How Much Should I Order? Par Levels vs Forecast Ordering for Restaurants

A Friday veg order that has to last until Tuesday should cover four days of expected use and a small buffer, less what is already in the cool room and on the truck, rounded up to whole bags. Restaurant ordering par levels handle that well for flour, oil and tinned tomatoes. For spinach, seafood and function menus, a forecast that is recalculated every order lands much closer.

Most kitchens order the way they always have. Someone walks the cool room on Thursday afternoon, eyeballs the shelves and texts the rep. Over-order and the herbs go slimy by Monday. Under-order and the lamb shoulder is 86’d at 8pm on Saturday. Neither shows up on the P&L until stocktake, by which point nobody remembers why.

What is a par level?

A par level is the amount of an item you want on hand after each delivery, set so you can get through to the next one without running out. If your par for 1L thickened cream is 12 and you have 5 in the walk-in, you order 7.

Pars work because they are simple. Anyone on the team can do the order: count, subtract from par, write the number down. A good par sheet is set out in shelf order, lists the pack size and is reviewed at least each season.

A sensible par is usually built from:

1. Average use per day over the last few weeks.
2. Days until the next delivery (the delivery cycle).
3. A safety buffer for a busy day or a late truck.

So par = average daily use × days between deliveries + buffer. In Excel, put daily use in B2, days between deliveries in B3 and the buffer in B4:

Where do par levels go wrong?

Par levels go wrong because they are a fixed number in a business where demand never is. A par set in a slow July will be short in a busy December, and one set in a big month leaves you carrying stock you do not need.

The common failure points:

• Delivery days are uneven. If your veg supplier comes Tuesday and Friday, the Friday order has to cover four days including the weekend; the Tuesday one covers three quieter days. One par cannot fit both.
• Bookings and functions are ignored. A par does not know there is a 60-guest birthday on Saturday. Our guide to planning orders and prep around bookings and functions covers how to build those in.
• Stock on order is forgotten. Someone orders against par on Monday, someone else on Tuesday, and the second order ignores what is already on the truck.
• Pars drift. They get bumped up after one bad night and never come back down.
• Pack sizes are fudged. You need 3kg of shallots but they come in 5kg bags, so the order is rounded in someone’s head, differently each week.

This matters because waste is expensive. Hospitality produces around 1.2 million tonnes of food waste a year, about 16% of Australia’s total, according to End Food Waste Australia. In the US, ReFED’s Restaurant Food Waste Action Guide puts pre- and post-consumer food waste at as much as 4.2% of total sales in an average restaurant. Some of that is plate waste, but a good share starts with stock that was ordered and never used.

How does forecast ordering work?

Forecast ordering works out how much you will use before the next delivery from what you expect to sell, then subtracts what you already have and what is already coming:

order = forecast use until next delivery − on hand − on order, rounded up to the pack size

The forecast use comes from recent sales of the dishes that use the item, adjusted for the day of the week, booked covers and any confirmed functions. It changes every order, which is the point.

A worked example

Say a bistro uses baby spinach in two dishes and a side. Its veg supplier delivers Tuesday and Friday, in 1.5kg bags, so the Friday order has to last until Tuesday morning.

Divide 6.4kg by the 1.5kg pack and round up: 4.27 becomes 5 bags (7.5kg). In Excel, with forecast in B2, buffer in B3, on hand in B4, on order in B5 and pack size in B6:

The MAX(0, …) part stops the sheet suggesting a negative order when you are already over-stocked. Compare that with a fixed par of 8kg: on a quiet weekend you would order the same 8kg regardless, and on a long weekend with a function you would be short.

Forecasting is moving from spreadsheet to software. In Toast’s survey of US restaurant decision-makers, 24% of operators said they already use AI for forecasting and demand planning, and 41% said they were extremely likely to adopt it. You do not need software to start, but the maths gets tedious across 200 lines and six suppliers.

Which items suit restaurant ordering par levels, and which need a forecast?

Use pars for items that sell steadily, keep well and are cheap enough that a bit of extra stock does not hurt. Use a forecast for items that are perishable, expensive, or swing with bookings and the season.

Good candidates for par levels:

• Dry store staples: flour, rice, sugar, tinned tomatoes, oils
• Cleaning chemicals and packaging
• Frozen items with a long shelf life
• Spirits and wine with steady sales (bar pars are standard practice)

Good candidates for forecast ordering:

• Fresh produce, herbs and leaves
• Fresh seafood and premium proteins
• Dairy with short dates
• Anything tied to a special, a seasonal dish or a function menu
• Bread and pastry with daily delivery

Most venues end up with a mix: pars for the long tail of cheap, steady lines, and a forecast for the smaller group of perishable lines where most of the spend and most of the waste sit. Start with the ten lines that cost you the most, not the whole list.

How do you set up forecast ordering without new software?

You can run a basic version with your POS sales report and a spreadsheet. It takes an hour or two to set up and about ten minutes per order once it is running.

1. Export four weeks of item sales from your POS, by day.
2. For your top ten perishable items, list which dishes use them and how much per portion.
3. Multiply forecast portions by the amount per portion to get forecast use per day.
4. Sum the days until the next delivery, then add a buffer.
5. Subtract on hand (from a quick count) and anything already ordered.
6. Divide by pack size and round up.

Check the forecast against what was actually used each week and tweak the buffer. If you keep throwing out the same item, the buffer is too big. If you keep running out, it is too small or the recipe quantity is wrong. The same sales forecast can drive your prep lists by station, and there is more on running orders this way in automated ordering and how to use it effectively.

How Restaurant AI Manager helps with how much to order

Restaurant AI Manager, Open Pantry’s AI restaurant operations software, builds a forecast order for every supplier, timed to their delivery days. (Open Pantry is an Australian ordering and operations platform for food and beverage suppliers and the restaurants they supply, founded in 2018.) It forecasts use from your POS sales, bookings and functions, subtracts what is on hand and on order, and rounds up to packs. Tap any line to see why it chose that quantity. For staples it can also order from pars.

Orders wait for your approval, and suppliers see nothing until you press Send, which is the main point in whether it is safe to let AI order stock. Any supplier can receive orders by email. Counts are done on a phone in shelf order, so the on-hand figure is current. Try it in the free live demo with no sign-up, or read more about stock control.

So, par levels or forecast ordering?

Use both. Restaurant ordering par levels are fine for steady, long-life staples. For perishables, functions and seasonal swings, order what you will use before the next delivery, minus what you have and what is already coming, rounded up to packs, and recalculate it every order.