Staff scheduling against the trading curve

Staff scheduling in most restaurants starts with last week's rota. The manager opens it, changes the names of anyone on holiday, adjusts a shift or two and publishes it. It is quick, and it keeps the team's routine steady. It also carries forward every old habit: the shift that has always started at five, the closing team that has always been four people, the lunch that has always had three servers whatever the trade.

The alternative is to build the rota from the trading curve: the shape of the venue's trade through each day, hour by hour. Every venue has one. Trade builds slowly at the start of a service, peaks, and tails off. A rota that follows that curve puts people where the guests are, and fewer where they are not. A rota copied from last week often has too many people at the edges of each service and too few at the peak.

This guide sets out how to do staff scheduling against the trading curve. It explains how to find the curve from sales by hour and covers, how to turn it into a staff rota, the two shifts most venues should trim first, how scheduling from forecast fits in, how to protect service at the peak, and how to measure the effect on labour cost percentage and on the guests.

The trading curve is the pattern of trade through a service or a day. For a typical dinner service, it might show a few covers at 18:00, a steady build through 19:00, a peak between 20:00 and 21:00, and a fall after 21:30. For lunch, it might show a sharp peak between 12:30 and 13:30. Weekends and weekdays usually have different curves, and so do seasons.

The curve is the most useful single picture for staff scheduling, because labour should rise and fall with it. Large operators use workforce management systems to match staff to demand this way; a single venue can do the same with a few weeks of hourly figures and a sheet of paper. A rota that matches the curve has enough people at the peak and fewer at the edges.

Each venue's curve is its own. A city-centre lunch spot, a neighbourhood bistro and a late-night bar all have very different shapes.

The easiest way to find the curve is from sales by hour: the sales rung in each hour of each service, from the till or point of sale. Look at a few weeks of the same day, such as four Saturdays, and average the sales for each hour. The result is the Saturday curve.

Covers by hour are even better for staffing, because the work of serving depends on the number of guests more than on how much they spend. Bookings by time, plus walk-ins, show when guests arrive; the length of a typical sitting shows how long they stay.

Our guide to sales by hour: the shape of your trading day covers how to read the hourly picture.

Three curves matter for staffing, and they peak at slightly different times. Arrivals peak first, when most guests sit down, which is when the host and servers are busiest greeting and taking orders. Seated guests peak a little later, when the room is fullest, which is when the floor team is busiest overall. The kitchen peaks when most orders are being cooked, often half an hour or so after the arrivals peak.

Schedule each part of the team for its own curve. Hosts and servers need to be there for arrivals; runners and the floor team for the full room; cooks for the kitchen's peak. A rota that brings everyone in at the same time misses all three.

To turn the curve into a staff rota, decide how many of each role each hour needs. A simple rule of thumb works: one server for every so many seated guests, one cook per station at the peak and fewer at the edges, a host for the arrival period, and so on. The exact ratios depend on the venue's style of service.

Then draw the shifts so the number of people on duty follows the curve. Staggered start times bring people in as trade builds; staggered finish times let them go as it falls. A dinner team might start in waves at 17:30, 18:30 and 19:30, and finish in waves from 22:00.

Check the result against the curve: at each hour, are there about as many people as the trade needs?

The first shift most venues should trim is the early start: the hour or more before the first guests arrive when the full team is already on duty. It happens because shifts are written in round blocks, starting at a fixed time whatever the bookings, and because "we have always started at five".

Some early work is genuine: setting up the room, preparing stations, briefing the team. But it rarely needs everyone. Work out what must be done before service and who needs to do it, then bring the rest of the team in closer to when guests arrive.

Look at the arrivals curve. If the first significant wave of guests arrives at 19:00, a server who starts at 17:00 has two hours of setting up for one hour's need.

The second shift to trim is the late tail: the end of service on quieter nights, when the full team stays until the last table leaves. On a busy Saturday, the tail is short and the team is needed. On a quiet Tuesday, the last two tables may keep four servers and two cooks for an hour.

Plan the tail. Decide who stays to the end and who can finish as the room empties, and let the floor manager release staff as trade falls. A clear plan, agreed before service, avoids the awkward moment of deciding who goes home while guests are still eating.

Watch for the kitchen's tail too. Once the last orders are in, not every station needs a cook.

Trimming the edges is only half the job. The other half is making sure the peak has enough people. Many rotas copied from last week are thin at the busiest moment, because hours have been spread evenly across the service rather than concentrated where they are needed.

Move hours from the edges to the peak. A server who starts later can also finish later, covering the busiest part of the evening instead of the quietest. A cook who comes in for the rush rather than the whole service adds capacity where it matters. The total hours may not change much, but the service gets better.

Watch service measures at the peak: ticket times, the gap before guests are greeted, complaints. If they improve after moving hours, the move was right.

The trading curve shows the typical shape; the forecast shows next week's size. Scheduling from forecast means combining the two: taking next week's covers forecast for each service and applying the usual shape, so the rota reflects both how many guests are expected and when they will come.

A Saturday expected to be busier than usual needs more people at the peak; a Tuesday expected to be quiet needs fewer at the edges. Our guide to scheduling from forecast, not from memory covers how to turn the forecast into shifts, including a buffer for walk-ins.

Every service needs a minimum team, whatever the trade: enough to run the kitchen safely, serve guests properly and handle the unexpected. On very quiet services, the minimum may be more than the curve suggests. That is fine; the minimum is a fixed cost of being open.

Write down the minimum for each service, and never schedule below it. Then build up from the minimum as the curve requires. Knowing the minimum also shows what level of trade a service needs to cover its fixed staffing.

Following the curve sometimes means shifts that do not fit neatly into a day: a short shift for the lunch peak, a split shift that covers lunch and dinner with a break, a late start for the dinner rush. These can help match staffing to trade, but they affect people's lives, and local rules may limit them.

