How to Reduce Restaurant No-Shows: Let an AI Call Every Guest the Day Before
2026-08-17 · Restaurant Operations
Friday, 19:40. Table eight by the window is set for four people who booked it ten days ago. The waiter has already turned away one walk-in couple because the table was blocked. At 20:15 it is obvious nobody is coming. No call, no cancellation, no message. That table earned nothing all evening, and on top of that it cost you guests who actually wanted to pay.
Every restaurant owner knows this evening. The loss is not only the average spend of four people. It is the food you prepped, the staff you rostered for a full room, and the guests you turned away while the table sat empty. And none of it lands anywhere you can see it. A bad review you can read. A broken fridge has a price on the invoice. A slow Tuesday shows up in the till at the end of the night. A no-show is the one loss a restaurant never counts.
The Stolio AI receptionist calls every guest the day before their booking. It asks one question: does the reservation still stand? That is the whole call. It takes the guest fifteen seconds, it happens in the late afternoon and early evening when people actually pick up the phone, and it happens for every booking in the book — not just the group of fourteen that somebody found time for.
The answer goes straight back into the reservation system. A confirmation leaves the table booked and marked as confirmed, so your host can see at a glance which tables are certain and which are still open questions. A cancellation frees the table on the spot, while there are still hours left to sell it — to the waitlist, to the next person who phones, to the couple who walk in at seven. Nobody has to transcribe anything afterwards, and nothing sits in a notebook waiting to be typed in.
This works where everything else fails for one reason: it is two-way. Every other method on the list below sends the guest information and then hopes. An SMS informs. An email informs. A confirmation link asks, but it accepts silence as an answer, and silence is the problem. A phone call ends with an answer, because there is a person on the other end who has to say something. That is the whole difference, and it is the difference between knowing before service and finding out at 20:15.
What it really does is convert silent no-shows into cancellations. It is worth stating that plainly, because it is not the same as making guests more reliable. Some of the people it calls were never going to come. The AI does not change their mind — it gets them to say so. A guest who admits at four in the afternoon that it is not going to work has cost you nothing: that is a table you can still fill. You cannot resell a table you do not know is free. And unlike a deposit, none of this costs you bookings at the moment of booking, because the guest is never asked for a card. There is nothing to enforce and nobody to chase for 200 CZK.
It also answers incoming calls when nobody on the team can pick up. During service, when the line rings and both hands are full. After closing, when the call would otherwise go to voicemail and the guest would simply phone the restaurant next door. It checks real availability in Stolio, creates the booking, and the guest gets an SMS confirmation like any other reservation. It speaks the guest's language — Czech, German, English, Italian and more — and the moment a request gets complicated, an allergy question, a change to a large group, anything that needs judgement, it hands the call over to a person.
It is not the only AI that answers restaurant phones, and it would be dishonest to pretend otherwise. There are general answering services that will pick up your calls and take a message perfectly well. The difference is where this one sits: it is built into the reservation system rather than bolted on beside it, so it works with your actual table availability instead of writing down a request for somebody to process later. A service that sits outside your booking system can only ever hand you a to-do list. This one changes the reservation itself. It is an optional add-on to any Stolio plan.
Why a phone call and not something cheaper. The rest of this article is the argument for that choice, and it is an honest one, because every method below genuinely achieves something and every one of them costs you something. Before you weigh any of them, know that the figure quoted around the industry — that somewhere between 10 and 20 percent of reservations end as no-shows — is a commonly repeated range, not a measured fact about your restaurant. A tasting-menu place in the centre of Prague, a neighbourhood pub with a regular lunch crowd and a restaurant that lives off tourists in summer are three completely different situations.
Deposits and card guarantees — these work. This is the part where a lot of marketing gets dishonest, so let us be clear: if a guest has 500 CZK on the line, or has handed over card details that can be charged, the chance they simply forget drops sharply. People protect their own money far more reliably than they protect your empty table. Anyone telling you deposits do not reduce no-shows is selling you something else.
The cost is that deposits lose you bookings before they ever exist. A share of guests reach the payment step and quietly close the tab. They are not bad guests. They are people booking a Thursday dinner for two who do not want to type card details for a 900 CZK meal, and who will just book the place next door that does not ask. That is why deposits fit fine dining, tasting menus, groups of eight and up, New Year's Eve and Christmas parties far better than a neighbourhood restaurant. For a large group or a special evening, a deposit feels normal. For an ordinary Tuesday, it feels like suspicion. There is also the admin nobody mentions: refunds, card disputes, and the deeply unpleasant moment when a guest who genuinely got sick asks for their money back and you have to decide who you want to be.
