DM triage: routing consultation enquiries to the booking desk
The Instagram DM automation doctors need is modest: acknowledge every message instantly, collect the minimum booking details, and hand the person to a human at the booking desk, usually via WhatsApp. Automation must never interpret symptoms or reports. Follow Meta’s and WhatsApp’s messaging rules, collect as little data as possible, and design for DPDP consent from day one.
For many doctors with an active Instagram account, the DM inbox has quietly become an appointment channel that nobody staffs. Requests for a slot sit next to lab reports nobody asked for, collaboration pitches and spam, and genuine enquiries wait until the doctor scrolls through them late at night. This piece covers the Instagram DM automation doctors can actually use, and, more importantly, the triage design and handoff that make automation safe.
The goal is easy to state. Every consultation enquiry reaches a human at the booking desk quickly, with enough context to book, and nothing clinical gets answered in a DM. Platform rules, WhatsApp policy and data protection law set the boundaries. It is one piece of a larger system described in personal branding for doctors in India.
What actually lands in a doctor’s DMs
Before building anything, read a few weeks of the inbox and sort it. The mix is usually some version of this:
- Booking intent: “how do I get an appointment”, “are you available on Saturday”, “do you see patients in Nagpur”.
- Logistics: fees, location, which hospital, cashless or TPA, whether video consultations are offered.
- Personal medical questions, often long, often with photos or reports attached.
- Existing patients: follow-up questions, report queries, prescription doubts.
- Urgent or distressing messages.
- Media, brand and collaboration pitches.
- Spam, scams and fake-account outreach.
Only the first two are clean booking work. The third and fourth look like booking work but carry the most risk, because the sender expects the doctor to reply personally with an opinion. The design has to handle them without ignoring the person and without advising by DM.
A three-lane triage model
Every conversation should leave the inbox in one of three lanes within a defined time. Keep it this simple, because complexity is where messages get lost.
- Book. The person wants a consultation or is asking about a procedure. Collect the minimum needed to book (name, phone, city, preferred location or video, the broad reason in their own words) and hand it to the booking desk.
- Route. The message belongs elsewhere. Existing-patient queries go to the clinic’s or hospital’s patient line, complaints go to patient relations, media goes to the PR contact or the doctor’s office, and urgent messages get emergency guidance at once plus a human alert.
- Close. Spam, scams, abuse and pitches you will not take. Archive, block or report.
One rule applies across all three lanes: nobody in the chain, human or bot, answers the clinical content of a message. “I can’t advise on this over chat, but here is how to see the doctor” is the whole clinical policy.
What Instagram DM automation doctors can use today
Native tools inside the app
A professional account (business or creator) already gets useful tools without third-party software: saved replies for common answers, suggested questions people can tap when they open a chat, and inbox folders that separate primary conversations, general ones and message requests. Meta’s business inbox also brings Instagram, Facebook and WhatsApp conversations into one view, with basic automated replies such as instant replies and away messages.
For a solo doctor with modest volume, this is often enough. An instant reply that sets expectations removes much of the anxiety that makes people send the same message five times: “Thanks for writing to Dr A. Sharma’s clinic (example). We reply between 9am and 7pm, Monday to Saturday. To book faster, message our desk on WhatsApp using the link in our bio.”
The messaging API and automation vendors
Anything beyond that runs on Meta’s Instagram messaging API, usually through a third-party automation or CRM vendor. Meta’s Instagram Platform documentation says apps can respond to messages sent to a professional account within 24 hours, and that a human agent tag allows a response within 7 days when a person needs more time. The same documentation covers private replies to comments, which is what powers “comment a keyword and we’ll message you” flows.
Meta’s Messenger Platform and Instagram Messaging API policy adds a rule every clinic should build in: automated chat experiences must disclose that the person is interacting with an automated service at the start of a conversation. Put that disclosure in your own first message rather than assuming the vendor has done it.
Before signing a vendor, ask where conversation data is stored, who at the vendor can read it, how long it is kept and whether you can delete a thread on request. The healthcare chatbot RFP checklist is a good starting set of questions, even for a small deployment.
Where automation stops and a human takes over
Automation is good at three things in a doctor’s inbox: acknowledging instantly, asking the same structured questions every time, and routing. It should do nothing else.
- It should not interpret symptoms, rank urgency clinically, or suggest which doctor or test someone needs based on their description.
- It should not ask for reports, photos or a detailed history. If someone sends them anyway, the reply should not comment on them.
- It should recognise a short list of distress and emergency phrases, agreed with the doctor and written in English and the regional languages you receive, and respond with emergency guidance plus an immediate human alert. Keyword lists miss things, so people still read the inbox.
