How much of your room to hold for reservations and how much to leave for walk-ins is one of the few decisions that changes your revenue without changing anything about your food. Get it wrong in either direction and you leave money on the floor.
The two failure modes
All reservations. The room is fully booked, and then eight per cent do not turn up. Those tables sit empty through the busiest hours while people who wanted to eat were turned away at six because you had nothing.
All walk-ins. You are full when you are full and empty when you are not, with no ability to plan staffing or ordering, and no way to smooth a Tuesday.
Both are worse than a deliberate mix, and most restaurants arrive at their mix by accident rather than by choosing it.
What decides the right mix for you
Your no-show rate. The higher it is, the more you need walk-in capacity to absorb the holes. Measure it before guessing — most restaurants either have not measured it or quote a number from memory that is comfortably low.
How far ahead people book. A restaurant booked three weeks out has a different problem from one booked the same day.
Your location. Passing footfall makes walk-ins viable. A destination restaurant with no street presence cannot rely on them.
Party size mix. Large parties need booking. A room of two-tops can run far more walk-in heavy.
Your service. A tasting menu cannot absorb a walk-in. A bistro can.
A workable starting point
Hold a portion of the room for walk-ins during your peak hours, and release unclaimed reservation tables on a defined timer.
Two rules make the mix work in practice:
Hold the bar and the small tables for walk-ins. Two-tops turn fastest and are the most common walk-in. Keeping a handful unbooked costs you little and catches the most people.
Release a no-show after a set grace period. Fifteen minutes is common. The important part is that it is a stated policy your host actually applies rather than a judgement made under pressure on a Friday.
Reducing the no-shows that force the trade-off
The less you lose to no-shows, the more of your room you can safely book.
Confirm, and make confirming easy. A text the day before with a one-tap confirm materially reduces no-shows, and it costs almost nothing.
Call the large parties. A table of ten not turning up is a hole you cannot fill. Those are worth a person's time.
Take a card for large parties or peak slots. Not necessarily charging it — holding it changes behaviour on its own. Whether a fee is worth it is a separate calculation, and we ran it in a reservation no-show policy you can actually enforce.
Make cancelling effortless. This is the counter-intuitive one. A guest who can cancel in one tap gives you back a table; a guest who has to ring and apologise simply does not show. Your cancellation path should be easier than your booking path.
The part that quietly breaks the whole plan
All of the above assumes you actually receive the bookings and the cancellations.
Most restaurant calls arrive during service, which is when nobody can reach a handset. A booking that rings out is a table you did not sell. Worse, a cancellation that rings out is a table you cannot resell — you hold it, the party does not come, and you have lost it twice.
That second one is the expensive and invisible half. A restaurant with a nine per cent no-show rate may genuinely have a five per cent no-show rate and a four per cent could-not-get-through rate, and no report anywhere distinguishes them.
Whatever you use to cover that — a booking page, a host who can step away, something that answers — it needs to handle cancellations as readily as bookings. What a restaurant phone setup actually has to do covers the options.
The measurement worth having
Three numbers, weekly:
No-show rate, as a percentage of covers booked. Walk-ins turned away at peak, counted by the host. Empty seat-hours during service.
The third is the one nobody tracks and the one that tells you whether your mix is wrong. A room that is never full and still turning people away has a distribution problem rather than a demand problem.