Kitchen inventory: how we found UAH 47K of monthly write-off losses and recovered half of it in 90 days
To stop losing money on write-offs, run your restaurant inventory count using a three-part method: blind counting, reconciliation against theoretical food cost, and root-cause analysis. Count class A items weekly and everything monthly. Compare actual and theoretical Food Cost, split write-offs by reason code, and verify the result with a recount by a second person.
Where the money goes: loss structure and the food cost gap formula
Kitchen inventory is not a stock count done for a report. It is a way to see the difference between how much raw material should have gone into the dishes you sold and how much actually disappeared. That difference is your hidden loss.
This is the base formula for analyzing losses:
Actual Food Cost, % = Actual usage ÷ Kitchen revenue × 100
Theoretical Food Cost, % = Σ (portions sold × recipe card cost) ÷ Kitchen revenue × 100
Gap, pp = Actual Food Cost − Theoretical Food Cost
These are the benchmarks we work with: a gap up to 1.5 pp is normal for a venue with well-run controls; 1.5–3 pp calls for an investigation by category; above 3 pp the system is leaking, and rarely for a single reason. They are working benchmarks, not an industry standard, so compare first of all against your own trend.
Four sources of loss
- Recorded write-offs. Spoilage, production waste, supplier defects, guest returns. They are visible and therefore manageable, provided they are logged with a reason.
- Unrecorded losses. Unlogged staff meals, tastings, portion overruns, kitchen-to-bar transfers without a document.
- Receiving errors. Short weight, wrong grade, substitution. This is where a supplier quietly takes 2–4% of the weight from every delivery.
- Data errors. Outdated recipe cards, wrong unit conversion factors, unaccounted cooking losses.
Here is what we saw in practice. In an 80-seat restaurant with kitchen revenue of UAH 1.4M per month, actual Food Cost was 34.4% against a theoretical 31.0%. A 3.4 pp gap is UAH 47,600 every month, almost UAH 571K a year. The breakdown of causes turned out to be unexpected.
| Cause of the gap | Share, pp | Loss, UAH/month | How it was found | What changed |
|---|---|---|---|---|
| Outdated recipe cards after an equipment change | 0.9 | 12,600 | Checked meat yield after cooking | Recalculated yields, updated 38 cards |
| Receiving: short weight, wrong grade | 0.7 | 9,800 | Control weighing of 10 deliveries | Receiving log with net weight, same-day claims |
| Portion overruns on the line | 0.6 | 8,400 | Portion control over 3 shifts | Portion spoons, scales at the pass |
| Staff meals and tastings | 0.6 | 8,400 | Reconciled with shift schedule | Separate write-off code, per-shift limit |
| Unrecorded spoilage | 0.4 | 5,600 | Inspected prep items and opened packs | Date labels, FEFO rotation |
| Unexplained, including theft | 0.2 | 2,800 | Residual gap after all checks | Control recounts |
| Total | 3.4 | 47,600 |
The table is an illustrative calculation based on a generalized case; substitute your own data from the variance report. Note the main point: theft and the unexplained account for only 6% of the gap. The rest are systemic causes, fixed by procedure rather than by firing people.
Owners who start by looking for someone to blame spend months and still miss most of the money. Start with the process: receiving, recipe cards, portions, records.
Preparation: item master, recipe cards and ABC analysis
The accuracy of a count is largely decided by what you did before the first item was counted. If the item list is cluttered with duplicates and units of measure contradict each other, you will count perfectly and still get a meaningless result.
Clean the item list and fix the units of measure
In a typical 420-item restaurant database we find 40–60 duplicates ("Sour cream 20%" and "Sour cream 20 % Halychyna") and 15–20 items with no pack-size factor. Each of them is a source of a phantom shortage or surplus.
Set one rule: dry goods in kg, liquids in liters, countable items in pieces, with pack weight or volume recorded in the item card. Then the count does not depend on who does it.