- It should always show a visible way to reach a person, and honour it quickly.
This is the same line I draw for hospital chatbots in using AI to triage enquiries without making clinical calls: triage by intent and logistics, never by clinical judgement. The most dangerous bot reply is the kind one: a reassuring line about a symptom is still clinical advice, however gently it is worded.
Handing off to the booking desk
The weakest point in most setups is not the bot. It is the handoff. The enquiry gets captured in an Instagram thread and then waits for someone to copy it into whatever the booking desk uses.
Move the conversation to where booking happens
For most Indian clinics and hospitals, booking happens on the phone or on WhatsApp, so the triage reply should move the person there in one tap. WhatsApp click-to-chat links (wa.me followed by the number) can carry prefilled text, which lets you put a short source tag such as “IG-DM” in the first message. The desk then knows where the enquiry started, and so do your reports.
WhatsApp has its own rules. The WhatsApp Business Messaging Policy says you may contact people only if they have given you their number and opted in to receive messages from you, that outside a 24-hour customer service window you may send only approved message templates, and that automated flows must offer prompt, clear escalation to a human. It also tells businesses not to send or request health information where applicable regulations require heightened protection for it. The platform detail is in what the WhatsApp Business API means for hospitals, and the broader argument in WhatsApp as a hospital’s front door.
What the booking desk should receive
- Name and a callback number.
- City, and preferred location or video consultation.
- The broad reason in the person’s own words, not a symptom history.
- Preferred doctor and time window.
- Source: Instagram DM, and the post or Reel if known.
- Whether the person has agreed to follow-up messages.
- The time the enquiry arrived, so response time can be measured.
Agree a response window with the desk and write it into the instant reply, so the promise and the practice match. If messages arrive after the desk closes, the away reply should say when a person will respond and give the emergency line.
Consent, DPDP and what not to collect
A DM enquiry is personal data from the first message, and anything describing symptoms is sensitive enough to deserve extra care. The Digital Personal Data Protection Act, 2023 requires a lawful basis for processing personal data, usually consent given against a clear notice, and the DPDP Rules, 2025 were notified in November 2025 with a phased 18-month window for organisations to comply. There has been public discussion of shortening parts of that window, so check the current position. Either way, build for these obligations now rather than retrofitting later.
- Collect only what is needed to book. The desk needs a name, number and broad reason, not a medical history.
- Tell people, in the first automated message or a linked notice, what their details will be used for and who will contact them.
- Ask separately before adding anyone to marketing or follow-up broadcasts. A booking enquiry is not consent for newsletters.
- Limit who can read the inbox, remove access the day someone leaves, and do not let staff or agency teams forward screenshots of patient messages.
- Set a retention rule for DM threads and vendor logs, and a way to delete on request.
- Put automation and CRM vendors under written agreements that cover how they process data on your behalf.
The consent mechanics for hospital marketing are covered in consent under DPDP for hospital marketing. This is not legal advice; take your own counsel on how the Act and Rules apply to your practice.
Example scripts for the first replies
These are illustrative scripts for a fictional practice. Have your doctor and clinical reviewer approve your own versions, in every language you receive messages in.
- First automated response (example): “Hi, this is an automated assistant for Dr A. Sharma’s clinic (example). I can help you book a consultation or find clinic details. I can’t answer medical questions here. Reply TEAM at any time to reach a person.”
- Booking lane (example): “To book, please share your name, city, and whether you’d like to visit in person or consult by video. Or tap the link to continue on WhatsApp with our booking desk.”
- Medical question received (example): “Thank you for explaining. The doctor needs to see you properly before advising, so we can’t answer this over chat. We can book a consultation for you. Would you like a call back?”
- Urgent language detected (example): “If you or someone with you needs urgent help, please call your local emergency number or go to the nearest emergency department now. A member of our team has also been alerted.”
- Existing patient (example): “For questions about your ongoing treatment, please call the clinic’s patient line at [official number] so the team can check your records properly.”
The DM triage and response script kit has a fuller set, including replies for reports sent unasked, fee questions and media requests.
Staffing, hours and a launch checklist
Automation does not remove the need for people; it changes when they are needed. Someone has to own the inbox every working day, clear the human-handoff queue and watch for messages the keywords missed. For a solo practice that is usually the front office; in a hospital group it is the contact centre or the digital team under a clear service level.
- Switch to a professional account and connect it to Meta’s business tools.
- Write the three lanes, the scripts and the emergency wording, and get the doctor’s sign-off.
- Set instant and away replies that state hours and the booking route.
- Create a WhatsApp click-to-chat link with a source tag, plus a phone fallback.
- Decide whether volume justifies an API vendor; if it does, run the data-handling questions before signing.
