What clinics shouldn't type in chat, even when the customer asks

Most clinics are careful with the words they use in posts and ads, because they know someone is checking. But in the chat inbox, where conversations are one-to-one, that care tends to disappear — even though chat replies are the clinic speaking too, and they're the easiest messages to screenshot and pass around.

Written from our work setting up and testing chat systems for healthcare providers. This is not legal advice and has not been reviewed by lawyers or external auditors. On legal matters, follow the regulators listed in the sources at the end of this article.

This didn't start with AI answering chats. Human admins slip in exactly the same places, because the pressure in a chat is different from writing a post — the customer asks straight out, "Will it go away?", and waits. The urge to close the sale ends in a sentence that shouldn't have been typed.

Why chat replies count

Healthcare advertising in Thailand is governed by rules enforced by the Department of Health Service Support (HSS) under the Ministry of Public Health. "Advertising" isn't limited to signs, posts or paid ads — it covers any communication that leads people to decide to use a service. Replies typed in LINE or a page inbox fall into the same group.

In practice, the trouble clinics actually run into rarely comes from proactive inspections. It comes from cases where the result wasn't what was discussed, and the patient attaches a screenshot of the admin typing "I guarantee it'll go away" to their complaint. One sentence in a chat becomes the hardest piece of evidence to answer.

Six wording traps clinics fall into most

1. Guaranteeing results

"It'll be gone for good." "Results guaranteed." "No results, your money back." This group is the most dangerous because it sounds like confidence in your own work. But the outcome of a medical procedure depends on each patient, so a promise made in advance is one you can't back up from the start.

2. Claiming to be the best or the only one

"The best in the area." "The only clinic in Thailand with this machine." "Number one for clear aligners." Superlative comparisons need verifiable data behind them. Without it, it's boasting — whether it's typed in a post or in a chat.

3. Saying it's safe with no side effects

"No side effects at all." "100% safe." These tend to slip out when a customer hesitates, yet every procedure carries its own risks. What you can talk about is the standard your clinic works to — who performs the procedure, what equipment is used, and how aftercare works.

4. Discounts and giveaways that push people to decide fast

Running a promotion isn't wrong in itself. But rushing a treatment decision with retail tactics — "Only 3 slots left, book before they're gone" — for a procedure that needs a doctor's assessment first is easy to single out, because it gets people deciding out of urgency instead of information.

5. Using reviews or other patients' words as proof

"Our last customer said it cleared up within a week." Citing other patients' experiences as proof of results is something healthcare advertising rules restrict — and in reality it doesn't mean the person you're talking to will get the same result.

6. Diagnosing or prescribing treatment over chat

"From the photo it looks like hormonal acne. Go for this program." This goes beyond advertising wording, because it's assessing a condition without an examination. What you can do is explain what kind of case each program is for, and invite the person in to see a doctor.

How to rewrite so it still sells

A common misconception is that cutting these phrases means you can't close. In reality, the safer sentence usually carries more information — and patients who decide based on real information cancel less, too.

If a system answers for you, what needs to be set

The upside of letting a system answer is that it has no urge to close, so it won't blurt out a hard-sell line the way a person might after being asked the same thing again late at night. But the lines have to be set from day one — not left for it to figure out.

• No promising results — never "gone for good", "guaranteed results" or how many years it will last
• No claiming to be better than others — only say what the clinic actually does
• No saying there are no side effects — pass on the doctor's explanation instead
• No diagnosing or calling it from a photo — if the clinic allows initial guidance, it must follow the doctor's criteria only and always lead to an in-person appointment
• No discounts or freebies on its own — negotiation goes to your team to decide
• On symptoms or side effects, stop and bring in a person — and alert your team immediately

The first five are about wording; the last is about timing. And the last is exactly what most systems do badly, because they're built to keep answering as much as possible, not to know when to stop.

Three checks before going live

1. Ask questions that bait a promise — "Will it go away?" "Can you guarantee it?" See whether the system sidesteps or slips into a promise.
2. Send in a photo — see whether the system diagnoses on the spot or points the person to a doctor.
3. Ask for a discount — see whether the system gives one or passes it to your team.

These three work for testing a system you're about to buy and for testing a newly hired admin. For the full test set to run before launch, read How far should AI answer your clinic's chat? And if you're comparing which kind of system can really control what it says, see the types of clinic chatbots.

Sources and further reading

1. Department of Health Service Support (HSS) — Ministry of Public HealthThailand's regulator for private healthcare facilities, including the rules on healthcare advertising — check the current version here before using new wording (in Thai)