Check recipe cards after any change of equipment or supplier
The non-obvious point: the food cost gap is most often created not by staff but by a recipe card that has fallen behind the kitchen. In one restaurant, after switching from a standard oven to a combi oven, the yield of roasted pork neck rose from 64% to 72%, while the recipe card still said 64%.
| Parameter | Oven, old recipe card | Combi oven, updated recipe card |
|---|---|---|
| Cooked yield | 64% | 72% |
| Raw meat per 160 g portion | 250 g | 222 g |
| Portion cost (raw at UAH 220/kg) | UAH 55.0 | UAH 48.9 |
| Raw meat for 1,500 portions per month | 375 kg | 333 kg |
| Monthly difference | — | 42 kg, about UAH 9,170 |
The system assumed the kitchen used 250 g of raw meat per portion when the real figure was 222 g. This produced a phantom surplus of raw material, which covered up a real shortage in other items, so the overall gap looked acceptable. That is how an outdated recipe card masks a problem rather than creating one.
The rule we leave with every owner: after replacing an oven, changing a cooking mode, a supplier or a cut, run a test batch with control weighing and recalculate the yield. One hour of the chef's time saves thousands of hryvnias every month.
Split items by ABC and XYZ
Class A items make up 70–80% of stock value (meat, fish, cheese, premium spirits), usually 12–18% of the item list. Class B is the next 15–20% of value, and class C is the rest: spices, small dry goods, supplies.
XYZ adds demand stability: X is steady, Z is sporadic. Class A items with Z demand, such as seafood for a seasonal menu, carry the highest spoilage risk. We count those more often than anything else.
Step-by-step restaurant inventory method: 9 steps
Below is the sequence we use. For a venue with 80 seats and 420 items, the full cycle takes 4–5 hours with a team of three.
Step 1. Fix the moment of the count
Count while goods are not moving: after the shift closes and the day is closed in the POS, before the first delivery. If movement cannot be stopped, keep a separate sheet of all receipts and transfers during the count. A single missed delivery on inventory day creates a gap that people then chase for a week.
Step 2. Appoint a two-person team, one of whom is not from the kitchen
One person counts, the other records. Include someone who is not responsible for the storeroom: the floor manager or the accountant. The head chef belongs on the team but not as the only counter: people check their own storeroom with bias, even with good intentions.
Step 3. Count blind
Do not show the counter the book stock. When a person sees "expected 12 kg," they fit the fact to the expectation, and the inventory turns into a confirmation of the error. The form or app should contain only a field for the actual quantity.
Step 4. Weigh opened packs and prep items
The non-obvious point: semi-finished items — sauces, stocks, marinated raw material, portioned prep — make up 8–15% of kitchen stock value, and they almost never make it onto the form. Add them to the item list based on the prep recipe card and count them in kilograms.
Weigh opened packs together with their containers and subtract the container weight from a tare table posted in the storeroom. At the bar, the unit of measure is bottle weight or liquid level on a scale in 10% steps: that way a bartender counts in a minute what would otherwise take half an hour.
Step 5. Build the variance report
The report should show book quantity, actual, difference in units and hryvnias, and deviation in percent. Investigate class A items with a deviation above 2%, class B above 5%, and any item with a difference above UAH 1,000. Record the rest, but do not spend the team's time on them.
Step 6. Find the cause using the "five control points" scheme
| Control point | Diagnostic question | Typical signal |
|---|---|---|
| Receiving | Does net weight on the delivery note match the actual weight? Is it the right grade? | Constant shortage from a single supplier |
| Storage | Is the opening date labeled? Does FEFO rotation work? | Expired items written off at month end |
| Prep | Does yield match the recipe card? How much trim goes to waste? | Raw material shortage with normal sales |
| Service | Is portion weight respected? Are all dishes rung up? | Overuse in items with a high portion price |
| Records | Are recipe cards, factors and units of measure current? | Surplus in some items and shortage in others |
Go through the points in order and stop at the first one where the hypothesis is confirmed by measurement rather than assumption. For each cause, assign one action and one owner.
Step 7. Record write-offs by reason code
A shortage that has been explained becomes a write-off with its own code. A shortage that has not been explained stays a shortage and goes into the P&L as a separate line. More on codes in the section below.
