Reducing no-shows: what actually works, and what just annoys people
Two reminders rather than four, the wording that turns a reminder into a rebooking, why the good version costs the same to send as the weak one, and the consent rule that decides whether you may send any of them.
Sohail Safdar · · 7 min read
A no-show costs you the slot, the staff time, and often the customer. Most practices respond by sending more reminders. That helps a little and then stops helping, because volume was never the problem.
What moves the number is when the reminder arrives, what it asks for, and what happens when somebody replies to it.
Two reminders, not four
24 hours before. Far enough out that the customer can still rearrange their day, or tell you they cannot come while you can still offer the slot to somebody else. This is the reminder that saves revenue.
Two hours before. Close enough to be the actual nudge, and the one that catches the person who genuinely forgot.
A reminder a week out is ignored, not because people are careless but because there is nothing to do about it yet: you cannot usefully act on Tuesday from the previous Monday. A fourth reminder is how you collect a STOP, and an opt-out applies to the whole business rather than to reminders, so one over-eager sequence costs you the ability to remind that customer about anything ever again.
If two well-timed reminders are not working, add nothing. The problem is the wording, or the booking.
The reminder has to ask for something
A statement gets read and dismissed. A question gets answered.
Weak:
"Reminder: you have an appointment at Bright Smile Dental tomorrow at 2:00pm."
Better:
"Hi Dana, reminder of your cleaning tomorrow (Tue) at 2:00pm with Sam at Bright Smile Dental. Reply C to confirm or R to reschedule. Reply STOP to opt out."
The second does four things the first does not: it names the person and the provider, it names the day as well as "tomorrow" in case the message is read the next morning, it gives a reply that takes one character, and it makes rescheduling exactly as easy as confirming.
And it costs the same to send. The weak version is 76 characters and the better one is 154, and the billed unit for a plain-text message is 160 characters, so both are one segment. There is no saving in the short version; it is simply worse. (Add one emoji and the better version becomes three segments, because a single non-GSM character drops the limit to 70. SMS character limits and segments has that arithmetic.)
The reschedule objection, answered
Practices resist making rescheduling easy, and the resistance is understandable and wrong.
Making it easy does not cause cancellations. The customer who cannot make Tuesday is not coming on Tuesday either way. The only question is whether you learn about it with 24 hours to refill the slot or at 2pm when the chair is empty.
An easy reschedule converts a silent no-show into a kept appointment. The rate going up is good news, which is why it is on the measurement list below.
What happens when the reply is not C or R
Most replies are neither. They are sentences: "can we do Thursday instead", "I'm running late", "does Sam still work there", "actually I need to cancel, my mum's ill".
So the reminder is not a one-way message, it is the opening of a conversation, and it needs somewhere to land. A reminder sent from a number nobody watches is a reminder that collects cancellations nobody reads, which is worse than not asking: the customer told you and you did not act, which from their side is identical to being ignored.
Whatever sends the reminders has to deliver the replies into the same place your team already works.
Timing in the customer's day, not your admin schedule
A reminder sent at 8:55am because that is when the morning admin runs is a reminder competing with somebody's commute.
Two rules:
Send in the customer's local time. For a single-location practice that is your time, and it stops being true the moment you have two locations in different zones, or a customer who kept their old mobile number after moving. Cutting "tomorrow" on the wrong clock puts a 9pm appointment on the wrong day.
Respect quiet hours, in the recipient's zone. Nothing about a reminder needs to arrive at 3am, and a system that cannot defer out of quiet hours will eventually send one. That message costs more goodwill than the reminder earned. Note that the recipient's zone is the one that matters, not the practice's: a California practice texting a Florida patient at 8pm Pacific is reaching them at 11pm.
The consent question, which most articles skip
You may not text everybody whose number you hold.
The distinction that matters is what the message is for. A reply to somebody who just messaged you is conversational, because they started it. An appointment reminder is transactional, and in practice it needs evidenced consent: a record of this person agreeing to be texted, and ideally the wording they agreed to.
The practical consequences are unglamorous and worth planning for.
- Collect consent at booking, as an unticked box with the disclosure next to it. Pre-ticked does not count.
- Do not back-fill. Assuming everybody who ever gave you a number consented is the assumption that produces complaints.
- Expect a gap. Every customer who booked before you started recording consent gets no reminders until somebody records that they agreed. That is the honest state of your records, not a fault in the software.
A system that silently sends nothing for those customers is worse than one that shows you the skip and the reason, because the second is a list you can work through.
Not legal advice. TCPA and state rules carry real penalties. Have counsel review your opt-in wording. Keeping a consent record for texting covers what to keep.
The reminder that never arrived
Worth one paragraph, because it is invisible by design.
A text can fail at the carrier: a disconnected number, a landline, a block, a filtered message. From your side the appointment looks reminded. From the customer's side nothing happened, and when they miss it they are a no-show in your figures and blameless in fact.
So whatever you use should tell you about failures rather than only about sends, and a run of failures to the same number is a number to correct rather than a customer to chase. What happens when a text fails to deliver goes through the categories.
Deposits work, and they are a trade
A deposit reliably reduces no-shows, because it converts a free option into a decision. It also reduces bookings, because some people will not pay one.
For long or high-value appointments that trade usually favours the deposit; for a fifteen-minute consult it usually does not. The mistake is one policy for every service. And if you take them, state the policy on the booking page before the customer confirms, because finding out afterwards is what produces a chargeback, and a chargeback costs more than the no-show did.
What to measure
Not the no-show rate alone. Three numbers:
- Confirmation rate. How many reply to the 24-hour reminder at all. Low means your wording is not asking for anything.
- Reschedule rate. How many move rather than vanish. Going up is good.
- No-show rate by service and by staff member. An aggregate hides the one service or the one provider where the problem actually is.
How to calculate your no-show rate has the formula and the splits.
How Frontiva handles it
Reminders are armed at the moment the appointment is created, inside the same transaction, so there is no scheduler sweep that can miss one. 24-hour and 2-hour SMS, deferred out of quiet hours rather than dropped, and sent only to customers with recorded consent, with nothing back-filled: an appointment booked before consent was recorded shows a "no consent" skip rather than silently sending nothing.
Two guarantees enforced by the database rather than by careful code: an appointment cannot be double-booked, and a reminder cannot be sent twice. The second is a partial unique index rather than a flag somebody remembers to check.
Quiet hours are measured against the recipient's own clock, derived from their number, with a per-state window where a state sets one, falling back to the business's zone. That is the rule from the timing section above, applied rather than remembered.
When a customer replies, it lands in your inbox as an ordinary conversation for a person to pick up. In a text the assistant has no calendar tool and will not offer slots, deliberately, so the customer rebooks on your public booking page, which shows real availability against the same rotas and buffers your staff calendar uses. On a call it can move the appointment itself; can an AI receptionist actually book appointments covers why those two channels are built differently.