The anatomy of a no-show: an interactive guide
One booking, four moments
How often do booked guests fail to turn up? The best measured answer we have read is the Resos No-Show Index 2026 (published 19 August 2026). Across 309,604 reservations at 316 US restaurants it recorded a no-show rate of 3.17%. Staff have to mark a no-show for it to count, so the authors say the real rate is somewhat higher.
Your restaurant may sit a long way from that. Caterer Middle East reported on 15 October 2026 that at Gymkhana Riyadh, no-shows and late cancellations together run at around 15% of bookings. So this guide does not hand you an industry number. It starts from yours.
A no-show is also not a single event. It is the end of a short chain, and each link is a place you can step in. This guide follows one reservation from the moment it is made to the moment the table is filled or empty. Set your own rate in the tool below, then switch a deposit and reminders on and off and watch what happens to 100 people who try to book.
Follow one booking from reservation to table
Starts at 3% because Resos recorded 3.17% for US restaurants. Staff have to mark a no-show for it to count, so real rates are likely higher. Move the slider to your own rate.
The easy yes
A booking that costs nothing to make costs nothing to drop. In the Resos data the bookings most likely to be dropped were small tables and ones made 2 to 6 days ahead, not the big parties booked a month out.
Resos recorded a 0.80% no-show rate on reservations that required payment protection and 2.47% on those that did not. That raw gap, 68%, is the default below. Resos says restaurants that ask for payment differ from those that do not, so the gap cannot be read as the deposit's effect. We found no measured figure for bookings lost to a deposit, so that slider starts at zero. Set your own.
Per 100 people who try to book
- Seated
- 97
- No-shows
- 3
- Never booked
- 0
No-show rate among bookings: 3.00%
This is a model you set, not a measured result. Defaults come from the Resos No-Show Index 2026 (published 19 August 2026, covering reservations from August 2025 to July 2026): a 16% lower recorded no-show rate with a delivered SMS reminder, and 0.80% against 2.47% with and without a payment requirement. Resos calls both observed associations and says they do not show cause. The model multiplies the two switches as if they were independent. In a real restaurant they overlap, so both together is a ceiling, not a forecast.
A note before you use it. An earlier version of this guide started from a much higher no-show rate and gave each switch a fixed effect. We could not trace those figures to a source, so we took them out. The starting rate is now yours to set. Each switch starts at the difference Resos recorded and lets you change it. Resos is clear that these are links in its data, not proof that a reminder or a deposit caused the difference. We also removed the switch for AI risk flags, because we have no measured effect to give it.
Moment one: the easy yes
A reservation with no commitment is free to drop. For a guest whose plans have changed, doing nothing is the easiest path.
It helps to know which bookings get dropped, and in the Resos data it was not the ones most operators fear. Tables for one or two had the highest recorded no-show rate (2.63%) and parties of ten or more the lowest (1.15%). Bookings made 2 to 6 days ahead were the riskiest (2.67%). Bookings made 30 or more days ahead were the most reliable (1.96%).
A deposit changes the sum for the guest. With money held, cancelling in time is easier than vanishing. Resos recorded a 0.80% no-show rate on reservations that required payment protection, against 2.47% on reservations that did not. Read that with care. Resos publishes it as a raw rate and says restaurants that ask for a deposit differ from those that do not, so the gap cannot be pinned on the deposit alone. It also followed restaurants across the date they switched a payment feature on, and says none of those estimates were firm enough to publish.
The cost is friction. Some people will not book if they are asked to pay first. We could not find a source that measures how many, so the tool starts that slider at zero and leaves the number to you. Here is one operator's view.
“Asking a loyal guest for a guarantee sends the wrong message.”
Gymkhana asks for a card guarantee only for private dining and for groups of 12 or more. That fits the data. A large party fails to show less often, but it takes far more with it when it does: a no-show of ten or more removed 13.44 covers on average in the Resos data, against 1.96 for a table for two. Our deposit policy guide covers who to ask and how much.
