Restaurant & bar shift scheduling: the complete guide

Restaurant demand is not a smooth curve. It is two or three sharp spikes with long valleys in between. A city-centre restaurant might seat 45 covers on a Tuesday lunch and 140 on a Friday dinner spread across two seatings. A schedule built around vague blocks like morning and evening hides exactly the difference that matters.

Break the week into services instead: prep, lunch, the afternoon lull, dinner and close. Each one has its own staffing profile. Weekends behave like a different business altogether — in many venues a single Saturday brings in more than several weekday services combined, with larger parties, longer tables and more bar work.

Start from your own numbers, not from intuition. Pull covers or tickets per service for the last few weeks, split by day of week, and write them down. Even a simple table of covers per service beats scheduling from memory, because memory overweights last week’s disaster and forgets the quiet Tuesdays.

A split shift covers both peaks with one person: a waiter works 12:30 to 16:00, goes home or waits, and returns from 20:00 to midnight. On paper it is efficient — you pay seven and a half hours and your strongest server works both services instead of one.

The cost lands on the employee. That same day runs eleven and a half hours door to door, with a dead gap in the middle that is only useful if they live close by. Teams with heavy split-shift loads tend to tire faster and leave sooner, and replacing a trained server mid-season is far more expensive than the hours you saved.

Use splits deliberately, not by default. Reserve them for people who live near the venue or genuinely prefer them, rotate who carries them, and mix in peak-only part-time shifts: a four-hour dinner-only shift for a student is often better than splitting a full-timer’s day. Whatever you decide, track who has taken splits recently — fairness is remembered.

Front of house means hosts, waiters, runners and bartenders; back of house means the kitchen line and the kitchen porter. The skills are not interchangeable: an extra waiter does nothing for a missing grill cook, and a brilliant chef cannot cover the bar on a Saturday night.

They also run on different clocks. The kitchen arrives hours before doors open — a cook starting at 10:00 for a 13:00 service is doing prep, not waiting around — while front of house arrives shortly before the first guests. At close it reverses: the kitchen shuts down early, the floor stays with the last tables, and the porter is often the last one out.

So schedule by role and skill, never by headcount. Three staff at dinner is not a plan; one grill, one cold section, one porter, three waiters and one bartender is. Keep a simple skills matrix — who can run the pass, who can open alone, who is trained on the till — and check it before you publish, not during service.

Daily totals hide the gap that hurts you. Six people on Friday can mean three staff standing around from 16:00 to 18:00 and the same six drowning from 21:00 to 23:00. The total was right; the service still failed.

Define what you need per role and per two-hour band. A realistic Friday might need four waiters, one bartender, three cooks and one porter from 20:00 to 22:00, but only two waiters, one bartender, two cooks and one porter from 22:00 to midnight. Lay the actual shifts over those needs and gaps and surpluses become visible at a glance.

This is exactly how ShiftCal models coverage — needs are set per role and time band, and gaps are flagged while you build the week — but the principle works on a whiteboard too. The point is to stop asking how many people work today and start asking who covers nine to eleven.

Publishing late is the quiet killer of hospitality teams. People plan childcare, second jobs and studies around their shifts; when the rota lands two days before the week starts, absences, last-minute swap requests and resentment all rise together.

A practical target: publish two weeks ahead and never less than one. Some jurisdictions also set minimum notice periods for schedules or schedule changes, so check the rules that apply where you operate. Then protect the published week with a freeze window — inside the last 48 hours, changes happen only by mutual agreement or true emergency.

Templates make early publishing achievable. If opening, lunch, split, dinner and close shifts exist as reusable patterns with role, times and breaks attached, building next week is assembly rather than invention — and the team learns the rhythm, which by itself reduces confusion and missed shifts.

Swaps are inevitable in hospitality, and informal ones are dangerous. When two waiters trade a Saturday over a private chat, the manager finds out at 20:00 when the wrong person — or nobody — shows up. The fix is not banning swaps; it is giving them one formal channel.

A sound flow has three steps: one employee proposes, the colleague accepts, the manager approves — with the system checking that the replacement has the right skills and is not running into excessive hours. For no-shows, decide the escalation before it happens: an open shift offered to qualified, available staff, first to claim it gets it, beats calling down a list at 19:45. ShiftCal handles swaps and open shifts this way, with every step recorded.

Finally, read the pattern behind repeated no-shows. They cluster around late-published rotas, chronic split shifts and always-the-same-person weekend duty. Fixing the cause is cheaper than staffing around the symptom.

A break in the middle of the Friday rush is not rest, and quietly skipping breaks is neither sustainable nor, in many countries, legal. Minimum break rules differ by country and by shift length, so treat them as a configurable rule of your operation, not an afterthought.

The craft is in the staggering. Kitchens typically eat before service — a staff meal at 12:00 ahead of a 13:00 opening — while the floor takes breaks in the 16:00 to 18:00 lull. Never everyone at once: define a minimum floor per band, for example at least two on the floor and one on the line even during break windows.

Plan breaks into the schedule explicitly, marked as paid or unpaid according to your policy, and record the breaks actually taken. Otherwise worked hours, overtime and labour cost are all quietly wrong.

Seasonal venues live two different lives. A beach restaurant serving 60 covers a day in May can be doing 200 a day in August. The workable model is a stable core team that runs the venue year-round, plus a seasonal layer hired for the peak.

Hire and train before the peak, not during it. Seasonal staff who start two weeks early learn the menu, the till and the station layout while mistakes are still cheap. Keep notes on who performed well and stay in touch — rehiring a proven seasonal worker next year is faster and safer than a new hire.

In low season, shrink the pattern rather than thinning every shift into uselessness. Shorter opening hours with properly staffed services beat a skeleton crew limping through a full schedule. Quiet months are also the time for holidays, cross-training and fixing everything the summer broke.

The number that matters is labour cost as a percentage of sales: total wages including employer social costs, divided by sales over the same period. Track it weekly and, ideally, per service. Healthy levels vary widely by concept and country, so benchmark against your own history rather than chasing a magic number.

Cost the schedule before publishing it. Estimate expected sales per service from recent weeks, price the scheduled hours, and look at the ratio. If Tuesday dinner reliably brings in modest revenue and you have six people rostered that evening, the arithmetic — not instinct — says trim to four or shorten the shifts. Comparing scheduled versus clocked hours afterwards catches the drift of early arrivals and unplanned overtime.

Cut the shoulders, not the peak. Start a shift 30 minutes later, end it when the last tables leave, merge two thin afternoon shifts into one. Understaffing the Friday rush to save hours costs more in lost sales, bad reviews and exhausted staff than it will ever save in wages.