Scheduling from forecast, not from memory

Scheduling from forecast means building next week's rota from an estimate of how busy each service will be, rather than from memory, habit or last week's rota. It sounds obvious, yet most venues schedule from memory. The manager knows that Saturdays are busy and Tuesdays are quiet, and writes the rota accordingly. That works, roughly, until a quiet Saturday leaves a full team standing around or a Tuesday with a large party catches the team short.

A forecast makes the rota specific. Instead of "Saturday is busy", it says "Saturday dinner is expected to be about 95 covers, mostly between 19:30 and 21:00". That tells the manager how many people each role needs and when. And because the forecast is written down, the rota can be checked afterwards against what actually happened, so next week's is a little better.

This guide sets out how to do scheduling from forecast, step by step. It covers the covers forecast, how to read next week's booking shape, the buffer for walk-ins and late bookings you cannot see yet, how to turn covers into hours for each role, how to build the staff rota, how to adjust during the week, and how to measure whether the rota matched the trade.

Memory is good at the general pattern and bad at the details. It remembers that Saturdays are busy, but not that this Saturday has fewer bookings than usual. It remembers last December, but not how the week before a public holiday behaved. It is also biased: busy nights that went badly are remembered vividly, so managers overstaff to avoid them, while quiet nights with too many staff are forgotten.

A forecast corrects those biases. The general practice of demand forecasting exists precisely because memory is a poor guide to next week's demand. It uses the venue's own figures, the diary and known events, and it produces a number that can be checked. Over time, the checks improve the forecast, which memory alone never does.

The first step is the covers forecast: the expected number of guests for each service next week. It combines the recent baseline, the seasonal pattern from last year, the bookings already in the diary and anything known about the week ahead, such as events or weather.

Our guide to sales forecasting for next week in twenty minutes covers the method, and our guide to a covers forecast your kitchen will believe covers how to build it by service, including walk-ins and no-shows.

For scheduling, covers matter more than sales, because the work of serving depends on how many guests there are.

The second step is the booking shape: when, within each service, the booked guests are expected to arrive. A service of 95 covers with most arriving between 19:30 and 20:30 needs a very different rota from one with arrivals spread from 18:00 to 21:30.

Read the shape from the diary: how many covers are booked into each half hour of each service next week. Compare it with the usual shape for that service, because bookings still to come tend to follow the usual pattern. Large parties deserve special attention, because one party of twelve arriving at 20:00 changes the peak.

The booking shape is what lets scheduling from forecast place people at the right times, not just in the right numbers.

The third step is the buffer: an allowance for walk-ins and late bookings that are not yet in the diary when the rota is written. A rota written on Thursday for the following week sees only part of next week's guests. Walk-ins, same-day bookings and late group enquiries will add more, especially at weekends and on sunny days.

Build the buffer from experience: how many covers each service usually gains between the day the rota is written and the service itself, and how many walk-ins it usually sees. Add it to the forecast, and spread it across the service in the usual walk-in pattern.

Keep the buffer flexible rather than fully staffed. A person on standby, or a shift that can be extended, covers the buffer without paying for it on nights it does not appear.

The fourth step turns covers into hours for each role. For the floor, a common approach is a ratio: one server for a certain number of seated guests at the peak, adjusted for the style of service. For the kitchen, it is usually per station: how many cooks each station needs at the expected peak, and fewer at the edges. For hosts, runners, bartenders and support roles, similar simple rules apply.

Work out the number of people each role needs in each hour of each service, from the forecast covers and the booking shape. That gives a staffing profile: a target number on duty for every hour.

Our guide to staff scheduling against the trading curve covers how arrivals, seated guests and the kitchen peak at different times.

The fifth step is building the staff rota so that the number of people on duty matches the staffing profile. Use staggered starts and finishes, so people come in as trade builds and leave as it falls. Respect the minimum team for each service, the team's contracted hours, rest periods and fair shares of popular and unpopular shifts.

Then check the rota against the profile hour by hour. Where there are too many people, such as before the first guests arrive, move start times later. Where there are too few, such as at the peak, add a shift or extend one.

Publish the rota in good time, so the team can plan their lives around it.

A rota built purely from the forecast could change dramatically every week, which is hard on the team. Balance accuracy with stability. Keep a core pattern of regular shifts for each role, so staff know roughly when they will work, and use the forecast to adjust the edges: who starts earlier or later, who is on standby, who finishes when the room empties.

Rotate the less popular shifts fairly, honour reasonable requests, and give notice of changes. A team that trusts the rota is more willing to flex when the forecast asks it to.

The forecast keeps changing after the rota is published. Bookings arrive, large parties confirm or cancel, the weather forecast changes. Check the diary each day for the next few days, and compare it with the forecast the rota was built on.

If a service is running well ahead of the forecast, call in the person on standby or extend a shift. If it is running well behind, agree with staff who can finish early or swap to another day. Make these changes as early as possible, so the team is not surprised on the day.

On the night, the manager on duty makes the final adjustments: releasing someone early when the room empties, asking someone to stay when a late group walks in. Give them clear authority and a simple plan, agreed before service, for who goes first and who stays.

Record what happened: covers by hour, staff by hour, and any changes. That record is what improves next week's rota.

After each week, compare the rota with what happened. For each service, look at forecast covers against actual covers, staff hours scheduled against hours worked, and sales per labour hour. Look at the peak: was it properly staffed? Look at the edges: were people on duty without guests to serve?

Look at service measures too: ticket times, how long guests waited to be greeted, complaints and feedback. A rota that saves hours but lets service slip is not working.

Note the patterns. If Saturdays are always under-forecast, raise the Saturday buffer. If Tuesdays always have too many people at the start, move start times later. Each week's comparison makes the next rota better.

