AI appointment reminder calls: cutting no-shows
A clinic's revenue is bounded by the doctor's time: when an appointment was booked and the patient does not arrive, that time does not come back and another patient waiting could not have it. This article builds the practical model for preventing no-shows: the five groups of reasons, the timing and number of reminders, the choice of channel, what the call says and the privacy limit, the confirm-move-cancel trio, filling the freed slot, repeat non-attenders, and the cuts to measure.
A patient who does not arrive is an empty slot
A clinic's revenue is bounded by the doctor's time. When an appointment was booked and the patient does not arrive, that time does not come back: the doctor sits idle while another patient waiting for a slot could not have it.
The loss is double. An empty slot on one side, a patient who did not get seen on the other — revenue lost and someone waiting at the same moment. The daily report usually shows only the first of those.
This article builds the practical model for preventing no-shows: the reasons, the timing and number of reminders, the choice of channel, what the call says and the privacy limit, filling the empty slot, and measurement. Booking the appointment is a separate subject, covered in appointment booking with a voice AI.
Why they do not arrive
The reasons usually fall into five groups, and only one of them is genuinely forgetfulness.
- Forgot: particularly where the appointment was booked a week or two earlier
- Plans changed: work, family, travel — and cancelling looked like a hassle
- Their condition changed: the complaint passed, or worsened and they went elsewhere
- A practical obstacle: transport, childcare, time off work
- Hesitation and anxiety: avoidance of the appointment — in this group the tone of the call matters most
The split matters, because each group needs a different intervention. Those who forgot need a reminder; those whose plans changed need an easy way to move the appointment — and it is this second group that reminders serve best, because they can tell you in advance.
One nuance about the fifth group: an insistent reminder loses those patients entirely. The tone has to be gentle and the question should not be «are you coming?» but «does this time still work for you?»
Timing and number of reminders
Too few reminders do not work and too many are intrusive. In practice two contacts hold the balance.
- First reminder: 24–48 hours beforeThe purpose is not only to remind — it is to learn about a change and release the slot. That is the main economic benefit of reminding at all.
- Second reminder: the morning of the appointmentA short confirmation, particularly useful for morning appointments.
- A separate call for unconfirmed appointmentsWhere neither contact got an answer, the slot is treated as at risk and the waiting list is kept ready.
- An extra early contact for distant appointmentsFor appointments more than a month out, an additional contact a week ahead helps — those are the most forgotten.
The hour matters too: answer rates differ between morning and afternoon, and the clinic's own data shows which. Reminders are not sent at night.
Channel: message, call, or both
The most used model in clinics is mixed, for practical reasons.
- Message: cheap, fast, and enough for most patients
- Call: after an unanswered message, and for first appointments
- Call: for older patients and groups that do not use messaging
- Call: where the procedure requires preparation — because it needs explaining
- Neither: where the patient has said they do not want contact
The fourth line matters particularly in clinics. Some examinations require preparation — arriving fasted, pausing a medication, bringing a document. That can be sent by message, but only a conversation confirms it was understood.
What the call says, and the privacy limit
In a clinic call, privacy is not a technical detail — it shapes the structure of the script.
The same rule applies to voicemail, where the risk is higher because the message stays on the phone. A voicemail should leave only the clinic's name and a request to call back.
The second limit concerns content: a reminder call gives no medical advice, explains no results and asks nothing about the patient's condition. If the patient asks, the conversation moves to reception or to clinical staff.
Confirm, move, cancel — all three should be easy
The purpose of a reminder is not only to say «please come». If the patient cannot come, the clinic knowing in advance is worth nearly as much as their arrival.
- Confirming should take one word
- Moving: an alternative time is offered immediately — «call us later» loses the slot
- Cancelling: simple and without blame, or the patient will simply not turn up
- In all three cases the result is reflected in the system at once
- A cancelled slot opens automatically to the waiting list
The third line is counter-intuitive: making cancellation easy lowers the no-show rate. When cancelling is hard the patient simply does not arrive and the clinic learns nothing; when it is easy, the slot frees up and can be filled.
Filling the empty slot
This is the second half of the system and it produces more revenue than the first: filling the time that was released.
- The waiting list has to existPatients who said «let us know if something earlier comes up» are kept on their own list. Without it, a freed slot stays empty.
- The notification has to be fastIf a slot frees a day ahead, the response window is hours. A call to the waiting list is the fastest channel.
- The offer has to be specific«Tomorrow at 14:00 has become available, does that suit you?» — a concrete time, not an open question.
- The first yes takes the slotThe same slot must not be promised to several patients — that creates a second problem.
Repeat non-attenders
Missing once is normal. Missing systematically needs its own approach, and it should be an adaptation rather than a punishment.
- After two misses, the confirmation requirement increases
- After a third, a deposit or a confirmation condition may apply — the rule is announced in advance and openly
- Ask the reason: it is usually a practical obstacle with a simple fix
- These patients are not placed in the hours with the lowest attendance
- The rule is the same for everyone — individual decisions generate resentment
The third line produces the most. Most repeat non-attenders are not uninterested — their schedule does not fit, transport is a problem, or they cannot come during working hours. One question surfaces it.
Measurement
A no-show rate on its own says little; it has to be read in cuts.
- Overall no-show rate and its distribution across days of the week
- By hour: morning and evening slots behave differently
- By lead time: appointments booked further ahead are lost more often
- Confirmation rate: the share responding to a reminder
- Advance cancellation rate: a rise here is good, not bad
- Filled slot rate: the effectiveness of the waiting list
- Split between new and returning patients
The fifth line is frequently misread. A rise in advance cancellations shows the system working: the patient tells you, the slot frees, and it can be filled. The bad indicator is not turning up without warning.
Limits
A reminder system does not solve every cause of non-attendance. Price, distance, waiting times and dissatisfaction with the service are not fixed by a call.
The second limit is tone. An insistent reminder can raise attendance in the short term and lose the patient over the longer one. For a clinic, a patient returning matters more than a patient attending once.
How Vexvon supports this flow
The parts a reminder flow needs exist on the Vexvon side.
- Reminders are stored in the system and checked every minute, with night hours protected
- A call campaign is built from a CRM filter: «tomorrow's appointments, unconfirmed» becomes a target list directly
- The scenario defines the fields to extract: confirmation, a request to move, the reason for cancelling
- The knowledge base limits what the agent may say — clinical content stays outside the scenario
- When the condition is met the call moves to reception
- Outcome codes produce the statistics on no-shows, cancellations and changes
- The waiting list for freed slots can be run as its own campaign
Inbound call handling is on the call center page, and the source of the agent's information on the knowledge base page.
First step
Take two months of appointments and produce the no-show rate in two cuts: by day of the week, and by how far in advance the appointment was booked. Those two tables usually show immediately where the problem concentrates.
Then build a two-contact reminder for one segment only — appointments more than a week out, for instance. To build the model for your own clinic, get in touch.