How to do a restaurant stocktake: a weekly count that finds the missing money

A step-by-step restaurant stocktake: when to count, who counts, how to handle open packs and prepped items, and how to turn the gap between expected and counted stock into money — then find out why it is there. Includes a counting sheet you can copy.

Restaurant storeroom shelves next to a stock count sheet with quantities filled in

Most owners know the feeling: sales look healthy, yet at the end of the month the cash does not match them. A good part of the answer is usually in the storeroom — chicken used faster than it sold, oil running out ahead of schedule, a delivery that came in lighter than its invoice.

A stocktake is the only thing that turns that feeling into a number. Done badly, though, it produces a reassuring figure that is simply wrong. Here is how to do it properly.

The one question a stocktake answers

Not “how many kilos of chicken do I have?” The real question is:

How much should I have, how much do I actually have, and what is the difference worth?

  • Expected (theoretical) stock: the last count, plus everything received, minus what sales consumed according to the recipes, minus recorded waste.
  • Counted stock: what is on the shelf today.
  • Variance: the money you need to account for.

Without an expected figure to compare against, a count tells you what you own and nothing about where it is leaking.

When to count

Weekly, for the expensive lines. In most kitchens a handful of items account for most of the purchasing spend: meat, chicken, cheese, oil, and in a café coffee and milk. Count those every week. It takes about half an hour and catches most of the leakage.

Monthly, for everything. A full count closes the month’s food cost so your food cost percentage is built on real numbers rather than invoices.

Same day, same hour. Counting before opening one week and after closing the next makes the two counts incomparable. Pick a fixed slot — before service or after close usually works — and make sure no deliveries arrive during it.

Before you start: four things to set up

  1. Freeze receiving. Anything delivered mid-count scrambles the numbers. Have the supplier wait, or receive after you finish.
  2. Order the sheet the way the store is laid out. Shelf by shelf, then the fridge, the freezer and the line. An alphabetical sheet sends the counter zig-zagging and items get missed.
  3. One unit per item. If the recipe measures cheese in grams, count it in kilos, not in blocks. An item counted by the case one week and by weight the next produces a variance that is not real.
  4. Two people. One counts, one writes. Ideally one of them is not the person responsible for the store — not because anyone is suspected, but because a fresh pair of eyes sees what familiarity hides.

The count, step by step

1. Count blind

The most important rule. If the counter knows there should be 40 kg, they will find 40 kg. The counting sheet lists item, unit and location — nothing else. Expected quantities are compared afterwards.

2. Count sealed cases, weigh open ones

A sealed 10 kg case of chicken is counted as a case. An open case, a half-used tub of sauce or a cut wheel of cheese gets weighed. Eyeballing expensive items ruins the whole exercise.

3. Don’t forget the other locations

Stock lives in more places than the storeroom: the walk-in, the freezer, the prep line, the bar, the shelf behind the till. An item in two places goes on one line as a total.

4. Decide how to treat prepped items

Made-up sauces, dough, cut vegetables: either count them as prepped items with their own recipe, or estimate the raw ingredients inside them. Choose one method and use it every time. Counting them one week and ignoring them the next guarantees a false variance.

5. Enter the numbers the same day

A count sheet that sits in a drawer for a week loses most of its value, because every day of sales and deliveries makes the comparison harder.

6. Re-check the biggest gaps before you leave

Look at the five largest variances while you are still in the building. If something looks badly short, count it again. Sometimes the “missing” chicken is a case in the second freezer.

Calculating the variance: a worked example

Say the last count of chicken breast was 30 kg, you received 60 kg during the week, sales consumed 48 kg according to the recipes, and 2 kg of waste was recorded.

  • Expected: 30 + 60 − 48 − 2 = 40 kg
  • Counted: 36.5 kg
  • Variance: 3.5 kg short

At $9 a kilo (an illustrative price), that is $31.50 in one week, on one item.

