Ordering from a forecast rather than a feeling

The head chef places the fish order on Monday evening, the way he always has: a look in the walk-in, a feeling about the week, the same as last week plus a bit for safety. By Thursday, half the sea bass is going in the staff meal. The following week, he orders less, and runs out of it by eight on Friday. Ordering from a forecast rather than a feeling takes the guesswork out of that decision. It turns next week's expected covers into the quantities each dish will need.

The idea is simple. If you know roughly how many guests are coming, and roughly what share of them order each dish, you know roughly how much of each ingredient the week will use. Subtract what is already on the shelf, add a sensible buffer and you have an order. The forecast does not have to be perfect; it only has to be better than a feeling, which it usually is.

This guide covers how to turn a covers forecast into an order list, how to handle the ordering schedule and par levels, the two kinds of ingredient that should ignore the forecast and how to check each order against what was actually used. Every venue and figure in it is an example.

Ordering from a feeling goes wrong in both directions. On a quiet week, the feeling says "the usual", and the kitchen over-orders, throwing away fresh food or freezing what should have been sold fresh. On a busy week, the same feeling says "the usual", and the kitchen runs out of popular dishes halfway through service.

Feelings also tend to remember the last bad night. A chef who ran out on Saturday orders extra for every day the following week, and the waste grows. Over a year, these swings cost a great deal in waste, in shortages and in the cash tied up in stock that is not needed.

Ordering from a forecast replaces the feeling with a few steps: the expected covers for each service, the expected share of each dish, the quantity of each ingredient in each dish, the stock already on hand and a buffer for uncertainty. The result is an order that matches the week ahead rather than the week just gone.

It sounds like a lot of arithmetic, and the first time it is. Once the dish shares and recipe quantities are written down, each week's order takes little more than updating the covers and the stock count.

Everything starts with the covers forecast: how many guests you expect for each service in the period the order must cover. Sales forecasting and demand forecasting provide it, from recent weeks, bookings already made, the season and local events.

If the forecast is good, the order will be good. If it is not, the order will inherit its errors. Our guide to demand forecasting without a data scientist covers how to build a forecast that is reliable enough to order from.

The next step is the dish mix: the share of covers that order each dish. If, over the last few weeks, about one guest in five ordered the sea bass, then a forecast of 300 covers suggests about 60 portions of sea bass.

Work out the mix from your sales records for the last few weeks, dish by dish, and update it when the menu changes. The mix is usually more stable than the number of covers, but it shifts with the seasons, with specials and when a dish is moved on the menu.

From the portions of each dish, work out the ingredients. If a portion of sea bass uses one fillet, sixty portions need sixty fillets. If a portion of risotto uses 80 grams of rice, a hundred portions need eight kilos. For ingredients used in several dishes, add up the quantity across all of them.

This needs a written recipe for each dish, with the quantity of each main ingredient per portion. Many kitchens have these for costing already. If yours does not, start with the dishes that use the most expensive or perishable ingredients.

Add up the quantities for each ingredient across all the services the order must cover, usually the days until the next delivery. The result is the amount the kitchen expects to use in that period.

Keep it to the ingredients that matter. There is no need to forecast every pinch of salt. Focus on the expensive, the perishable and the ones that cause trouble when they run out, such as the main proteins, fresh fish, dairy and fresh produce.

Subtract what is already on the shelf. If the week needs sixty fillets and twelve are in the walk-in, order forty-eight, before any buffer. That means counting the stock of each key ingredient before placing the order, which many kitchens skip.

A quick count of the main items before each order is enough. It does not need to be a full stocktake. Our guide to how to run a restaurant stock count properly covers counting when accuracy matters more.

Forecasts are never exact, so add a buffer, often called safety stock: a little extra to cover a busier week than expected. The size of the buffer should match how accurate the forecast usually is and how bad it would be to run out.

For an ingredient where running out would hurt, such as the main protein of a best-selling dish, a larger buffer makes sense. For one with a short shelf life where waste is costly, a smaller buffer, and a plan for what to cook if it runs low, may be better.

Suppliers sell in pack sizes: a case of twelve, a five-kilo bag, a box of fillets. Round each quantity to the nearest sensible pack, usually up for ingredients you cannot do without and down for those with a short life that you can substitute if needed.

