Friday, half past seven. Two six-seat tables by the window, both booked, both empty. You turned away three phone calls that evening because you were full. That is the expensive part of a no-show: not the money on the table, but the guests you sent away to keep it free.
It is usually not deliberate
The conversation about no-shows tends to be about how inconsiderate people have become. My experience is duller: most of them simply forget. They booked three weeks ago, plans changed, and cancelling means calling and explaining yourself. Not turning up is easier.
Which gives you a fairly concrete instruction. If you want fewer no-shows, make cancelling easier than not turning up. Not stricter. Easier.
What works, best value first
A reminder the day before with a one-click cancel link. It is the cheapest thing you can do and it returns the most. A guest who realises the night before that they will not make it will free the table if it costs one tap. If it costs a phone call, they do nothing.
A confirmation sent the moment they book. Obvious, and plenty of venues still skip it, leaving the guest unsure for a week whether the booking exists at all. An uncertain booking gets cancelled less often than a certain one, because people barely remember it.
Deposits, but only where they earn their keep: New Year, groups over six, tasting menus, limited-capacity events. A blanket deposit on an ordinary Tuesday puts off the guests you want and buys you arguments at the door.
A waitlist. When someone cancels at two in the afternoon you can still fill that table, but only if you know who to call.
Measure yours first
Before changing anything, spend a month writing down how many bookings did not turn up. Guesses are wrong in both directions. Some people I have talked to were certain they had a problem and found three empty tables a month and a completely different issue to solve.
The goal is not zero. It will never be zero. The goal is going from five empty tables a week to one, and getting your Friday evenings back.