Step 8. Run a control recheck with a different person
After 24–48 hours, someone who did not take part in the main count recounts 10–15 class A items and 5 random ones. The acceptable match with the main count is at least 97%. If the match is lower, recount the whole zone. The control recheck is what turns inventory from a ritual into a tool.
Step 9. Lock the result into the P&L and into actions
Put separate lines in the P&L: "Write-offs by code" and "Shortage from inventory results." Every variance above the threshold gets an owner, an action and a check date. For analysis, build a loss heat map: categories in rows, weeks in columns, deviations in percent.
Here is what we saw in practice: when the same category stays hot for three weeks in a row, the cause is systemic — a recipe card, receiving or a portion. A one-off spike almost always has an event behind it: a banquet, a delivery failure, a power outage.
Frequency: what detection delay costs
A full monthly inventory is the minimum, not the solution. A leak that starts the day after a count will be visible only 28 days later. That is why the size of losses depends on how often you check more than on the accuracy of a single count.
To see this, take a leak of UAH 1,500 per day: short-weight deliveries or oversized portions. The average detection delay equals half the interval between counts.
| Counting frequency | Average detection delay | Loss before detection |
|---|---|---|
| Monthly, full | 15 days | UAH 22,500 |
| Every 2 weeks | 7 days | UAH 10,500 |
| Weekly, A and B items | 3.5 days | UAH 5,250 |
| Daily, high-value A items and bar | 0.5 day | UAH 750 |
The practical routine we recommend: daily, the 8–12 most expensive class A items and spirits, about 10 minutes; weekly, A and B items by zone, up to 90 minutes; monthly, the full inventory; quarterly, a control check by an outside person.
Run an extra inventory when the head chef or the person accountable for stock changes, after a large event, and after a long power outage. Ten minutes a day on ten items replaces three days of investigation at month end.
TCO of counting methods: paper, software or scales
Most owners choose a method by the price of the subscription. We calculate the full TCO of inventory: team labor, software, equipment and the cost of error. The figures below are a model for a venue with 80 seats and 420 items.
Inventory cycle productivity — items per person per hour, including weighing prep items and opened packs — differs sharply: 18–22 on paper, 30–40 in a mobile app with a scanner, 45–55 in an app with integrated scales.
| Method | Labor for a full inventory | Monthly cost | Item-level error | When the result is ready |
|---|---|---|---|---|
| Paper forms and Excel | 27 person-hours | about UAH 4.9K | 3–5% | In 2–4 days |
| Inventory software with mobile counting | 12 person-hours | UAH 3.7–5.2K | 1–2% | Same day |
| Software, integrated scales, scanner | 8.5 person-hours | UAH 3.7–5.6K | up to 1% | Same day, broken down by zone |
Model assumptions: a person-hour costs UAH 180, inventory software costs UAH 1.5–3K per month, scales and a scanner cost UAH 25–40K amortized over 36 months. Check the current pricing of your system.
The cost difference between methods is up to UAH 1K per month. The difference in error and speed is far larger: a 4% error on class A stock worth UAH 250K is UAH 10K of phantom shortage or surplus that the team then has to investigate.
Here is what we saw in practice: moving from Excel to mobile counting cut the inventory time for a team of three from 9 hours to 4. But the main effect was different — results were available the same day, and the chef could fix the cause that very week.
Write-offs: reason codes and benchmarks
A write-off without a code turns into a catch-all bin: spoiled sour cream and a missing steak both end up in it. A reason code shows which write-offs are justified by the business (tasting, marketing) and which are simply losses.
| Code | Example | Who confirms | Benchmark |
|---|---|---|---|
| 1. Spoilage, expiry | Chilled fish, dairy | Sous-chef and manager | Up to 1.0% of purchase cost |
| 2. Production error | Overcooked, cooking mistake | Head chef | Up to 0.5% of purchase cost |
| 3. Supplier defect | Short weight, wrong grade | Purchasing manager | Claim on the day of delivery |
| 4. Guest return | Complaint, wrong order | Duty manager | Logged together with the receipt |
| 5. Staff meals | Per shift schedule | General manager | Per-person allowance |
| 6. Tasting, marketing | Guest compliment, trial portions | General manager | Within budget |
| 7. Emergency event | Power outage, cold room failure | Committee with a report | Temperature log and photos |
The benchmarks are working figures: set your own after 2–3 months of measurement. In a venue with well-run controls, total recorded write-offs are usually 1.5–3% of purchase cost; above 4% signals problems in purchase planning or rotation.
