How to Do Restaurant Inventory (7 Steps)

Most kitchens that distrust their food cost number do not have a food cost problem. They have a counting problem, and from the office the two look identical. The percentage arrives late, moves four points for no reason anybody can explain, and eventually stops being discussed because nobody believes it. Underneath that is almost always a physical count that was taken in a different order, in different units, at a different time of night, by a different person, than the one it is being compared against. Two numbers produced that way are not comparable, and no amount of analysis downstream can rescue them.

This walkthrough is about the count itself rather than the ratio it feeds. Seven steps: build the count sheet in the order you physically walk your storage, standardize a count unit for every item, freeze the period, count in pairs, price the sheet one way and keep it that way, convert the counts into cost of goods sold, then compare actual usage against theoretical usage to find out what the gap is made of. For the ratio the count produces, our walkthrough on calculating food cost percentage covers the arithmetic in detail, and our rundown on cutting restaurant food waste covers one of the causes the variance will point at. Run your own opening, purchases, closing and sales through the equipment ROI calculator as you read.

Before you start

Four things need to exist before the first shelf gets counted, and none of them costs money. Assembling them takes an afternoon and removes most of the reasons a count comes out wrong.

• A written definition of what is in the count. Decide once whether beverage, alcohol, paper and packaging are inside your food inventory or tracked separately, write it on the top of the sheet, and apply it identically every period. A definition that drifts makes this period incomparable with the last one.
• A map of your storage areas in walking order. Walk-in, freezer, dry store, line stations, overflow shelving, wherever else food actually sits. Number them. That numbering becomes the section order of the count sheet.
• Current purchase prices. The most recent invoice price for every item you carry, in the unit you buy it in, so pricing the sheet is a lookup rather than an argument.
• Food sales for the exact same window, pulled from your point-of-sale and separated from beverage and other revenue, so the ratio at the end has a matching denominator.

Time and difficulty are modest and worth stating plainly. The first full count in a kitchen that has never had a stable sheet commonly takes three or four hours because you are inventing the sheet while you use it. Once the sheet settles, two people can walk an illustrative small full-service kitchen in about two hours, with pricing and data entry adding perhaps forty-five minutes. Difficulty is low, the arithmetic is fourth-grade, and the only genuinely hard part is doing it the same way twice. Nothing here is accounting or tax advice, and every dollar figure is an illustrative planning shape rather than a quoted price.

What a physical count is actually measuring

It helps to be precise about what the count is for, because the answer is not “knowing what you have”. You mostly know what you have; the walk-in is right there. The count exists to establish two boundary values so that the flow between them can be calculated. Opening inventory is the value of food on the shelf at the instant the period began. Closing inventory is the value at the instant it ended. Everything that happened in between, every delivery, every plate, every dropped tray and every staff meal, is captured by the difference plus the invoices.

That is why the count is a boundary measurement and not a stocktake in the retail sense. Retail counts units to reconcile against a unit-level system. A kitchen count values a shelf so that consumption can be inferred. It follows that the accuracy that matters is not absolute accuracy but consistent accuracy. A count that is systematically five percent light in the dry store, every single period, in the same way, produces a nearly correct cost of goods sold, because the error appears in both the opening and the closing number and largely cancels. A count that is perfect one period and casual the next produces two wrong periods.

There is a second thing the count measures, which most kitchens ignore for years and then discover all at once. Because the sheet carries a dollar value against every line, it is also a map of where your working capital is parked. Money sitting as frozen product on a shelf is money not in the bank, and an inventory that quietly grows by a few thousand dollars over a season is a cash flow event that never appears on the profit and loss. Our rundown on restaurant profit margin covers why that distinction matters more in food service than in most businesses.

Step 1: Build the count sheet in storage order

Open a blank sheet and resist every instinct to organize it the way a spreadsheet wants to be organized. Do not sort alphabetically. Do not group by supplier. Do not group by category. Organize the sheet in the exact physical order a person walks your storage: walk-in first, shelf by shelf from the door and going clockwise, then the freezer the same way, then dry storage aisle by aisle, then the line stations and reach-ins, then overflow. Every line on the sheet should appear in the order the counter’s eyes will meet it.

The reason is boring and decisive. A counter working from an alphabetical sheet has to walk the room repeatedly, or hold a partial list in memory, and both produce misses. A counter working from a sheet in storage order walks once, in a straight line, and any item that is on the shelf but not on the sheet becomes immediately obvious because it interrupts the sequence. Missed items are the single most common cause of an inventory value that jumps around, and they are almost entirely a sheet design problem rather than a diligence problem.