Rounding is where a lot of over-ordering creeps in. If the forecast suggests half a case, decide deliberately whether to order a whole case or find another use for the rest, rather than rounding up by habit every time.

The ordering schedule, when you order and when deliveries arrive, shapes every order. An order placed on Monday for Tuesday delivery must cover the kitchen until the next delivery, perhaps Thursday. Another placed on Wednesday must cover Thursday to Saturday, the busiest days.

Know each supplier's order deadline and delivery days, and plan each order to cover exactly the period until the next delivery. Ordering too far ahead ties up cash and space; ordering too little leaves the kitchen short before the next van arrives.

Each order covers a window: from this delivery to the next. Forecast covers for each service in that window, and work out the ingredients for those services only. A Thursday delivery covering Thursday to Saturday needs a much larger order than a Monday delivery covering two quiet days.

Ordering by window makes the forecast directly useful. The weekly forecast, broken down by service, tells you how big each order needs to be, rather than ordering the same amount on every delivery day.

Par levels are the quantity of each item a kitchen aims to have on the shelf at the start of a period. Ordering to par means topping up to that level each time: if the par is ten and there are four, order six. Par levels are simple and work well for items used steadily.

Par levels and forecasts work together. A par level can itself be set from a forecast, and adjusted for busy and quiet weeks. Our guide to par levels: how much stock to hold, and why covers how to set them.

A practical approach is to use par levels for steady items and the forecast for items whose use rises and falls with covers. Dry goods, cleaning supplies and long-life items are ordered to par. Fresh proteins, fish, dairy and produce are ordered from the forecast.

For items in between, set the par level from the forecast: a higher par for busy weeks and a lower one for quiet weeks. That gives the simplicity of par levels with some of the accuracy of the forecast.

Not every ingredient should follow the forecast. Two kinds are usually better handled another way: long-life staples, which should be ordered to par, and the ingredients for dishes that are meant to sell out, which should be ordered to a fixed amount whatever the forecast says.

Treating them separately keeps the forecast where it helps most, on the fresh, expensive and variable ingredients, and stops the kitchen spending time forecasting things that do not need it.

Long-life staples, such as flour, rice, pasta, oil, tinned goods, frozen items and most wine and spirits, keep for weeks or months. A small change in next week's covers makes little difference to how much you should hold, and over-ordering them causes little waste, though it does tie up cash and space.

Order them to par, checked weekly, and ignore the forecast. The time saved is better spent on the fresh items where forecasting makes a real difference.

Some dishes are meant to be limited: a special made with a few fish from a particular boat, a dish of the day, a tasting course with a set number of portions, bread baked once each morning. Their quantity is decided by the kitchen, not by the forecast.

Order exactly what the kitchen has decided to make, and let the dish sell out when it is gone. Forecasting more of it would defeat the point. Tell the floor how many portions there are, so servers can let guests know when it is nearly finished.

Over ordering is the most common and most expensive ordering mistake. It fills the walk-in, ties up cash, leads to waste when fresh food goes past its best and encourages the kitchen to push dishes that need using up rather than those guests want.

Its usual causes are ordering from a feeling, rounding up by habit, ordering the same amount every delivery and adding extra after one bad night. Our guide to how much does restaurant food waste really cost covers what the waste adds up to.

Under ordering has its own costs: dishes running out mid-service, disappointed guests, last-minute trips to a cash-and-carry at higher prices and a kitchen under pressure. A dish that is off the menu by eight on a Friday loses sales and goodwill.

When a dish does run out, take it off the menu straight away so guests are not told after they order, and note it. If the same dish runs out often, the forecast or the dish mix for it is too low.

Specials and new dishes have no history, so the dish mix has to be estimated. Start cautiously, perhaps a little below what similar dishes sell, and watch how it goes in the first few days. Adjust the next order quickly.

A new dish that sells faster than expected is a good problem to have, and the next order can correct it. One that sells slower than expected and was over-ordered leaves the kitchen with ingredients to use up.

Busy weeks, holidays and local events change both the covers and sometimes the dish mix. A match night may bring more orders for quick dishes; a festive week may shift guests towards a set menu. Adjust the order for both.

