Operated function · booked time
An unpaid invoice can still be collected next quarter. A Tuesday at ten cannot — at ten past ten it has gone, and the staff who were paid to be there were paid anyway. Which is why almost everything worth doing about a no-show happens before it, not after.
The booking was made three weeks ago, when it mattered. Life moved. The person who booked it has since arranged something else, or resolved the thing they booked for, or simply lost the plan — and they have not thought about you since, because nothing has asked them to.
A reminder went out. It was sent to the address on file, which for a substantial share of any book is out of date or belongs to a channel they never check, and it said the time and the place and nothing that would prompt a person to say "actually, no".
The cancellation route is worse than the booking route. Booking took two taps; cancelling takes a phone call during opening hours to a queue. So the rational thing for a busy person is to say nothing, and the business receives silence where it could have had twenty-four hours’ notice and a chance to refill.
When the slot does empty, the waitlist exists but nobody works it. Filling a gap at short notice means phoning down a list, most of whom will not answer, during the exact part of the day the front desk is busiest — so it happens when someone has a quiet moment, which is rarely.
And the pattern is knowable but unread. The same appointment types, the same times of day, the same intervals between booking and appointment, and frequently the same individuals, produce most of the misses. Everyone at the desk knows this and nothing in any system records it.
A missed appointment cannot be recovered after it occurs, so the only leverage is confirmation and refill inside the window before it — and both are manual work that competes with the busiest part of the day.
That constraint sounds obvious and it is routinely violated, because after-the-fact work is easier to schedule. A no-show report reviewed weekly is a record of expired inventory; it changes nothing about the inventory that expires next week.
Confirmation is the highest-leverage action available and it is mostly about reachability rather than persuasion. A message that arrives on a channel the person actually uses, close enough to the appointment to be actionable, and that makes rescheduling as easy as replying, converts silence into notice — and notice is what makes a refill possible.
Making the exit easy is counter-intuitive and it is the strongest single move. A business that makes cancelling a one-tap reply gets told about tomorrow’s gap today. A business that makes cancelling awkward gets an empty room and no warning, which is strictly worse for both parties.
Refill is then a queue problem rather than a phone problem. If the waitlist is ranked and contactable, a gap that appears at four in the afternoon can be offered to the right people immediately rather than when the desk clears — and short-notice availability is genuinely valuable to a subset of any waitlist.
And the pattern data closes the loop: knowing which slots and which intervals miss most turns scheduling itself into the fix, which is cheaper than any amount of chasing.
Notice before an appointment is missed — measured by releases received in advance as a share of all appointments not attended, against a baseline of silent non-attendance.
Slots that go empty and stay empty — measured by unfilled slots per week against your own prior period, with the same schedule shape.
Time from a slot opening to it being refilled — measured by elapsed minutes to first offer and to acceptance, where a baseline is often unmeasured entirely.
Whether the contact route on file actually works — measured by deliverability and reply rate by channel, which usually reveals a share of the book unreachable by the default channel.
Which slots miss most — measured by non-attendance by appointment type, time of day and booking lead time — a breakdown most schedules cannot produce.
Consistency of policy application — measured by deposit or policy actions applied at the threshold your policy defines, as a share of cases that crossed it.
any clinical, diagnostic or advisory judgement, and no decision about a person’s care, eligibility or standing. Nothing here triages urgency, decides who should be seen sooner, or applies a penalty. Where an appointment involves professional judgement about priority, that stays with your staff and is escalated rather than handled.
This reads and, where you delegate it, writes to your existing scheduling system through its documented interfaces. No second calendar is created — two places holding the truth about who is coming at ten is a double-booking waiting to happen, and it will happen in front of a customer.
Contact goes out on channels you already own, from your number and your domain, so a confirmation looks like it came from the business the person booked with rather than from an unfamiliar service.
Where a scheduling system exposes no interface, that is reported as a limitation rather than worked around by automating its screens. A screen-driven booking integration breaks silently at the next vendor update, and it breaks on somebody’s appointment.
Consent and channel preference are read from your record and honoured. Somebody who asked not to be texted is not texted, and a preference does not get overridden because another channel converts better.
Frequency is capped by your rule. A confirmation sequence that maximises attendance by messaging repeatedly will annoy the people who were always going to come, and they are the majority.
Every contact carries a receipt naming the channel, the message, the time and the response — which is what settles a dispute about whether somebody was told, and in a regulated setting is what a reviewer will ask for.
Where the appointment involves protected information, the messaging is scoped so that content is not disclosed on an unsecured channel. A reminder can confirm a time without stating what it is for, and where a Business Associate Agreement governs the work, obligations attach because that agreement exists.
Operational access is not permission to train. Appointment and contact data does not become material improving anything serving another organisation.
The person who owns the schedule owns the thresholds: how early to confirm, how many contacts, who may be offered a short-notice slot, and what happens when nobody accepts. Those are operational decisions with real customer consequences and they should not be vendor defaults.
In a clinical or otherwise regulated setting, the privacy question is about message content on an unsecured channel, and it is answered by scoping content rather than by asserting encryption. A reminder that confirms a time without naming a service is safe on any channel.
Where an obligation attaches through an agreement or a data class — a BAA, a messaging rule, a residency requirement — it is marked applicability-gated rather than presented as already in force.
A single location and appointment type — observed first, with no message sent and no calendar written — to establish the non-attendance pattern and how much of the book is reachable at all.
Observation produces two things worth having regardless of what follows: which slots, times and booking lead times actually miss, and what share of the contact routes on file are dead. The second number is usually the surprise, and it caps what any messaging programme could achieve.
Where the reachability number is poor, the correct first action is a contact-data exercise rather than a reminder programme — sending better messages to wrong numbers is effort applied at a nonconstraint.
The first delegation is the confirmation message on one appointment type, with an easy release path, on approved language, with the waitlist offer added only after confirmation is working.
Nearly everyone does, and the two questions that decide whether they work are whether they arrive on a channel the person uses and whether replying to cancel is as easy as attending. If your reminders are one-way and cancelling requires a phone call in office hours, you are systematically converting would-be notice into silent non-attendance. The observation phase measures reachability directly, and that number usually settles the discussion.
It will increase cancellations and decrease no-shows, and those are very different things. A cancellation twenty-four hours out is a slot you can refill; a no-show is expired inventory plus staff who were paid to wait. The people who cancel when it is easy were not going to attend anyway — the only thing a hard cancellation route changes is whether you find out in time to sell the slot to somebody else.
Then texting them would be the wrong channel and the operation should not use it. Channel preference is read from your record and honoured, and voice remains a first-class route rather than a fallback. This is exactly what the reachability measurement in the first phase is for — it tells you what your book actually answers, rather than assuming a channel mix that suits the tooling.
Deposits reduce non-attendance and they also deter booking, and they fall hardest on the people least able to absorb them — which in some settings is a fairness problem before it is a revenue one. They also do nothing to refill the slot, which is where the recoverable money is. If your deposit policy is working, keep it; the measurement worth having is still which slots go empty and how fast they could be filled.
It is, which is why message content is scoped so an unsecured channel carries a time and no purpose, why the language is approved by you before anything sends, and why nothing here makes a judgement about urgency or priority. Where a scope involves protected information, the agreement that creates the obligation is executed before any of it is reached. If a message set cannot be made safe for your setting, the honest scope is observation only.