Every write-off has four elements: a photo, a weight, a code and the signatures of two people. The manager approves it on the day of the event. A write-off backdated at month end is a sign that the system is not working and that the numbers are being fitted to the stock.
A separate note on rotation. FEFO ("first expired, first out") matters more than FIFO where one category contains different shelf lives: cheeses, chilled fish, herbs. A date label for the opening day on every pack is the cheapest fix in this section.
Local context: power outages, prices and paperwork
A power outage is a separate cause of loss for Ukrainian venues. After a long outage, the person responsible for HACCP decides on write-offs from the temperature log. Without the log you cannot prove the write-off was justified, and both the committee and inspectors judge the document.
A digital temperature logger with a phone alert pays for itself in one blackout: 40 kg of chilled meat at UAH 220/kg is UAH 8,800. For venues that keep smaller stocks because of unstable logistics, a daily count of A items is more realistic than a full inventory.
The second local factor is unstable prices. If you value stock at the last purchase price, Food Cost swings by 1–2 pp with no actual loss. We move to weighted-average valuation and record the revaluation date. For venues that use imported ingredients, add the effect of the exchange rate on purchase prices.
Legally, the procedure and timing of inventory are governed by the Regulation on Inventory of Assets and Liabilities (Ministry of Finance order No. 879 of 02.09.2014). Inventory is carried out, among other cases, before annual financial statements are prepared, when the accountable person changes, when theft or damage to property is found, and after emergencies. Results are recorded on standard forms, including the inventory list (INV-3) and the comparison statement (INV-19).
Control inventory checklist
Below is the core of the checklist we use to verify the result after the main count. The full version, with signature fields and deviation thresholds, can be downloaded as a separate file.
Before the count
- The day is closed in the POS and the count start time is recorded
- All deliveries and transfers are posted; new ones go on a separate sheet
- A two-person team, one of whom is not from the kitchen
- Item list free of duplicates, pack-size factors filled in
- Container tare table posted in the storeroom
During the count
- Blind count, without book stock
- Prep items, sauces and stocks counted separately
- Opened packs weighed, tare subtracted
- Bar spirits counted by scale or bottle weight
- Zones marked with "counted" stickers
After the count
- Variance report built on the day of the count
- A cause found for A items with a deviation above 2% and B items above 5%
- Control recheck done by another person after 24–48 hours: 10–15 A items and 5 random ones
- Match with the main count is at least 97%, otherwise the zone is recounted
- Write-offs recorded by code, with photos and two signatures
- Separate P&L lines for write-offs and shortage; owners and check dates assigned
Download the full control inventory checklist with deviation thresholds and a variance report form.
Get the checklist30-60-90 day plan for owners and investors
- Days 1–30: baseline. Clean the item list and recipe cards, introduce blind counting and count 10 class A items every day. The goal is to measure the current gap, not to reduce it.
- Days 31–60: loss channels. Introduce write-off codes, a receiving log with net weight, and a weekly count of A and B items. The goal is to cut the gap by about 1 pp.
- Days 61–90: stability. Add control rechecks by another person, a loss heat map and an owner for each category. The goal is a gap of 1.5–2 pp and stability from week to week.
In the case example, cutting the gap from 3.4 to 1.6 pp at kitchen revenue of UAH 1.4M per month returns UAH 25,200 every month, or UAH 302,400 a year. The cost of the method in the TCO table is no more than a fifth of the recovered margin.
FAQ: short answers to related questions
Author: Ruslan, Independent HoReCa Analyst, Expert Technologist at eeat.com.ua
