Instagram DMs are where aesthetic patients start. Most clinics lose them there.

For an aesthetics, dermatology or dental clinic in this region, Instagram is not a marketing channel that feeds a website. It is the whole funnel. A patient sees a before-and-after reel, taps through to the profile, and sends a DM saying “price?”. There is no landing page, no form, no CRM entry — just a message in an inbox that a receptionist checks between patients.

That last hop is where the money goes. This guide is about closing it: what Instagram actually allows you to automate, how the three entry points behave differently, and the specific mistakes that turn interested people into people who messaged three clinics and booked with whichever answered first.

The three entry points, and why they are not the same

Clinics tend to think of “the Instagram inbox” as one thing. Meta treats it as three surfaces with different rules, different urgency and different intent.

Direct messages

The highest-intent surface. Someone who opens a DM has already decided to make contact. The questions are predictable — price, availability, does it hurt, how many sessions — and almost all of them are answerable from written material. This is also where the after-hours problem is worst, because DMs peak in the evening.

Story replies

A story reply lands in the same inbox but carries context that gets lost: the patient is reacting to a specific story, and “how much is this?” means nothing without knowing which story. Any system handling story replies must pull in what the story was, or its answers will be confidently about the wrong treatment.

Comments on posts and reels

Public, and therefore double-edged. A comment asking “price?” is a lead. Answering it publicly with a number is a decision about your pricing strategy and your competitors’ research. The standard pattern is a brief public acknowledgement and a private reply with the detail — which Meta permits once per comment, within a limited window after the comment is posted.

What Meta lets you automate, and what it does not

Instagram messaging through the API follows rules close to WhatsApp’s but not identical, and the differences matter.

The human-agent extension is worth understanding rather than skimming. It exists precisely so that a business can escalate a conversation to a person who is not on shift at that moment and still reply later. For a clinic, it is the difference between “a nurse will look at this in the morning” being possible and being a policy violation.

The questions patients actually send

Pull a month of DMs from any aesthetic clinic and the distribution is remarkably consistent. Roughly in order of volume:

1. 01Price. Usually one word, sometimes with a treatment name, often replying to a specific reel.
2. 02Availability. “Do you have anything this week?” — needs the live calendar, not a document.
3. 03Suitability. “Would this work on my skin type?” — the dangerous one. Informational in tone, clinical in substance.
4. 04Location and hours. Trivially answerable and still often unanswered for hours.
5. 05Downtime and pain. Answerable from aftercare material, and the answer strongly influences whether they book.
6. 06Session count and packages. Answerable if your catalogue is written down properly. Frequently is not.
7. 07Photos of themselves. Asking for an assessment. Never answerable by software.

Four of those seven are pure documentation problems. Two need a live system integration. One must go to a human, always. That ratio is the honest version of “how much can be automated” — and it is high, provided the seventh case is handled properly rather than approximately.

Five mistakes that cost bookings

1. A menu instead of an answer

The classic automation reply is a numbered list: “1 for prices, 2 for booking, 3 to speak to someone”. A patient who typed a specific question and receives a menu has learned that nobody read their message. Answer the question that was asked.

2. Answering in the wrong language

Instagram DMs in this region arrive in Arabic, English, and Franco-Arabic typed in Latin characters with numbers standing in for letters. A system that detects language once and locks it will get the rest of the thread wrong. Detect per message and mirror what the patient used.

3. Public prices in comments

Decide this deliberately and apply it consistently. Whichever you choose, the comment should get a reply — silence under a comment asking “price?” is visible to every other person reading the post.

4. Losing the thread when a human steps in

If your staff answer from the phone app while the automation runs elsewhere, you get two systems replying to the same patient, and eventually they contradict each other. Handoff has to mean the agent stands down and the human continues in the same thread, with the history visible.

5. Treating photos as inputs to an answer

Patients send pictures of their own skin, of scans, of other clinics’ results. A system that describes what it sees and offers an opinion has crossed into clinical territory. The correct behaviour is to route, with the image attached, and say so.

Measuring whether it worked

Follower count and reach tell you nothing about whether the funnel closed. Four numbers do, and all four are available from the inbox rather than from Instagram’s analytics.

• Median first response time, split by hour of day. The gap between your 2pm number and your 11pm number is the size of the problem.
• Answered-without-escalation rate, which should rise as the knowledge base improves and then plateau. If it approaches 100%, something is being automated that should not be.
• DM-to-booking rate, measured on conversations rather than on people.
• Escalation response time — how long a patient waits after a handoff. This is the number that quietly destroys trust when the automation is fast and the humans are not.

We go through the full set, and what a healthy range looks like, in the five numbers worth tracking on a clinic front desk.