Check the local events calendar when you forecast, and make sure the order covers the busiest services in the window. Our guide to local events: the match night nobody told the kitchen covers building the calendar.

After each delivery window, compare what you ordered with what was actually used and what was left. If you ordered sixty fillets, sold fifty-two and had eight left, the order was a little high. If you ran out, it was too low.

Keep a simple record for the key ingredients. Over a few weeks, patterns appear: some ingredients are always over-ordered, others always short. Fix the dish mix or the buffer for those, and the orders improve.

The order can only be as accurate as the forecast behind it. Track forecast accuracy alongside the order checks: if the covers forecast was ten per cent high, the orders will have been too, and the problem is the forecast rather than the ordering.

Separating the two helps fix the right thing. Our guide to forecast accuracy: measure it or stop forecasting covers how to score the forecast.

One person should own each order, usually the head chef or sous chef for food and the bar manager for drinks. They count the stock, update the forecast numbers, work out the quantities and place the order by the supplier's deadline.

Write the routine down, so someone else can do it when the owner is away. An order placed in a hurry by someone who does not know the routine is how over-ordering and shortages happen.

Suppliers often have minimum orders or delivery charges below a certain value. These can push the order up, especially on quiet weeks. Plan around them: combine items into fewer, larger deliveries on quiet weeks, or ask suppliers whether minimums can be relaxed.

Talk to suppliers about the forecast too. A supplier who knows a busy week is coming can make sure the stock is available, and one who knows a quiet stretch is coming may help with smaller deliveries.

Ordering from a forecast works best as a fixed routine: forecast covers on Monday, count the key items before each order, work out the quantities from the dish mix and recipes, adjust for stock and buffer, round to packs and place the order. After each window, check what was used.

Once the dish mix and recipe quantities are written down, the routine takes little more time than ordering from a feeling, and the waste and shortages fall steadily.

An invented bistro orders fish twice a week, by feeling, and either wastes sea bass or runs out of it. The head chef starts ordering from the forecast. For the Thursday to Saturday window, the forecast is 310 covers, and over the last month about one guest in five ordered the sea bass, so about 62 portions. There are 10 fillets in the walk-in, so he needs 52, plus a buffer of 6 for the busy Saturday: 58, rounded to five cases of twelve.

He orders rice, flour and wine to par, ignoring the forecast, and orders exactly 16 portions of the Saturday market fish special, which is meant to sell out. After the weekend, 4 sea bass fillets are left and nothing ran out. Over the next month, fish waste falls noticeably, and the Friday shortages stop.

servearound does not suggest orders: stock, recipes, suppliers and purchasing are not in servearound today, and inventory tracking is coming, so the order list in this guide is built in your own sheet. What it gives the sum is the demand side: the menu report shows units sold for every dish over any dates, which is the dish mix, and reservation analytics show covers by day and by meal period.

Bookings already made for the days ahead show on the reservations screen, date by date. When an ingredient runs out, an 86 on the till takes the dish off the digital menu at once, and in the menu library a sold-out dish can come back by itself tomorrow or stay off until you turn it back on. See Analytics.

Working out how much of each ingredient the coming days will use, from expected covers and the dish mix.

The share of covers that order each dish, taken from recent sales.

Enough to cover the forecast's usual error, more for ingredients you cannot run out of.

No. Focus on fresh, expensive and variable items. Order staples to par.

The quantity of an item to hold at the start of a period, topped up with each order.

Long-life staples, ordered to par, and the ingredients for dishes meant to sell out.

Compare each order with what was used and what was left, delivery by delivery.

One person per supplier area, following a written routine.

Ordering from a forecast rather than a feeling turns next week's expected covers into an order list: forecast covers for each service, apply the dish mix to get portions, use recipe quantities to get ingredients, add them up for the window until the next delivery, subtract what is on the shelf, add safety stock and round to pack sizes. Order each window to fit the ordering schedule and the supplier's delivery days.

Use par levels for steady items, and let two kinds of ingredient ignore the forecast: long-life staples, ordered to par, and dishes meant to sell out, ordered to the kitchen's fixed amount. Avoid over ordering by counting stock, rounding deliberately and resisting extra after one bad night. Check every order against what was used, and track forecast accuracy so you fix the right thing.