business account, possibly in a shared inbox that four receptionists can open. It is health information about an identifiable person, and it arrived without anybody deciding to collect it.
You did not ask for the report. You now hold it anyway, and that is a decision you have to make deliberately rather than by accident.
Decide in advance: who may open the thread, whether images are retained at all, how a patient's request to delete is honoured, and what happens to the phone when a receptionist leaves. India's data protection framework applies to personal data and health data sits at the sensitive end of anybody's definition — what exactly it requires of your practice is a question for your own advisers. The DPDP overview sets out the shape of it; it is not advice and neither are we qualified to give any.
What is left is genuinely useful
Strip out symptoms, diagnosis, medication, prices and reports, and a real and valuable channel remains. It is administrative, and administration is most of what frustrates patients.
Safe to automate
Person, immediately
The appointment reminder is the honest centre of this. It is about a booking the patient made, it reduces no-shows, and it needs no cleverness whatsoever — date, time, doctor, and how to change it.
Two details worth getting right. Reminders should not contain a diagnosis, a department that reveals one, or a procedure name; "your appointment with Dr Rao at 11:15" is enough, and a phone screen is read by whoever is standing nearby. And the confirmation of an appointment should be sent only when it actually exists in the system, never as an automated yes to a request.
The platform, briefly
The WhatsApp Business Platform requires a Meta Business account, verification, and templates Meta reviews. Message categories, approval requirements and pricing are Meta's, vary by country, and change; take current specifics from their documentation on the day you set up.
And a general point that healthcare should take seriously: any vendor claiming their product makes you compliant with a health or data regulation is selling something nobody can sell. A tool can offer mechanics — access control, export, deletion, an audit trail. Compliance is a property of how your practice operates.
What we run today
PostEngage answers Instagram comments and DMs on the official Graph API, and that is the entire product. There is no WhatsApp in it. This post is about the channel because clinics search for it and deserve a straight answer rather than a pitch.
On Instagram the same refusals apply and the clocks are fixed: seven days on a comment, twenty-four hours in a DM thread, restarted only by their own next message. Templated replies are free and unlimited, a credit is spent only when the AI writes new text, the free tier is 100 credits with no card, and packs start at ₹499.
Related: clinics and doctors, dentists, and healthcare on Instagram.
