Rostering a Restaurant: Lunch, Dinner and Front of House

A restaurant roster is really two rosters that happen to share a building. The kitchen and the floor have different shift boundaries, different skill constraints, and very different answers to the question “what happens if someone doesn’t turn up?”

Running them as one list is the most common structural mistake in restaurant rostering, and it produces a specific, recognisable failure: a floor manager approving a kitchen swap they have no ability to evaluate.

The service structure

Most restaurants with a lunch trade run three rosters — two in the kitchen, one on the floor. The floor typically doesn’t warrant a separate lunch roster unless lunch is a genuine service rather than a handful of covers.

Note that dinner carries an extra line cook, an extra dishwasher and a pastry chef over lunch. That’s not padding. Dinner covers are higher, the menu is longer, and dessert is a real course rather than an afterthought — the roster has to reflect the menu, not just the hours.

“Line cook” is not a role

This is the single most useful correction to make to a restaurant roster, and it costs nothing.

Rostering four line cooks tells you the count is right and tells you nothing about whether service will work. Larder, grill, sauce and pass are not interchangeable. A cook who runs a flawless grill section may be genuinely unable to hold the sauce station on a Saturday, and finding that out at 19:30 is the worst possible time.

Name the section on the roster. Not “line cook ×4” but grill, sauce, larder, pass. It takes the same amount of time to write and it makes an unworkable swap visible before service instead of during it.

Service hours are not shift hours

The table shows lunch service at 10:00. Nobody in the kitchen starts at 10:00.

Prep starts one to two hours before doors, and pack-down runs 45 minutes to an hour past last table. A roster written to service hours systematically under-counts kitchen labour by around three hours a day, which is how restaurants end up simultaneously over-rostered on paper and short-handed at 09:00.

• Roster prep explicitly. If sauce needs two hours of prep before a Saturday dinner, that’s a rostered block with a name on it, not something the sous chef absorbs.
• Roster the close. The last dishwasher out at 23:45 is doing an hour of work that has to appear somewhere.
• Stagger the starts. Not everyone needs to be in for prep — a 14:00 start for the pass and a 15:00 for the grill is usually fine when doors are at 17:00.

Gate dinner. Don’t gate lunch.

The reason to require approval on dinner swaps and not lunch swaps isn’t seniority — it’s recoverability.

A lunch service running one line cook short is unpleasant and survivable. Covers are lower, the menu is shorter, and the head chef can step onto a section. A Saturday dinner running one sauce cook short is not survivable in the same way; you either refuse bookings or send out food you wouldn’t want your name on.

The floor works the same way in the opposite direction. Six waiters going to five on a Tuesday is a slightly slower service. Losing your sommelier on a Friday means nobody can sell wine, which is where the margin is.

Roles with no redundancy

Look at the roster above and count how many roles appear exactly once per service: head chef, pastry chef, floor manager, sommelier, host. Five single points of failure, and most restaurants have no plan for any of them.

You don’t need a second sommelier on the roster. You need to know, in advance, who covers wine service when yours calls in sick — and that person needs to have been told. A named backup for each singleton role is twenty minutes of thinking that saves a service.

The split shift problem

The gap between 15:00 and 16:00 in the table is inviting. It looks like you could roster one sous chef across both services and save a wage.

You can. It is also the fastest way to lose experienced kitchen staff. A 10:00 – 23:00 day with an hour in the middle isn’t a split shift in any meaningful sense — there’s nowhere to go and nothing to do, so it’s a thirteen-hour day at eleven hours of pay.

If you genuinely need a split, make it a real one — four hours or more between blocks, and rotate it so the same person isn’t carrying it every week. Check your local award or agreement too; split-shift allowances and minimum-break provisions exist precisely because this pattern was abused for decades.

Don’t under-roster the dish pit

Dishwashers are the first cut when labour looks high and the most expensive one you can make.

A backed-up dish pit doesn’t degrade gracefully. It stops the line — no clean pans means no plating, and within twenty minutes you have three chefs standing still and a floor manager apologising to a full room. The roster shows three dishwashers on dinner and two on lunch for a reason: dinner generates more plates per cover and more pans per dish.

Front of house can flex, so let it

The floor is the one part of a restaurant where genuine self-management works well, with two carve-outs.

Waiters swapping sections and shifts among themselves is normal and healthy — they know who wants the Friday and who needs the Tuesday off. The carve-outs are the sommelier and the floor manager, where the swap needs to be checked because there’s no second person holding the skill.

The practical version: let the floor trade freely, require approval on the two roles that can’t be covered, and make sure the person approving actually works the floor. A kitchen manager approving a wine service swap is theatre.

Putting it together

• Two kitchen rosters and one floor roster — not one combined list.
• Sections named on the roster, not just job titles.
• Prep and pack-down rostered as real hours.
• Approval on dinner and on any singleton role; open trading on lunch and general floor shifts.
• A named backup for every role that appears once.
• Dish pit staffed to plate volume, not to labour target.

The structure is not complicated and most head chefs already carry it in their heads. What usually goes missing is writing it down in a form where a swap request can be evaluated by someone who wasn’t in that conversation — which is where most restaurant rosters quietly fail.