← Back to blog

The WhatsApp 24-hour window: why clinics lose patients overnight, and how an AI receptionist keeps them

September 2026 · 7 min read · AI receptionist

Bu yazıyı Türkçe oku → · Srpski →

Short answer: the 24-hour window cannot be kept open by any software; WhatsApp closes it. What loses patients is not the rule but the silence inside it. An AI receptionist removes the silence: it replies within seconds, ends every message with something the patient can answer, secures a phone number before the window closes, and delivers any later reply the moment the patient writes again.

Every clinic that sells on WhatsApp eventually discovers the same thing: a patient writes at 22:40, nobody answers until 09:15, and by then the patient has messaged three other clinics. Not because your clinic was worse. Because it was quiet.

Then someone on the team reads about the 24-hour window and the problem gets a name. This post explains what the rule actually is, what it does not allow, and what we built so that a clinic stops losing patients to it.

What the rule actually says

When a customer messages a business on WhatsApp, Instagram or Facebook, the business may reply freely for 24 hours from the customer's last message. Every new message from the customer resets the clock. On WhatsApp a call from the customer resets it too, which most systems ignore.

When the window closes, the business can only send pre-approved template messages — short, categorised texts that the platform has reviewed. A reminder for an existing appointment is fine. A free-form “hi, are you still interested?” is not.

Two details matter more than the rest. First, if the patient arrives from a click-to-WhatsApp ad or the WhatsApp button on your Facebook page, the window is 72 hours instead of 24, and messages inside it are free. Second, a quick-reply button tap counts as a message from the patient and reopens the window; a link button does not.

Why the window closes on clinics, specifically

Clinics lose the window for a boring reason: the reply arrives late, and when it arrives it ends the conversation. We measured this on real conversations before building anything. 38% of replies invited no response at all — they answered the question and stopped. A patient who reads “Yes, we do rhinoplasty. Have a nice day.” has nothing to reply to, so the clock runs out.

The second reason is that nobody is awake. A receptionist works eight hours; the window runs twenty-four.

What an AI receptionist does inside the window

The design goal was never “beat the rule”. It was “never be silent while the window is open, and never lose the patient when it closes”. Five layers, each one deterministic — meaning they run in code, not as a polite request to the language model:

The part that actually keeps patients

Answering fast is table stakes. What moved the needle in our tests was what the receptionist does when a patient starts to leave.

Patient: “I'm not sure yet, I'll think about it.”
Before: “Of course, take your time, write whenever you like.”
After: “Of course. If it helps, I can hold tomorrow at 09:00 for you — your name and phone number are enough.”

The first version closes the conversation itself. The second offers one concrete step and, if the patient still declines, asks for a phone number so the clinic can follow up. No pressure, no discounts it cannot promise, one honest offer. The nearest open slot is calculated from the clinic's real working hours, so a patient asking for Saturday hears “we're closed Saturday; the earliest is Monday 09:00” instead of being booked into an empty building.

What it will not do

It does not invent prices, recovery times or transit directions. When it does not know, it says so and takes the patient's details. It does not diagnose. It does not send marketing to anyone who has not opted in, and a patient who says no once is never messaged again. And it does not pretend the window is open when it is not — your team sees the real state of every conversation in the panel.

Key numbers from our own testing

If your clinic runs on WhatsApp and the night shift is the problem, we can show you the receptionist on your own knowledge base in a 20-minute call. Setup takes a few days; we connect the channels and test with you before a single patient sees it.

Frequently asked questions

What is the WhatsApp 24-hour window?

After a customer messages a business on WhatsApp, the business can reply freely for 24 hours. Once that window closes, only pre-approved template messages can be sent until the customer writes again. The same rule applies to Instagram and Facebook messaging.

Can an AI receptionist keep the 24-hour window open?

No system can prevent the window from closing; the rule is enforced by the platform. What an AI receptionist can do is answer instantly while the window is open, ask a question that invites a reply so the window keeps resetting, capture contact details before it closes, and queue any later reply so it is delivered the moment the customer writes again.

Does the AI receptionist reply in the patient's language?

Yes. It detects the language of the incoming message and replies in it — Turkish, English, German, Arabic and Russian were verified in our testing. On the website widget, the greeting is shown in the visitor's browser language before they type anything.

What happens when a patient asks for a closed day, like Saturday?

It says the clinic is closed that day and offers the nearest open slot, for example “we are closed on Saturday; the earliest slot is Monday 09:00”. The slot is calculated from the clinic's configured working hours, not guessed.

Does it invent medical advice or prices?

No. It answers only from the clinic's own knowledge base. If the information is not there, it says so and captures the patient's details so the team can follow up. Medical assessments are routed to the doctor.

— Written by the Bulsu Labs team · Talk to us