Print section headers for each storage area with space at the bottom for write-ins, because you will find items that are not on the sheet and you want them captured rather than skipped. Leave a column for count, a column for unit, a column for unit cost and a column for extended value. Number the sections so a half-finished count can be handed over without ambiguity. Then, and this is the part that gets skipped, physically reorganize the storage once so it matches the sheet rather than the other way around. The sheet is the standard; the shelves should conform to it.

The watch-out at this step is building the sheet from your supplier’s order guide. An order guide lists what you buy in the order the supplier’s catalogue happens to run, which correlates with nothing in your building. It is a useful source of item names and pack sizes and a terrible source of sequence.

Step 2: Standardize the count unit for every item

Now decide, item by item, the single unit you will count in, and write the conversion from the purchase unit next to it. This is the step that separates a count somebody can repeat from a count only its author understands. The purchase unit is what the invoice says: a case, a bag, a jug, a tub. The count unit is what a person can plausibly stand in a walk-in and tally: a can, a pound, a jug, a quart. Where those differ, the conversion has to live on the sheet in writing.

Take an illustrative set of conversions from the kitchen used throughout this walkthrough:

• Crushed tomatoes, bought as a case of six #10 cans at $38.40. Count unit is the can, so the unit cost is $38.40 divided by 6, or $6.40 a can.
• All-purpose flour, bought as a 50 lb bag at $27.50. Count unit is the pound, so $0.55 a pound, and partial bags get weighed rather than eyeballed.
• Olive oil, bought as a case of four 3 L jugs at $86.00. Count unit is the jug at $21.50, with part-jugs recorded to the nearest quarter.
• Chicken thighs, bought as a 40 lb case at $103.60. Count unit is the pound at $2.59, counted on a scale rather than by case.
• Heavy cream, bought as a case of twelve quarts at $47.40. Count unit is the quart at $3.95.
• Fryer oil, bought as a 35 lb jug at $32.90. Count unit is the sealed jug, and the oil already in the fryer is excluded by written rule rather than by whoever is counting that night.

Two rules make these hold up. First, one unit per item, forever, until you deliberately change it and note the change. An item counted in cases one period and pounds the next can move a single line by a factor of forty, and that error is large enough to swamp everything else on the sheet. Second, the conversion is arithmetic you do once and print, not arithmetic the counter does at 1am in a cold room. Every conversion performed under fatigue is a coin flip.

Step 3: Freeze the period and set the count window

A count is a photograph of a moving object, and if the object moves during the exposure the photograph is useless. So fix the moment. The standard that works in nearly every kitchen is this: count after close on the last night of the period, after the line has been broken down and everything has been put away, and before the first delivery of the new period arrives. That window is usually a few hours long and it is the only time the building is genuinely still.

Write the rule down and apply it identically every period, including the same start time, because a count taken at 11pm and compared to one taken at 6am after a morning prep session is comparing two different kitchens. If your operation runs overnight prep, the boundary moves to whatever your genuinely quietest hour is, but it moves permanently rather than per period.

Then handle the exceptions in advance, because they will happen. If a delivery arrives during the count, do not put it away and do not count it. Quarantine it in a marked area, leave it sealed, and record the invoice in the new period’s purchases. If a delivery arrives before the count but after close, you have a choice, and either answer is fine as long as it is the same answer every time: count it and put the invoice in the closing period, or quarantine it and put the invoice in the next one. What is not fine is counting the stock in one period and the invoice in the other, which double-counts the food and is the single most common way a food cost percentage comes out impossible.

The watch-out here is the transfer nobody logs. Food moving between locations, to a catering van, to a second site, or borrowed from a neighbor at 7pm on a Saturday, has to be recorded as a transfer out or in, or it lands in your usage as consumption that never happened. A transfer log is one line on a clipboard and it prevents a category of variance you would otherwise spend a month chasing. If you also run offsite events, our walkthrough on starting a catering business covers why that flow needs its own paperwork.

Step 4: Count in pairs and record as you go

Send two people, and give them different jobs. One counts and calls the number out loud. The other holds the sheet, writes it down, and reads back the next item on the list. That division is not ceremony. A single person counting and writing does both tasks badly, loses their place, and, worse, tends to write what they expect rather than what they see. Two people with separate jobs produce a measurably steadier sheet, and the count goes faster despite using twice the labor, because nobody is stopping to find their place.

Count what is physically there, not what the shelf tag, the par sheet or the system says should be there. This sounds obvious and it is violated constantly, especially late in a long count when the remaining lines are small. If the sheet says nine cans and the shelf holds seven, the answer is seven, and the discrepancy is information rather than an error to be smoothed over.

