twimco.Let’s talk ↗︎
Menu
Field notes / Product & engineering

Somebody cancelled. Who gets the appointment?

When a license or permit appointment frees up, your booking system already has a rule for who gets it, even if nobody wrote it down. Usually it's whoever refreshes fastest. Here are four rules a public office can choose instead, and what each one asks of staff.

Editorial engraving of a bright municipal back office where a coordinator moves a token from an open calendar square to the first in a row of waiting tokens.

A cancelled government appointment goes to someone by some rule, written or not. The default, release it instantly to whoever is looking, rewards people and software that refresh constantly. Offices can instead offer it to a waitlist in order with a short hold, release freed slots in batches at a published time, or keep some slots for phone and counter booking. Pick one per service and publish it.

The rule you already have

Somebody cancels a driving test, a passport interview or a building inspection for next Tuesday. The slot goes back into the calendar. Who gets it?

In most booking systems, the answer is whoever happens to be looking at that moment. The slot appears, the first person to click takes it. Nobody decided that. It's just what the software does when nobody tells it otherwise.

For years that mostly meant the people who checked the website most often: the persistent, the ones with flexible jobs and time to refresh. That's already a quiet bias. Now there's a new kind of visitor. OpenAI's release notes for ChatGPT Work's signed-in browsing, from August 27, give "Book a DMV appointment or fill out your passport renewal forms" as examples of what it can do for someone, asking them to confirm before it completes a reservation. Other tools, and plain scripts, can watch a calendar too.

That isn't a reason to panic or to block assistants outright; plenty of people who use them are just trying to get an appointment. It is a reason to choose the rule on purpose. Every booking system already has a rule for cancelled slots. The question is whether anyone chose it.

Rule 1: first come, released immediately

How it works. The slot goes back on the calendar the moment it's cancelled. Anyone can book it.

Good for. Services with plenty of capacity, short lead times and few cancellations, where the slot will probably be filled at the counter anyway.

What it costs. Nothing to build, which is why it's the default. But in a busy service it rewards whoever checks most often, person or software, and tells everyone else the calendar is always full.

What staff need. Nothing new, apart from fielding complaints that "it was full every time I looked."

Every booking system already has a rule for cancelled slots. The question is whether anyone chose it.

Rule 2: offer it to the waitlist, in order, with a hold

How it works. People who want an earlier date join a waitlist, saying which locations and times they can make. When a slot frees up, the system offers it to the first person who fits, by text or email, and holds it for them for a short window. If they don't confirm, it goes to the next person.

Good for. High-demand services where an earlier appointment really matters, such as licensing, inspections or immigration-related interviews.

What it costs. A waitlist, a message, a hold timer and a fallback. The hold has to be short enough that the slot isn't wasted and long enough that someone at work can reply. Write down what happens if the slot is tomorrow morning and nobody confirms tonight.

What staff need. A way to see the waitlist, move someone up for a documented reason and explain to the public how the order works.

One more decision belongs here: one place per person. If someone can join the waitlist five times under five email addresses, the order stops meaning anything. Tie the place to whatever the service already uses to identify an applicant, such as a license or case number, and let people change their preferences without losing their place.

How the work moves

Waitlist offer with a hold

  1. Slot cancelled

    Next Tuesday, 9:30, one location.

  2. Find the first fit

    Waitlist order, location and times they can make.

  3. Offer and hold

    Text or email; the slot is held for a short window.

  4. Confirmed

    Booked; the waitlist moves up.

  5. No reply

    Offer the next person, or release by your fallback rule.

Illustrative flow; set your own hold window and fallback. Not a measured result.

Rule 3: release freed slots in batches, at a published time

How it works. Cancelled slots don't go back immediately. They're collected and released together at a time you publish, for example every weekday at 10 a.m.

Good for. Services where you can't run a waitlist yet but want to stop the constant-refresh race. Everyone knows when to look, so the advantage shrinks.

What it costs. Small, but slots cancelled late for the next day may need a separate rule, or they'll go unused. A batch release can also bring a rush of traffic, so the booking site has to handle it.

What staff need. A short public note on how release works and an answer for people who can't check at that time.

Rule 4: hold some slots back for phone and counter booking

How it works. A share of appointments, including some freed by cancellations, can only be booked by phone or in person.

Good for. Services used by people who are less likely to book online: older residents, people without reliable internet, people who need help in another language.

What it costs. Staff time on the phone and at the counter, and a clear rule about how many slots are held back and when unused ones go online.

What staff need. The same booking screen as the public uses, with the held-back slots visible to them and a way to release unused ones.

Decision guide

Four rules for a cancelled slot

ConsiderationWho it favorsWhat it asks of you
First come, immediateWhoever checks most often.Nothing to build.
Waitlist with a holdPeople who said they want an earlier date, in order.A waitlist, messages and a hold timer.
Batch releaseAnyone who looks at the published time.A schedule, a public note and capacity for the rush.
Held for phone and counterPeople who won't book online.Staff time and a release rule.
Decision framework; illustrative, not a measured result or any agency's policy.

How to choose, and where to start

Most offices don't need one rule for everything. Pick per service.

If the service is quiet, keep first come. If it's busy and an earlier date matters, run a waitlist with a hold. If you can't build a waitlist this year, publish a batch release time. In almost every case, keep some slots for people who won't book online.

Whatever you pick, publish it in plain words on the booking page: how cancelled appointments are offered, and what someone should do if they need an earlier date. People are far more patient with a rule they can see.

Then check whether it's working. Look at how many freed slots go unused, how many waitlist offers are confirmed, and whether complaints about "always full" change. And make sure assistants and people meet the same rule: the same booking limits, the same confirmation step, the same hold windows. Fairness gets harder to explain if they don't.

For what happens once people arrive, see Let people wait for the counter somewhere other than the lobby.

This week, pick your busiest bookable service and ask one question of whoever runs the booking system: when someone cancels, what exactly happens to the slot? Write the answer down. If it's "it goes back on the calendar," you've found your current rule, and you can decide whether to keep it. Then ask the counter staff what residents say about booking. They hear the complaints the website never records, and they'll know which services people give up on.

We build government and public-sector services that work for residents and the staff behind the counter, including booking that's fair and explainable. If your appointment calendar is always full and nobody's sure why, talk to us.

Related work: PatientIQ.

twimco.

We build alongside the people doing the work.
Bring us the problem you’re thinking about.

Related work

PatientIQ ↗︎