How to write a restaurant cancellation policy

Most restaurant cancellation policies are really no-show policies wearing a different name. They set a fee for guests who never call, then say nothing about the guest who cancels at 6:45pm for a 7pm table. That gap matters more than it looks: in the Resos No-Show Index, 317,541 of 566,049 recorded cancellations, 56%, arrived within 24 hours of the booking or after it had already started. A table released that late is nearly as hard to resell as one that simply never fills. If your policy only talks about no-shows, it is addressing less than half the problem.

Key takeaways

• 56% of cancellations (317,541 of 566,049) arrive within 24 hours of the booking or after it has started, and 7.1% (40,360) arrive after the reservation’s start time.
• A late cancellation and a no-show cost you the same thing: a table you can no longer resell in the time you have left.
• A working policy needs a notice window, a party-size threshold, and a clear consequence for a release that comes in too late to matter.
• Deposits and no-show fees protect exposed services; they are not a proven fix. We publish no figure for how much they change guest behavior.
• Dinner, Mondays, and January are where the exposure concentrates, so that is where a stricter policy earns its keep.

The cancellation problem hiding behind the no-show number

The headline figure from the Resos No-Show Index is a 2.33% recorded no-show rate: 87,953 no-shows across 3,768,761 advance reservations at 2,417 restaurants, from August 2025 to July 2026. That number gets all the attention because it is the clean, final outcome: guest never came, table sat empty. But no-shows are only one way a table goes unsold. Cancellations are the other, and there were far more of them, 566,049 in the same dataset.

The distribution of when those cancellations landed is the real story. Broken down by notice given:

• After the booking’s start time: 7.1%
• 0 to 2 hours before: 17.4%
• 2 to 6 hours before: 15.0%
• 6 to 24 hours before: 16.6%
• 24 to 48 hours before: 11.6%
• 2 to 6 days before: 17.4%
• 7 or more days before: 14.6%

Add up the first four bands and you get to 56% of all cancellations arriving inside a 24-hour window, including the 7.1% that came in after the guest was already due at the table. A cancellation with a week of notice is a non-event; you rebook the slot without thinking about it. A cancellation with twenty minutes of notice on a Saturday at 8pm is functionally a no-show with slightly better manners.

Why late cancellations deserve their own policy line

A cancellation policy that stops at “no-shows will be charged” leaves the entire 56% untouched, because those guests did cancel. Technically they followed the rules. Practically, they handed you a table with no time to resell it, which is the exact outcome you were trying to prevent.

This is where a digital waitlist earns its place next to the policy rather than instead of it. A late cancellation that lands with two hours of notice still has a shot at being resold if you have a list of waiting guests to text the moment the slot opens. A policy without a backfill mechanism just documents the loss after it happens.

What a cancellation policy needs to cover

Most templates you’ll find online cover one thing: what happens if the guest doesn’t show. A policy that actually protects your service needs three more pieces.

A notice window. State the number of hours before the reservation that counts as a “cancellation,” and set it based on how hard your busiest services are to resell, not a round number you saw somewhere else. A Tuesday lunch table might be resellable with two hours’ notice; a Saturday dinner table for eight probably needs 48.

A party-size threshold. Large parties don’t fail often, but when they do, the damage is disproportionate. In the index, parties of 10 or more no-show at just 1.15%, well below the overall rate, but each one that fails takes an average of 13.4 covers with it. A policy that treats a table for two and a table for twelve the same way is protecting the wrong risk.

A stated consequence for a late release. Not just “you may be charged,” but what specifically happens: a fee, a hold on a card, or simply a note on the guest’s profile if this is a repeat pattern. Ambiguity here is why staff hesitate to enforce the policy at all.

A cancellation policy you can adapt

Here is a short, concrete version you can rewrite in your own voice:

Cancellation policy. We hold your table for 15 minutes past your reservation time. Please cancel or change your booking at least 24 hours in advance so we can offer the table to another guest. For parties of 6 or more, we ask for 48 hours’ notice given the size of the table involved. Cancellations made less than 2 hours before the reservation, or after the reservation time, will be treated as a late release: the card on file may be charged our no-show fee, since at that point we usually cannot resell the table.

Adjust the numbers to your own service. A neighborhood cafe with walk-in overflow might set the notice window at 2 hours across the board. A tasting-menu room that preps for exact covers might need 72 hours and a firm party-size cutoff at 4. The structure stays the same: a notice window, a size threshold, and a plainly stated consequence for a release that comes in too late to help you.

Where reminders and confirmation fit in

A policy only works if guests see it before they book, and if you give them an easy way to act on it before the window closes. A clear booking confirmation step does the second part: it surfaces the policy at the moment of booking and gives the system a chance to re-confirm or release a reservation while the slot is still sellable, instead of finding out at the door.

Reminders do the first part. They don’t just nudge a guest who forgot, they give a guest who already knows they can’t make it a clean, easy way to cancel before the window narrows to nothing. In the Resos No-Show Index, reservations with a delivered SMS reminder showed a 16% lower recorded no-show rate than reservations at the same restaurants without one, based on 46,884 exposed reservations at 257 restaurants that had bookings both with and without a delivered reminder. That is an observed association from operational data, not a controlled experiment, so treat it as a signal that reminders correlate with fewer no-shows rather than a guaranteed result for your restaurant.

Where to make the policy stricter

Not every service carries the same exposure, so a single blanket policy leaves money on the table in some slots and annoys guests unnecessarily in others. The index points to where the risk actually concentrates: dinner runs a 2.63% no-show rate against 1.84% at lunch, 43% higher. Monday is the worst day at 2.69%, Wednesday the best at 2.09%. January is the worst month at 2.62%, August the best at 2.14%. If you only have the bandwidth to enforce a strict notice window and party-size threshold on a few services, dinner, Mondays, and January are where a working strategy to reduce no-shows pays off first.

What deposits and fees can and can’t promise you

A no-show fee or a required deposit is a reasonable way to protect an exposed service, especially the large-party and peak-dinner bookings above. What it is not is a proven way to lower your no-show rate. We publish no figure for that, and we would not stand behind one. Use it to recover the cost of a table that goes unsold and to signal that a reservation is a real commitment, not as a claim that it will change guest behavior on its own.

Put the policy where guests will actually read it

Write the policy into your confirmation message and your booking page, keep the notice window and party-size rule short enough to fit in two sentences, and pair it with a reminder guests can act on and a waitlist that can backfill the table when a late cancellation does come in. That combination, not the policy text alone, is what turns a released table back into a filled one. Resos handles the confirmation, reminder, and waitlist pieces automatically, so the policy you write is one your booking system actually enforces.