SMS and email reminders — cheap, easy, and you should absolutely have them. An automated confirmation the moment the booking is made, and a reminder the day before, cost you almost nothing and catch the honest failure mode: the person who booked three weeks ago and genuinely forgot. Any reservation system worth paying for does this without you touching anything. Stolio sends the confirmation and the reminder automatically, in Czech or English, depending on the guest.
The limit is structural, and it is the reason the AI receptionist exists. A reminder is a one-way message. It informs, it does not commit. The guest reads it on the tram, thinks "right, that's tomorrow", puts the phone away, and then does not come and does not cancel. You gave them information. You got nothing back. The table is still blocked and you still do not know whether it will be used. Reminders reduce the forgetting problem. They do not touch the "changed my mind and could not be bothered to tell you" problem, which for most restaurants is the bigger half.
Confirmation links — a step up from a plain reminder. Instead of just informing, the message asks for one tap: confirm or cancel. This is better, because it forces a small decision instead of passive reading, and a cancellation that arrives at 11:00 the day before is a table you can still sell. The weak point is silence. Most people who are not planning to come will not tap "cancel" — they will simply not answer. And silence is ambiguous. You cannot tell the guest who is definitely coming but ignores links from the guest who has already decided to skip. So you keep the table blocked either way. A call closes exactly that gap: it does not let silence be the outcome.
Overbooking — the airline trick, and it translates badly. Airlines overbook because seats are interchangeable, the compensation rules are fixed and published, and a bumped passenger is dealt with by a system, not by a host standing in a doorway. In a restaurant, a table is not interchangeable. It has a size, a location and a time window. When you overbook and get it wrong, you are not paying a fee — you are telling a couple who booked a week ago and just arrived in the rain that there is no table. That guest does not come back, and they tell people. Mild overbooking on evenings where you have real historical data can be defensible. Overbooking as a policy, in a restaurant of thirty tables, mostly converts a no-show problem into a reputation problem.
Waitlists — genuinely useful, with one condition. A waitlist only recovers a table if you find out the table is free while somebody still wants it. A cancellation at 16:00 is gold: you message the list, someone takes it, the evening is whole. A no-show discovered at 20:15 is worthless, because the people who wanted that table already ate somewhere else. So a waitlist is not really a no-show tool. It is a cancellation tool, and its value depends entirely on how early you learn the truth about a booking. That is why the two belong together: the calls produce early cancellations, the waitlist turns them back into revenue.
Cancellation policies and fees — the problem is not writing the policy, it is enforcing it. Most restaurants put a line on the website saying cancellations under 24 hours may be charged, and then never charge anyone. Think about what enforcement actually requires: you have no card on file for most bookings, so you would have to call a guest, argue about 200 CZK, and either lose the argument or win it and lose the guest permanently. Almost nobody does this, and honestly almost nobody should. An unenforced policy still has some deterrent value for the conscientious, but the guests who no-show are, by definition, not the conscientious ones. If you want a fee to have teeth, you need a card, which brings you back to the deposit trade-off and its cost in abandoned bookings.
Guest history and loyalty — works, but slowly. If your system records who came, who cancelled and who did not show, then over months you build something useful: you can see that a particular name has failed to appear twice, and you can treat their booking for a full Saturday differently. Stolio keeps this history, including cancellations, so it accumulates whether or not you think about it. But be realistic about the timescale. This helps you in a year. It does not help you this Friday, and it does nothing at all for a first-time guest, which is most of the risk in a restaurant that depends on new custom.
Which leaves the method every restaurateur already knows is the best one. The day-before call beats everything above because it is a two-way conversation that requires an answer, and the reason nobody does it is not doubt — it is arithmetic. Thirty bookings is thirty phone calls. During service there is no one free to make them. Between services, the staff who are around are prepping, ordering or going home, and the calls have to land in the window when guests actually answer, which is exactly when your kitchen and floor are busiest. So the calls get made for the big group, and skipped for everyone else. That has been the honest state of the trade for as long as restaurants have taken reservations. The only thing that has changed is who makes the calls.
Two things to do next. First, measure. For one month, mark every no-show in your reservation book and multiply by your average spend per person. That number is yours, not an industry average, and it is the only one that should decide how much effort this deserves. Second, if the number is worth fixing, get in touch and we will set the AI receptionist up for your restaurant — starting with a number you can ring yourself, so you hear exactly what your guests will hear before a single one of them is called. If you are not on Stolio yet, start with a free trial and add it when you are ready.