Not every role rises and falls with guests. Morning preparation, deliveries, cleaning, administration and maintenance follow their own rhythm, set by the menu, the suppliers and the building rather than by bookings. Schedule these roles from their own needs: the prep list, the delivery days, the cleaning routine.

The forecast still helps. A busy weekend ahead means more preparation on Thursday and Friday; a quiet week may allow deep cleaning or training. Link the preparation rota to the covers forecast, even if the people involved never see a guest.

Staff holidays and leave have to fit around the forecast. A rota built for a busy week with two key people on holiday will be stretched; one built for a quiet week can absorb absences more easily. Look at the forecast when approving leave requests, and plan cover for the busiest weeks well in advance.

Keep a simple calendar of approved leave beside the forecast, so the person writing the rota sees both at once.

New staff need training shifts, usually alongside an experienced colleague, before they work alone. Those shifts are real hours, and they should be scheduled deliberately, ideally on services that are busy enough to learn from but not so busy that training is impossible.

Use the forecast to choose them. A moderately busy weekday dinner is often better for training than a quiet Monday, where there is little to learn, or a packed Saturday, where there is no time to teach.

The kitchen rota follows the forecast slightly differently from the floor. Cooks are needed for preparation before service, for the kitchen's peak, which comes a little after the arrivals peak, and for closing down. Large parties and set menus can change what each station needs.

Share the forecast and the booking shape with the head chef, who knows how many cooks each station needs at each level of trade. Our guide to pacing covers covers how bookings affect the kitchen's peak.

Publish the rota on a fixed day each week, far enough ahead for the team to plan, using the latest forecast. Then limit changes after publication to genuine needs: a large booking, a forecast that has moved sharply, an absence. Tell affected staff as early as possible.

A rota that changes constantly after publication loses the team's trust. A rota that never changes, whatever the bookings, loses money. Aim for a stable rota with a small, agreed number of flexible shifts that absorb the changes.

In a group, each venue should schedule from its own forecast, because each has its own pattern of trade. A group-wide rule, such as a fixed number of servers per venue, ignores the difference between a busy city site and a quiet suburban one. Shared staff who work across venues can help absorb the peaks, if their shifts are planned from each venue's forecast.

Share the method across the group, so every manager builds the forecast and rota the same way, and the results can be compared fairly.

Scheduling from forecast shows up in labour cost. When the rota matches the trade, hours on quiet services fall, the busy services are properly covered, and labour cost percentage becomes steadier across the week. Our guide to labour cost percentage, by service not by month covers how to read the effect by service.

Set a labour budget for the week from the sales forecast, and check the rota against it before publishing. If the rota is well over budget for the forecast trade, look for hours at the edges that can move.

Venues that do not take bookings can still schedule from forecast. The forecast relies more on the recent baseline, last year and known events, and the shape comes from the usual pattern of arrivals for each service, from sales by hour. The buffer is larger, because there is no diary to show what is coming.

Track walk-ins carefully and note what affects them, such as weather and local events. Over time, the forecast becomes reliable enough to schedule against with confidence.

Large parties and private events change the rota. A party of twenty may need an extra server and an extra cook for the duration, and may need people at a different time from the main service. List them separately in the forecast, with their timing and menu, and schedule for them specifically.

Confirm numbers a few days ahead, and adjust the rota if they change. A party that grows or shrinks by ten changes what the team needs.

Scheduling from forecast is the most practical use of sales forecasting a venue can make. The forecast earns its keep when the rota built from it puts people where the guests are. A venue that forecasts but still writes the rota from memory gets little from the forecast; one that schedules from it gets the benefit every week.

Here is an illustrative Friday dinner, scheduled from forecast. The figures are examples.

servearound does not write rotas or track hours. What it provides is the forecast's raw material and the booking shape. The book shows next week's bookings by service and time, with party sizes, and the live waitlist shows walk-in demand on the night. Reservation analytics shows covers by day and hour, how much notice guests give and no-show rates, which are what the buffer is built from.

Payment analytics shows revenue by hour, so sales per labour hour can be calculated for each service afterwards. See Analytics and the reservations module.

Building the rota from an estimate of covers and their timing for each service, rather than from memory.

Based on how many covers each service usually gains after the rota is written. Keep it flexible, such as a standby shift.

As early as the forecast allows while still being reasonably accurate, often a week ahead, with adjustments as bookings change.

No. A simple sheet with forecast covers and a staffing profile is enough to start.

Adjust during the week and on the night, record what happened, and improve the forecast.

Keep a stable core pattern, rotate the less popular shifts and use the forecast to adjust the edges.

The manager closest to each team, usually the general manager for the floor and the head chef for the kitchen, both working from the same forecast.

No, it supports it. The best rotas combine the forecast's numbers with the manager's knowledge of the team and the venue.

Explain the reason, keep a stable core of regular shifts, and change the edges gradually. Most teams accept a rota that clearly follows the trade, especially when the busiest nights become less stretched.

Usually, by trimming hours at the quiet edges of services. It also improves service at the peak, which is often worth as much as the hours saved.

A few weeks of measuring forecast against actual is usually enough to see the pattern and set sensible buffers for each service. Keep measuring after that, because patterns shift with the seasons.

Scheduling from forecast replaces memory with a number. Start with the covers forecast for each service, read the booking shape from the diary, and add a buffer for the walk-ins and late bookings you cannot see yet. Turn covers into hours for each role, build a staffing profile by hour, and draw the staff rota to match it with staggered starts and finishes.

Balance accuracy with fairness and stability, adjust during the week as bookings change, give the manager on the night clear authority, and compare the rota with what happened every week. Staff scheduling built this way puts people where the guests are, and it is the most practical use of sales forecasting a venue can make.