Moment two: the quiet days
Between booking and dinner there is usually silence. A table booked days ahead can slip anyone's mind, and plans change. If cancelling means finding an old email and clicking through a form, many guests do not bother.
Reminders are the cheap fix. In the Resos data, reservations with a delivered SMS reminder had a 16% lower recorded no-show rate than reservations at the same restaurants without one (46,884 reminded reservations at 257 restaurants). The authors say plainly that this is an observed association and does not show the reminder caused the difference. In the same data, asking guests to confirm again before service showed almost no link with no-shows at all.
Send one reminder a day ahead, so a guest who cannot come can say so while you still have time to resell the table. Send another shortly before service for the guest who has simply forgotten. Every message should make cancelling as easy as confirming.
Moment three: the late change
On the day, the bigger risk is not the guest who vanishes. It is the guest who cancels late. Of 566,049 cancellations in the Resos data, 56% came inside 24 hours of the booking or after it had started, and those late cancellations outnumbered recorded no-shows by more than three to one. Resos did not measure how many of those tables were rebooked, so treat them as tables at risk, not as empty tables.
This is why the tool has no switch at this stage. What helps here is practical: a cancel link that works in one tap, a way for a guest to say they are running late, and a waitlist of people your staff can offer the table to.
Some booking systems, TableShift among them, also give each booking a risk score. Treat a score as a way to pick the few bookings that deserve a call or a deposit request before service. We know of no published measurement of how much a score lowers no-shows, so we make no claim about it. Our guide to AI in restaurant bookings covers what a score can and cannot tell you.
Moment four: arrive, cancel or ghost
By showtime a booking ends one of three ways. The guest arrives. The guest cancels and the table comes back to you. Or the guest does not appear, and you have held a table you could have given to someone else.
“What was happening is that we were turning away walk-ins, and then the reservation wouldn't show up.”
Pizzata's answer, as The Philadelphia Inquirer describes it: a card to hold the booking, a $15 per person no-show fee, a 20-minute grace period, and three reminder texts plus a courtesy call before any charge.
An empty table is not always lost money. At Gymkhana Riyadh, Roberto Paterna does not call the 15% a major financial problem, because walk-ins, the waiting list and concierge referrals usually fill the gaps. In his words: “Riyadh dines late and decides late.” What a no-show costs you depends on whether you can refill the seat.
Keep a way to be fair. A guest with a real emergency should get the deposit back. The aim is not to punish someone having a bad day. It is to make cancelling in time the easy choice.
Putting it together
Go back to the tool, set your own rate, and turn both switches on. Then try a cautious version: halve each effect and add a few lost bookings. If the picture still looks worth it, you have a plan. If it flips, you have learned where your risk is.
Notice how little room there is at a low starting rate. With the starting values, a deposit moves about two guests in a hundred from no-show to seated. If asking for it costs you more than about two bookings in a hundred, you seat fewer people, not more. At a higher starting rate the same deposit has far more to work with. That is why your own number matters more than anyone's average.
If you want the cost in dollars, run your numbers in the no-show cost calculator. If you would rather learn this step by step, the No-Show Playbook course walks through reminders and deposits with exercises.
In TableShift the pieces map across like this. Guests book on your own booking page or website widget. A refundable deposit is paid to your own Stripe account, and you can require one on every booking or request one on a single booking. Email reminders go out 24 hours and 2 hours before, with text messages on paid plans. Every confirmation email carries links to change the booking, cancel it, or say the guest is running late. Each booking gets a risk score from 0 to 100, and your staff run the waitlist. Plans are Starter (free), Essentials ($99), Growth ($149) and Scale ($249) per location per month. TableShift takes 1% of the deposits you collect, never more than $149 a month (£119 in the UK), and nothing extra when a guest does not show.
Check what you followed
0/31. Where does the guide's starting rate of 3% come from?
2. Resos found a 16% lower recorded no-show rate where an SMS reminder was delivered. What does that show?
3. Why does the deposit switch start with zero bookings lost and let you set the number?