Use them with care and consent. Some staff like short shifts or split days; others find them hard. Balance the curve with what makes the rota fair and sustainable for the team, because a rota that drives people away costs more in recruitment than it saves.

A rota is not only a cost plan; it shapes people's lives. Staff need predictable hours, fair shares of good and bad shifts, enough rest and time off they can plan around. A rota that follows the curve perfectly but changes every week, or always gives the same people the late finishes, will lose good staff.

Publish rotas in good time, keep patterns as stable as the trade allows, rotate the less popular shifts fairly and listen to requests. Our guide to staff retention and the four reasons people leave covers why this matters.

Weekdays and weekends usually have different curves, and the rota should treat them differently. A weekday dinner may build slowly and peak modestly; a Saturday may fill quickly and stay full for hours. A weekday lunch may have a sharp, short peak driven by office hours; a weekend lunch may be longer and more relaxed, with families and groups.

Build a separate curve for each day of the week, or at least for weekdays, Fridays, Saturdays and Sundays. Copying the Saturday pattern onto a Wednesday, or the other way round, is one of the commonest causes of overstaffing and understaffing.

The curve changes with the seasons. Summer evenings on a terrace may peak later; winter evenings earlier. Holiday periods may bring busy lunches to a venue that is usually quiet at midday. In the Gulf, Ramadan moves the whole day, with a sharp peak at iftar and later trade into the night.

Rebuild the curves at the start of each season from recent weeks, rather than relying on a pattern from months ago. A curve that was right in spring can be wrong by midsummer.

A rota that follows the curve must still give staff proper breaks and rest. Local rules often set minimum breaks and rest periods, and good practice goes further: tired staff make mistakes and leave. Plan breaks into the quieter parts of each service, before the peak and after it, so the team is fresh when it matters.

Staggered starts help here too. A server who starts later can take a break later, keeping the floor covered at the busiest moment.

Scheduling against the curve is easier when staff can do more than one job. A host who can also serve, a runner who can cover the bar, a cook who can work two stations: each gives the manager flexibility to match people to the trade without adding hours. When the arrivals peak passes, the host can help on the floor; when the bar quietens, the bartender can help run food.

Train staff across roles where it makes sense, and recognise the extra skills. Flexibility makes the rota cheaper to run and the work more varied for the team.

Sometimes the curve shifts for a reason: a new menu that changes how long guests stay, a promotion that fills the early evening, a nearby office that closes or opens. Watch for these changes in the hourly figures, and update the rota when the shape moves. A rota built around last year's curve can be wrong for months before anyone notices.

Staff scheduling against the curve shows up in labour cost percentage, especially when it is read by service. Trimming the early start and the late tail usually lowers the percentage on quieter services, while protecting the peak may raise it slightly on the busiest ones, which is often the right trade.

Our guide to labour cost percentage, by service not by month covers how to split the figure and read it.

Before publishing the rota, check it against three things: the curve for each service, the forecast for next week, and the labour budget. At each hour, is the number of people on duty about right for the expected trade? Does any service have a full team before guests arrive, or after they leave? Is the peak properly covered?

A quick visual check helps: plot the number of staff on duty by hour against the expected covers by hour. Where the lines diverge, adjust.

No rota is perfect, because trade varies. Give the manager on duty the authority to adjust on the day: calling in an extra person when bookings surge, releasing someone early when a service is quiet. Agree in advance who can be called in and who is happy to finish early, so adjustments are fair.

Record the adjustments and why. Over time, they show where the rota is consistently wrong.

Measure the effect of scheduling against the curve on both cost and service. On cost: hours worked by service, labour cost percentage by service and sales per labour hour. On service: ticket times at the peak, guest feedback, and how often the team felt stretched.

Compare a few weeks before and after the change. A good result is lower hours at the edges, the same or more at the peak, a slightly lower overall labour cost, and service that is as good or better.

Here is an illustrative Tuesday dinner, before and after scheduling against the curve. The figures are examples.

servearound does not build rotas or track hours. What it provides is the curve. Payment analytics shows revenue by hour, order analytics shows when checks are rung across the week, and reservation analytics shows covers by day and hour, so the shape of each service is clear. The table report shows when the room fills, and kitchen analytics shows ticket times by day and hour, which marks the kitchen's own peak.

The book shows next week's bookings by time, so the expected arrivals for each service can be read straight from it. See Analytics.

The pattern of trade through a service or a day, hour by hour, from arrivals to the last tables.

From sales by hour and covers by hour, averaged over several weeks of the same day.

The early start before guests arrive and the late tail after the room empties on quiet nights.

Not if hours are moved to the peak rather than simply cut. Watch service measures as you change.

Weekly, from the curve and the forecast, rather than copying last week's. Keep the core pattern stable and adjust the edges.

It helps a great deal. Most tills can show sales by hour, and covers by hour from bookings make the curve even clearer for staffing.

Balance the curve with fairness and predictability, because a rota that drives staff away costs more than it saves.

Yes. The kitchen's peak usually comes a little after the arrivals peak, so cooks and servers need slightly different start and finish times, and the head chef should help build the kitchen's part of the rota. Agree the two together, so the floor and the kitchen are ready for the same peak.

Staff scheduling works best against the trading curve: the hourly shape of each service, found from sales by hour and covers by hour. Schedule hosts and servers for arrivals, the floor for the full room and the kitchen for its own later peak, using staggered starts and finishes so the number on duty follows the trade.

Trim the two shifts that waste most: the early start before guests arrive and the late tail on quiet nights. Move hours to the peak, respect the minimum team, combine the curve with scheduling from forecast, keep the staff rota fair and predictable, and measure the effect on labour cost percentage and on service.