Do the same for every expensive line and put them in one table:

ItemExpectedCountedVarianceUnit costVariance value
Chicken breast (kg)4036.5−3.5$9.00−$31.50
Mozzarella (kg)1817.2−0.8$11.00−$8.80
Frying oil (l)2521−4$2.50−$10.00
Bread rolls (each)400380−20$0.15−$3.00

Quantities and prices are illustrative.

The bread rolls show the biggest gap by count and the smallest by value. Rank variances by money, not by quantity, and work from the top down.

For the month as a whole, actual usage is simply:

Actual usage = opening stock + purchases − closing stock

Compare it with theoretical usage from sales and recipes. That gap is the difference between the food cost on paper and the real one.

What the variance is telling you

The variance itself is not the problem; not knowing its cause is. The usual suspects, roughly in the order most kitchens find them:

  1. Unrecorded waste. Burnt, dropped or spoiled food thrown away with no note. The easiest to fix: log waste with a reason as it happens and the unexplained gap shrinks. More in waste and yield.
  2. Portions above the recipe. The recipe says 150 g of chicken; the cook uses a handful. Watch the line during a rush, or give the station a scale or a sized scoop.
  3. Food that left without a bill. Staff meals, the owner’s guests, an order cooked and then cancelled. Record all of it, even at zero price.
  4. Short deliveries. Invoiced for 20 kg, delivered 18. Receive by weight, not by paperwork — see purchase orders and goods received.
  5. A wrong recipe. An item that comes up short by the same proportion every week may simply have a recipe written lighter than reality. Fixing it also fixes the plate cost and the price.
  6. Theft. Possible — but do not start there. Once the causes above are ruled out, what remains is worth investigating.

A positive variance matters too. Finding more than expected usually means sales recorded against the wrong item, or a recipe written heavier than the real portion. Don’t celebrate it; correct it.

Not every line will match

Don’t chase zero on every item. Weighing, trimming and rounding create some natural noise. Two things matter more:

  • The trend. Chicken short by 1 kg, then 2 kg, then 3.5 kg is a change worth chasing.
  • Value. A small gap in beef matters more than a big one in salt.

A counting sheet to copy

If you count on paper or in a spreadsheet, these columns are enough:

LocationItemUnitCounted quantityNotes
Freezer 1Chicken breastkg
Shelf 2Frying oill

There is no column for the expected quantity, on purpose: it belongs on the comparison sheet after counting, not on the sheet you count with.

Stocktakes on GoSufra

The hard part of a spreadsheet stocktake is that someone has to work out the expected figure by hand. In GoSufra inventory management that figure keeps itself up to date:

  • Stock is held per branch, and the quantity on screen is the expected stock.
  • Every sale on the till deducts its ingredients from the recipe, including sizes and modifiers. Nobody types up usage at the end of the day — see recipes and automatic stock deduction.
  • Waste is logged with a reason, so whatever variance remains after a count is the part that genuinely needs explaining.
  • When you count, you enter the quantity you found; the system works out the difference and records it in the stock movement history, so any past variance can be traced back to the day it appeared.
  • Low-stock alerts mean the count is for checking, not for discovering that the oil ran out.

To turn what you learn into a true plate cost, see recipe costing.

Mistakes that spoil a stocktake

  • Counting during service, while orders are still pulling stock.
  • Changing the unit from one count to the next.
  • One person counting and checking their own work.
  • Leaving the numbers until the comparison is impossible.
  • Ignoring positive variances as “free stock”.
  • Accusing someone on the strength of one count. A single count tells you where to look, not who to blame.

The short version

Every week, count your ten to fifteen most expensive items, at the same time, with two people, without anyone seeing the expected figure. Rank the gaps by money and start at the top. It is about an hour a week, and it usually returns more than any promotion you could run.

Run all of this from one system

POS, kitchen, inventory, recipe costing, staff and accounting — connected, and free to start.

Create your free account

← All articles

Contact us