For high-value items, count blind. Blind counting means the recorder does not tell the counter what last period’s number was and does not show them the system’s expected quantity. Anchoring is powerful and unconscious: a counter who knows the expected figure is twelve will find twelve more often than the shelf deserves. On the handful of lines carrying real money, proteins, seafood, cheese, oil, spirits if they are in your definition, blind counting is worth the small extra friction.

Record as you go rather than reconstructing afterward, and record in ink. A count that gets typed into a spreadsheet from memory forty minutes later is not a count. Where an item genuinely cannot be resolved, a sealed case you cannot open or a container whose contents are unclear, write a question mark and a note rather than a guess, and resolve it before the sheet is priced.

Step 5: Price the sheet with one costing method

Now put a dollar value against every counted line, and make one decision that you will not revisit: last cost or weighted average. Last cost means every unit on the shelf is valued at the most recent invoice price for that item. Weighted average means you blend the value of what you opened with the cost of what you bought during the period, and price the closing count at the blended figure. Both are legitimate. Neither is more honest than the other. The only wrong answer is switching between them.

Watch what the choice actually does on one line. Take the crushed tomatoes: you opened with 8 cans valued at $6.10, bought 24 cans at $6.40, and counted 9 cans at the close. Under last cost, those nine cans are worth 9 times $6.40, or $57.60. Under weighted average, the goods available were 8 times $6.10 plus 24 times $6.40, which is $48.80 plus $153.60, or $202.40 across 32 cans, giving $6.325 a can, so the nine cans are worth $56.93. The difference is 67 cents.

Sixty-seven cents on one line is nothing. Across a full storeroom in a period where supplier prices moved, the same effect compounds into a few hundred dollars, and a few hundred dollars of inventory value is a few tenths of a point of food cost. That is small enough to ignore and large enough to be mistaken for an operating result if the method changes underneath you. Pick one, write it at the top of the sheet next to your definition of food, and leave it alone. If you do change, change at the start of a fiscal year and note it, so the step change is labeled rather than diagnosed.

The watch-out is stale prices. A sheet carrying last quarter’s costs values a shelf at a price nobody can buy at, and it produces a food cost that lags reality by a quarter. Refresh the price column from live invoices every period as part of pricing the sheet, and flag any item whose cost moved by more than a few percent, because a supplier price move is a cause you can act on before it reaches the plate. Our walkthrough on pricing a restaurant menu covers what to do once you have found one.

Step 6: Turn the counts into cost of goods sold and food cost

With two counts and a stack of invoices, the arithmetic is short. Cost of goods sold equals opening inventory plus purchases minus closing inventory. Opening is what you had, purchases are what you added, and subtracting what is still on the shelf leaves what left the building as meals, waste or loss. Then food cost percentage equals cost of goods sold divided by food sales for the identical window, times one hundred.

Run the illustrative period. Opening inventory, which is last period’s closing count, is $14,200. Purchases across every food invoice in the window total $38,600. The closing count prices out at $13,400. Cost of goods sold is $14,200 plus $38,600 minus $13,400, which is $39,400. Against $124,000 of food sales, that is 31.8%. Note that the kitchen consumed $39,400 of food while buying only $38,600 of it, because it drew $800 off the shelf, and capturing exactly that difference is why the two counts were worth two hours.

The single most common shortcut, dividing purchases by sales, would have reported $38,600 over $124,000, or 31.1%, which is about six tenths of a point flattering and would be wildly wrong in any period where you stocked up for a holiday or ran the freezer down. The counts exist to strip stocking swings out of the number, and skipping them turns a management metric into a description of your ordering rhythm.

On targets, be careful about borrowed benchmarks. A commonly used planning band for full-service concepts runs somewhere around 28% to 35%, but treat that as a planning band that varies by concept rather than a verified standard: a high-volume format with light labor can carry a higher food cost and make it back on throughput, while a scratch kitchen with heavy prep labor often needs a lower one. Set the target against your own labor and occupancy, which our explainer on restaurant labor cost percentage sizes, rather than against a number from somebody else’s business.

Step 7: Compare actual usage against theoretical usage

The food cost percentage tells you the result. It does not tell you the cause, and a kitchen that stops at step six spends the next month arguing about it. The step that produces a cause is the comparison between actual usage, which your counts and invoices just gave you, and theoretical usage, which is what your recipes say you should have used given what you actually sold.

