"One server per sixteen covers" is the number that circulates most, and like every circulating number it is true in exactly one case: the restaurant it was measured in. Applied anywhere else it produces either an uncovered floor or unnecessary cost, sometimes in the same venue on two different nights.

It is worth understanding what it actually depends on, because that is what ends up in the rota.

The ratio is not one number, it is four

What we call covers per server depends on at least four things, and changing any one of them moves the answer more than the number of guests does.

The style of service. An à la carte menu with wine and dishes explained asks three or four more passes per table than a fixed formula. Those minutes multiply across every table of the evening.

Table turns. A hundred covers in one sitting and a hundred covers across two sittings are different nights: the second one also contains a hundred resets, and the reset is exactly when the floor goes thin.

Physical distance. A room across two floors, or a kitchen far from the pass, burns time in walking that no standard ratio accounts for.

Who is on that night. An experienced server who knows the menu covers a section that a new hire needs two people for. A plan that treats everyone as interchangeable is the plan that falls over on Friday.

The useful number is per slot, not per night

The costliest mistake is not getting the ratio wrong, it is applying it to the whole service. A 120-cover night is not 120 covers spread evenly, it is a wave: nearly empty at 19:30, full between 20:30 and 21:30, tailing off after.

If the rota puts the same number of people on from 19:00 to 23:00, you are paying for staff in the first stretch and running thin in the second. It is the same reasoning that applies to shops, which we covered in peak footfall and staffing: coverage is declared per slot, not per day.

Getting to your own number

A method that works and needs no software, at least to start:

Take four weeks of bookings and count covers in half-hour bands rather than per night. It only takes an hour and it shows you the shape of the wave, which in most venues is surprisingly stable.

Mark the nights that went badly and ask what was missing: a person on the floor, a person in the kitchen, or neither of those and instead a reset that never had a chance. The three causes get blurred together and need separating.

Test the number, do not argue about it. Adding half an hour of overlap in the peak band is cheap and verifiable within two weeks. Theoretical debates about the right ratio are not.

The two mistakes you see most

Sizing for the busiest night. Putting the worst-case headcount into every service pays for coverage that is unnecessary seventy per cent of the time, and sooner or later gets cut all at once.

Sizing for the average. The opposite gives you a floor that holds four nights out of seven and collapses on the three that carry the revenue.

The answer sits in between and is not a compromise: it is a plan with different numbers in different bands, which is laborious by hand and therefore rarely done.

And the kitchen?

Floor ratios and kitchen ratios condition each other: a slow kitchen fills the room with waiting tables, and at that point even a correct ratio looks wrong. The reasoning on the brigade is in kitchen shift planning.

In Sked Solve coverage is declared per time slot and per role, so front of house and kitchen can carry different numbers within the same service. See how it works at Sked Solve.