- Brief the booking desk on the fields they will receive and the response window.
- Name an inbox owner and a backup for leave, weekends and evenings.
- Test every path with a colleague posing as a patient, including an emergency phrase in a regional language.
- Review a sample of conversations every week for the first month, then monthly.
Measuring whether triage works
Keep measurement close to the booking desk rather than the platform dashboard. Five numbers tell you most of what you need: time to first human response, the share of booking-lane conversations that became booked appointments, booked appointments that were honoured, messages that waited past the promised window, and any conversation where something clinical was answered or an emergency was missed. The last one should be reviewed individually, every time.
Public comments feed the same inbox, so align this with comment moderation for doctors, where medical questions are routed into DMs or WhatsApp. And when you want to put a value on the whole channel, measuring a doctor’s social media ROI shows how to carry the source tag through to honoured appointments. Done well, the Instagram DM automation doctors set up becomes invisible: patients simply find it easy to book, and nobody gets advice by chat.
Questions people ask
It means using Instagram’s built-in tools, or Meta’s messaging API through a vendor, to acknowledge messages instantly, ask the same booking questions every time and route each conversation to the right place. For a doctor, its only jobs are acknowledging, collecting booking details and handing over to a human. It must never interpret symptoms, comment on reports or give anything that reads as medical advice.
Meta permits automated replies through its messaging tools and API, subject to its policies, including disclosing at the start of a conversation that the person is talking to an automated service. Professional rules on medical communication still apply to what the bot says, so keep it to logistics and routing. Check the current platform policy before launch, and take your own legal advice on your specific setup.
No. There is no such thing as a simple medical question over chat, because the bot has no history, examination or context. Anything about symptoms, medicines, doses or reports should receive a polite script saying it cannot be answered in chat, with an offer to book a consultation. Distress or emergency language should trigger emergency guidance and an immediate alert to a human.
The native Instagram and Meta business inbox tools cost nothing to switch on, so a small practice can start with staff time alone: writing scripts, getting sign-off and running the inbox. An automation or CRM vendor adds a subscription and some set-up effort. The larger, ongoing cost is people to handle the human-handoff queue, which automation reduces but never removes.
A basic setup with instant replies, saved replies, a WhatsApp link and approved scripts can be ready once the doctor has signed off the wording. An API-based flow takes longer because of vendor selection, data-handling checks, testing and staff briefing. Plan a first month of close weekly review before you trust it, whatever route you choose.
A single clinic with one desk and moderate volume can usually work well on the WhatsApp Business app. The API becomes worth it when several people need to answer from one number, when you want CRM integration, or when you need templates and reporting at scale. Either way, WhatsApp’s opt-in and messaging rules apply, and health information needs careful handling.
Do not comment on them, even to reassure. Reply with the approved script that thanks the person, explains the doctor needs a proper consultation, and offers to book. Decide in advance whether staff should delete such threads after the booking is made, and who may view them. Never forward patient reports or screenshots on personal phones or to agency teams.
Every DM enquiry involves personal data, so you need a clear notice of purpose, consent where required, minimal collection, access controls, retention rules and agreements with vendors that process data for you. The DPDP Rules notified in November 2025 set a phased compliance window, so build now. Health details deserve extra care. This is not legal advice; get counsel on your own setup.
Whoever owns booking should own the inbox, not the doctor and not a content agency on its own. In a private practice that is usually the front office; in a hospital group, the contact centre or digital team with a written service level. The doctor approves scripts and handles escalations. The owner needs a named backup for leave, weekends and evenings.
Set an away reply that says when a person will respond, gives the booking route and points to emergency care for urgent situations. Keep a short list of distress phrases that alert an on-call contact even out of hours, and agree who that contact is. Clear the overnight queue first thing, oldest first, and track how long those messages waited.
Ask where conversation data is stored, who at the vendor can read it, how long logs are kept and how deletion requests are handled. Ask how the bot discloses it is automated, how human handover works, whether escalation keywords can be set in regional languages, and how it hands records to your CRM or booking desk. Get data-processing terms in writing.
Track time to first human response, the share of booking-lane conversations that became appointments, how many of those were honoured, and messages that waited beyond the promised window. Review every case where something clinical was answered or an emergency was missed. Compare month on month rather than chasing a single target, and read a sample of actual conversations regularly.
Yes, if the group standardises the lanes, scripts, escalation rules and data handling, and lets each doctor’s account route into a central booking desk with the doctor’s name and source attached. Doctor-specific details such as clinic days and locations sit in a shared knowledge base. Governance matters more than software here: one policy, one owner and regular clinical review of scripts.
Read my takes first in Google Search
