A patient calls at 7:30 pm and the reception desk is already closed. At best they hear silence or an automated message, at worst they never call again. For a dental clinic, every missed call is a potential patient who will find another clinic. The problem is especially acute in dentistry, where a new patient is not just one visit but a long-term relationship plus referrals to family and friends.
On this page you will find: four after-hours service models with specific criteria, a formula to calculate the cost of missed calls using your own numbers, legal requirements (AI Act, GDPR, Electronic Communications Law), and four concrete steps to check how your clinic handles calls after closing.
Who calls after the clinic closes
After-hours calls can be divided into four groups. Each has a different level of urgency and different consequences if nobody answers.
Acute pain, swelling, fever — such reports shouldn't wait until morning; what to do with them is set by the clinic. A tooth hurts so much the patient can't sleep, has swelling or fever. If nobody answers, the patient seeks help elsewhere, and the clinic loses the chance to provide help and build loyalty.
New registration — desire to book an appointment. The patient is looking for a dentist, calls after hours because they're at work during the day. If you don't answer — they probably call a competitor. The loss isn't immediate — some of these patients don't call back if the first contact fails.
Rescheduling or cancelling an appointment. The patient wants to change the time or cancel. Without contact, the clinic loses the doctor's time — a free slot that could have gone to another patient.
Question about price or range of services. The patient wants to know how much whitening, an implant, or root canal costs. Without an answer — they most likely search elsewhere or choose the clinic that answered first.
Each group requires a different approach. Acute pain needs immediate connection with a person who can assess the situation. Registration and questions can wait until morning, but it's better if the patient can leave a message or book an appointment themselves.
Priority handling — how to decide
The simplest rule: which calls go directly to a person is set by the clinic. The system doesn't assess health condition. If they're asking about an appointment — can wait until morning, but better if they can book themselves through the calendar. The rest (price questions, cancellations) — standard queue.
How much a missed call costs your clinic — formula for your own numbers
You can't calculate the exact loss without specific data, but you can estimate your own cost. The formula is simple:
losses_per_month = after_hours_calls × share_of_new_patients × share_who_book_elsewhere
Example on conditional numbers — substitute your own: the clinic receives 300 calls per month, of which 40 fall after hours (between 6 pm and 8 am the next day). Of those 40 calls, half are new patients (20).
Loss = 20 × 0.4 = 8 new patients per month.
This isn't statistics — it's a mechanism. The more after-hours calls, the bigger the loss. Calculating exactly with your own numbers lets you decide if the problem is worth investing in. More about calculating the cost of missed calls is in our article How much do missed calls cost in a company.
Four after-hours service models
Every clinic faces the choice: who answers phones when reception isn't working. There are several options, each with different consequences for costs, patient data, and legal risk.
| Model | Availability | Who sees patient data | Typical problem |
|---|---|---|---|
| Automated recording | 24/7 | None | Patient can't leave a message or book; you don't know who called |
| Forwarding to doctor's/owner's phone | Depends on availability | Doctor/owner | No boundaries between private life and work; risk of missing calls; no call documentation |
| External medical secretariat (call centre) | Per contract | Call centre + clinic | Flat fee or per-minute cost; sensitive data with third party; limited knowledge of clinic specifics |
| AI receptionist with online calendar | 24/7 | System + clinic | Requires calendar integration; struggles with unusual situations; must clearly inform patient they're talking to a machine |
The automated recording is cheapest but practically doesn't solve the problem — the patient has no way to register themselves. Forwarding to the doctor's phone is the most expensive in terms of life quality and burnout risk. A call centre adds cost and a third party to data processing. AI reception requires a calendar and clear rules but works around the clock.
Cost vs quality — what to choose
When choosing a model, it's worth looking not only at the price but at the total cost for the clinic. The value of a new patient over time is significant — it's not just the first visit, but also treatment and referrals to acquaintances. Even a small number of missed calls can mean real loss of revenue.
More about how AI systems work in customer service can be found in the article AI agent vs chatbot: the difference visible in three process characteristics.
What the system does when a patient reports acute pain at night
Automation cannot and should not replace medical assessment. If a patient reports acute pain, swelling, fever, or bleeding, the system doesn't diagnose — it tells them where to seek help according to the clinic's rules.
What the system says in such a situation is set by the clinic in the scenario — e.g., provides the emergency number according to clinic rules or transfers to the on-duty doctor if the clinic has one.
What the system doesn't do
The system doesn't assess the patient's health condition. It doesn't advise on treatment. It doesn't say "take ibuprofen and wait until morning." It doesn't suggest the problem isn't urgent. If the patient says "it hurts a lot" or "I have swelling" — the system proceeds according to the clinic's scenario. This isn't a technological limitation — it's a conscious limitation to avoid exposing the clinic to liability for incorrect medical advice.
More about the limits of automation in healthcare can be found in the article Query handling automation: one queue instead of five inboxes.
The calendar as the only source of truth — confirmation chain
There's no automation without a good calendar. Double bookings, manual windows in the doctor's schedule, appointment changes — all of this causes the registration system to show slots that don't exist in practice. Result: the patient arrives and the doctor already has another patient.
The confirmation chain works like this:
- 01Appointment booking
- →02SMS with confirmation
- →03Reminder day before the visit
- →04Patient response
Each step is a separate automation. Just putting a slot in the calendar isn't enough — you need confirmation from the patient. Without it, the clinic doesn't know if the patient will actually show up.
Patient data — what the system knows and who has access
During after-hours phone registration, the system collects minimum data: name, phone number, preferred appointment time, reason for visit (free text). That's basic data. Full medical documentation is created only on-site, during the appointment.
