Restaurant Stock Count: Do It Right Without Shutting Down

Your books say you have 42 kilos of rice. You walk to the shelf and there's less. You recount, same answer. Then month-end comes and you keep pushing the stock count back, because for you it means closing for a day, a pile of paper, and numbers you don't believe anyway. This guide walks you through counting your restaurant's stock end to end: why the numbers drift, what to count monthly versus weekly, how to work out a variance percentage with an example you can redo, and how to finish in about two hours without locking the door.

Why the paper never matches the shelf

Variance isn't one thing, it's four things stacking up quietly over a month, and each one has a different fix. Don't treat them all the same and don't assume the worst on day one.

Here are the four causes, most common first:

• Recipes that aren't locked down: the recipe says 180 grams of chicken and the line puts out 220. The system deducts 180, the shelf loses 220, and the gap compounds with every plate. Start with recipe costing and nail the grams.
• Short or unrecorded deliveries: the supplier wrote 20 kilos and dropped 18, and someone signed without weighing it. The system booked 20, the shelf got 18, and you started the week two kilos down without knowing it.
• Undocumented waste: a plate burned, a tub spilled, something expired and went in the bin. Waste that never gets logged comes back to you as a shortage at count time.
• Theft: product walking out the back door, or a sale that was never rung up. It happens and we won't pretend otherwise, but it's the last explanation, not the first.

That order is deliberate. Three of the four are operational and get fixed with training and paperwork; only one is stealing. Start at the top, not the bottom, or you'll lose your team's trust and the gap won't have moved.

Full count or cycle count? Both.

A full count means counting everything in the building, usually at month-end, and it's what closes your books and gives you cost of goods sold. The catch is that it's heavy going: it takes people and hours, and you can't run it every week without hurting service.

Cycle counting is the fix: count a small slice every week, only the expensive and fast-moving items, in about twenty minutes before you open. That way a problem is a week old when you catch it instead of a month, and you can still remember which shift it happened on.

So what goes on the weekly list? Use ABC, a standard inventory rule: sort your items by value, not by count. Typically around 20% of your items carry roughly 80% of your inventory value, and that's your A group: meat, chicken, cheese, coffee, canned drinks. The C group (salt, napkins, forks) is fine on the monthly.

Before the first count: four things to set up

Most counts fail before they start, not during. Lock these four down once and every count after it is a copy of the last one:

• One unit per item: rice in kilos, eggs by the piece, sauce by the tub. The moment the unit changes between counts the variance is fake and you're chasing a ghost.
• Recipes mapped to items: every menu item needs a recipe that deducts its ingredients automatically at the point of sale. Without it there's no system balance to compare against, you're writing down a number with nothing behind it.
• A fixed quiet slot: after close or before open, the same day every month. Counting mid-service gives you a wrong number, because stock is moving under your hands while you count it.
• Count sheets in shelf order: print the sheet in storage order (fridge, freezer, dry, bar), not menu order. Whoever counts walks it row by row without doubling back, and that alone cuts the time in half.

The monthly count routine: seven steps

Let's walk it with an example we follow to the end: a small restaurant counting at month-end, with a system balance of 42 kilos of rice. Follow the seven steps in order and you're done in about two hours, trusting your numbers.

1. 1
   Pick the slot and freeze receiving

   Choose a window with no movement: after close or before open. An hour before the first count, stop all deliveries and all issues out of the store. Anything moving in or out while you count means your number is wrong from the start.

2. 2
   Print the sheets in storage order

   One sheet per area: fridge, freezer, dry, bar. The unit is printed next to each item, and the system balance is hidden from whoever counts. Anyone who sees the expected number will count until they reach it.

3. 3
   Split the team, two on the expensive shelf

   Meat, chicken, cheese and coffee get counted by two people: one calls it out, the other writes. This isn't about suspecting anyone, it's that most of your inventory value sits on those shelves, and one wrong figure there costs you more than the rest of the sheet put together.

4. 4
   Count every item in a single unit

   If the item is tracked in kilos, count it in kilos even when it's stacked in cases, and weigh anything opened instead of eyeballing it. A half-empty tub gets written down by weight, not as the word half.

5. 5
   Enter the count as the new balance, not an addition

   The number you counted is the balance, not something added to the old one. In our example: the system says 42 kilos of rice, the count is 39.5, so the balance becomes 39.5 and the 2.5 is recorded as variance with its own name and time.

6. 6
   Pull the variance report and sort it

   Once everything is entered, pull the difference and the percentage for each item and sort highest first. Don't read the sheet line by line, read the top five, that's where the story is.

7. 7
   Investigate and write the decision down

   Every item over your internal threshold gets a written answer: what caused it and what you're doing about it (fix a recipe, retrain receiving, start a waste log, watch a shift). Without a written decision the same numbers come back next month.

Put the result on a single variance sheet, and this is all you need:

The percentage is the difference over the system balance. Rice: 2.5 divided by 42 is 0.059, so about 6%. If your internal threshold is 3%, rice, chicken and drinks are all over it while the mozzarella is inside it and not worth your time. The drinks are the ugliest line: 10% and twenty cans gone, and a number like that doesn't come from a recipe or from kitchen waste.

Turn it into money so it lands, using made-up purchase prices: rice, 2.5 times 8 is 20. Chicken, 1.2 times 25 is 30. Cheese, 0.1 times 30 is 3. Drinks, 20 times 3 is 60. That's 113 from one count, and across 12 months it's 1,356 leaking out of your profit without ever touching an invoice. Put your own prices in and run the same math.

The investigation: four questions before you accuse anyone

Variance came in over your threshold? Don't line the team up for questioning. Work these four in order, and most cases are solved by the first or the second:

• Is the recipe deducting correctly? Take one item, weigh what actually goes on the plate, and compare it to the recipe. A 20 gram gap on a dish that sells 300 times a month is 6 kilos disappearing with nothing suspicious behind it.
• Was the delivery booked in full? Go back through the month's invoices and compare supplier quantities against what entered the system. An invoice that never got entered, or got entered at the paper quantity instead of the weighed one, leaves you short by exactly that much.
• Is there waste nobody logged? Ask about burns, spills and expiry. And if you don't have a waste log at all, that's your answer: the waste happened, nobody wrote it down, so it showed up in the count.
• Is there a repeating pattern? Compare this month's variance to the two before it: is the same item always short, and is it worse on a particular shift? Random variance is operational; variance with a pattern is something else.

The difference between a system that fixes your stock and one that corrupts it

Here's the line between a system you trust and one that makes things worse: a count has to set the balance, not add to it. When you enter 39.5 kilos of rice, the balance afterwards is 39.5, and the system books a 2.5 adjustment and keeps it as a variance record. Weak systems treat the counted figure as a fresh delivery, add it on top of the 42, and hand you 81.5 kilos of rice that never existed. The worst part is that it compounds: every count corrupts your stock a little more, until you stop counting because the numbers have become a joke. In Loqma, counts, movements and the variance report live in one place, and the count sets the balance while keeping the difference as a record you can go back to.

Read the count next to your register numbers

The count tells you stock is missing; it doesn't tell you where it went. Your register numbers are the other half of the story: heavy refunds on specific items, voids after the food was already made, and no-sale drawer opens landing on the same shift. If canned-drink variance is high and that same shift is full of voids, you have two threads pulling toward the same place. The X and Z report guide walks you through reading those signals.

What you walk away with

A good count isn't one with zero variance, that doesn't happen in a working restaurant. A good count gives you three things: a new per-item balance you actually trust, a short list of items over your internal threshold, and a written decision for each. Do that three counts in a row and the variance drops on its own, because the team learns the numbers get read.