Where the call recording is stored: if the system records it — there must be a legal basis. UODO imposed a PLN 10,000 fine on an entity that recorded sound in a monitoring system without a legal basis (the case concerned monitoring, not a helpline, but the principle is the same — voice recording requires a legal basis and clear information for the person). More about data protection in automation can be found in the article Automation and GDPR: where your customer data physically ends up.
Who has access: only persons authorized by the clinic. In the case of AI reception — access to transcripts and recordings should be limited to the system administrator and the person responsible for patient service. There's no reason for dentists to listen to all conversations — unless you conduct systematic quality audits.
AI Act since August 2, 2026 — what you need to know
Since August 2, 2026, AI Act transparency requirements apply. A person using an AI system must be clearly informed that they are interacting with a machine if it is not obvious — informs the Ministry of Digital Affairs. This applies to chatbots, virtual assistants, and automatic user service systems — exactly what an AI receptionist in a clinic is.
How it should sound in the first phrase of the call: "Good afternoon, this is the automatic registration system for [clinic name]. Can I help you book an appointment?" Clear, simple, in Polish (or the language of the country). The patient knows from the first second that they're talking to a system.
If the system is integrated with website chat — there must be a clear indication "This chat is AI-powered." If it takes calls — the first phrase speaks for itself. You don't need to repeat it in every sentence, but the beginning of the call must be unambiguous.
Appointment reminder vs marketing — the boundary in Electronic Communications Law
The Electronic Communications Law (Article 398) prohibits using automatic calling systems and telecommunication terminal equipment for direct marketing purposes without the prior consent of the subscriber. An appointment reminder is usually treated as a service message related to the service.
Example usually treated as service: SMS "reminding you of your appointment on Thursday at 10 am". Example of promotional content requiring consent: SMS "a new year is approaching, we invite you for a check-up". The boundary in a specific case will be assessed by a lawyer or DPO.
More about what happens to an inquiry after the first conversation can be found in the article Follow-up automation: what happens to the inquiry after the first conversation.
What the patient sees before they call — results of our audit
Before calling, the patient sees three things in Google: phone number, opening hours, and a contact form (if it exists). Whether these elements work determines whether a conversation happens at all.
Where these numbers come from: we analyzed public websites of Warsaw businesses in this industry (July 2026); the share is calculated from the pages that could be opened.
Among 259 Warsaw dental clinics whose websites we analyzed, 226 pages opened.
This means a patient who prefers writing to calling will either give up or choose a clinic with a form.
15 clinics out of 259 registered exclusively through a platform like Booksy or another external system, without their own website with contact details. Another 13 had a domain entered in Google but the website didn't exist or was unavailable.
These numbers aren't national statistics — it's a specific measurement in one city. But the mechanism is universal: the fewer entry points, the fewer patients reach the clinic.
Google asks to provide real opening hours. In the Google profile, you must provide the actual hours when a patient can contact the clinic — and it is worth mentioning after-hours automatic booking in the profile description.
Check it yourself in a week — four practical steps
You don't have to assume anything — you can check it yourself in a week. Four simple steps:
Call yourself on Friday at 7:30 pm and Saturday at 11:00 am. Use a phone other than your work phone to see what the patient hears. Write down exactly what you heard — whether there's a message, silence, or the call disconnects.
Check in your phone system (PBX) panel how many missed calls you have during hours when reception is closed. If you have access to reports — see how many calls come in between 6 pm and 8 am. If you don't have reports — just call back all unanswered numbers from evenings and nights for several days.
Verify hours in Google and on your website. Does what the patient sees in Google match what's on your site? Are the hours current? Is there information about automatic registration working after hours?
Listen to your answering machine. Literally: call and record what the patient hears. Is it clear what to do? Is there an option to leave a message? Is there a link to online registration?
After a week you'll have concrete data instead of assumptions. And you'll know whether the problem is real or just imagined.
How it looks when the system manages it
After-hours scenario:
- 01phone after 6 pm
- →02AI receptionist introduces itself
- →03patient chooses available slot
- →04SMS confirmation
- →05morning list in CRM
What remains for humans: handling unusual situations that the system can't solve (specific medical questions, complaints, requests for urgent same-day slots), and final confirmation of the visit and preparation of the clinic. The rest is automatic.
Aura offers services that let you implement such a scenario: AI receptionist for after-hours phone handling, online booking for patient self-service scheduling, automated SMS for confirmations and reminders, CRM for patient management, and calendar integration to connect with your clinic's schedule.
Frequently asked questions
Can AI reception handle acute pain reported at night?
The system doesn't diagnose or assess health condition. What the system says in such a situation is set by the clinic in the scenario: e.g., emergency number according to clinic rules, information about the nearest care point, or the first available slot for the next business day.
Do I need an online calendar for AI reception to work?
Yes. AI reception registers patients in available slots, so it needs access to the current schedule. Without an integrated calendar, the system can at most accept messages for later processing by reception. Full functionality (registration without human involvement) requires connection to the clinic's calendar.
Is recording conversation with AI reception legal?
Whether conversations are recorded and what is stored is set by the clinic during implementation — and must be informed about. UODO indicates that voice recording requires a clear legal basis and informing the person.
Is an appointment reminder marketing?
A reminder about a scheduled visit is usually treated as a service message related to the service. Promotional content requires consent. Whether a specific message is marketing — a lawyer or DPO will assess.
How much does AI reception implementation cost in a dental clinic?
The cost depends on the chosen model and feature set. An exact quote can be obtained after an audit of the current registration process. See how the AI receptionist service works for more details.