Build theoretical usage by multiplying each menu item’s sales count for the period by its costed recipe, then totaling. In the illustrative period, that totals $37,200, or exactly 30.0% of the $124,000 of food sales. Actual cost of goods sold was $39,400, or 31.8%. The variance is $2,200, about 1.8 points of sales, and if the gap holds across thirteen periods it is roughly $28,600 a year. That is a number worth an afternoon.

Then take the variance apart at item level, because the total is only a headline. Follow one line all the way through. The recipe for the braised chicken plate calls for 6 oz of raw thigh, which is 0.375 lb at $2.59, or $0.97 of chicken per plate. The point-of-sale says 620 of those plates sold, so theoretical usage is 232.5 lb. Actual usage from the counts is 84 lb opening plus 260 lb purchased minus 76 lb closing, which is 268 lb. The gap is 35.5 lb, worth about $92, and 15% more chicken than the recipe called for. That is a portioning conversation with a specific station, not a general complaint about food cost.

The watch-out is treating a first variance as an accusation. The first time most kitchens run this comparison, a large share of the gap turns out to be stale recipes, uncosted modifiers and menu items whose yields were never tested, rather than anything happening on the line. Clean the recipe file first, re-run it next period, and only then start looking at people.

Where the count value sits by storage area

Before the worked example, look at where the money on the sheet actually sits, because it determines where your counting attention belongs. The bars below break the illustrative $13,400 closing count into the five storage areas it was counted in. Each bar is drawn from its share of the largest area.

Two things fall out of that shape. The walk-in and the freezer together carry about $8,100, roughly six tenths of the count, which is where the slow, careful counting belongs and where blind counts are worth the friction. The overflow shelving carries $600, and it is the area most often skipped entirely, which costs you $600 of accuracy in a number that swings on far less. Counting it takes four minutes. The habit of skipping it is worth more error than every rounding decision on the rest of the sheet combined.

What the usage variance is made of

The $2,200 variance is a total, and totals do not tell you what to do on Monday. Break it into causes. The split below is the illustrative attribution for the same period after the kitchen worked through it line by line, with the five causes summing to the full variance.

Read the shape rather than the exact shares. Portioning is usually the largest single cause and the cheapest to fix, because it responds to scales, portion tools and one focused retraining session. Unlogged waste is second and is a measurement problem before it is an operational one, which our rundown on cutting restaurant food waste works through in detail. Stale recipes are pure paperwork and are entirely your own fault. Comps and staff meals are usually a reporting fix in your point-of-sale rather than anything happening in the kitchen.

The residual matters most of all, and it should be small. A kitchen that resolves the first four causes and is left with 7% of the variance unexplained is running a tight room. A kitchen where the unexplained share is half the variance has either a counting problem or a shrinkage problem, and those two are worth separating carefully before anybody says anything out loud.

A worked example: one 28-day period end to end

Put the whole procedure through one period with every number attached. The kitchen is an illustrative small full-service restaurant running thirteen 28-day periods a year, with a walk-in, a freezer, a dry store, four line stations and a rack of overflow shelving. The sheet is printed in that order, every item has one count unit with a printed conversion, and the costing method is last cost.

The period opens with a closing count carried forward from the previous period: $14,200. Across the 28 days, food invoices total $38,600, with beverage, alcohol and paper coded to separate accounts per the written definition on the sheet. On the last night, after close and before the morning delivery, two people walk the sheet in about two hours. The closing count prices out at $13,400: $4,700 in the walk-in, $3,400 in the freezer, $3,000 in dry storage, $1,700 across the line stations and reach-ins, and $600 on the overflow rack.

Cost of goods sold is $14,200 plus $38,600 minus $13,400, or $39,400. Food sales for the identical window, pulled from the point-of-sale and stripped of beverage, are $124,000. Food cost percentage is $39,400 divided by $124,000, which is 31.8%. Average inventory across the period is the mean of the two counts, $13,800, so inventory turned $39,400 divided by $13,800, about 2.9 times in the period, which is roughly 37 times a year and about 9.8 days of food on hand.

Then the part that pays for the exercise. Theoretical usage, built from recipe costs times items sold, is $37,200, or 30.0% of sales. The variance is $2,200, about 1.8 points, and about $28,600 a year at that rate. Broken down, roughly $880 is portioning drift, $550 is waste that never made the log, $396 is recipes and yields nobody has updated since the menu changed, $220 is comps and staff meals not backed out of sales, and $154 is unexplained. Every one of those is a different Monday morning conversation, which is the whole reason the count was taken. Put your own four numbers into the equipment ROI calculator and the companion beside this walkthrough to see your version.

Reading inventory turns and days on hand

The same two counts that produce food cost also produce a second number most kitchens never look at, and it is the one that tells you whether you are storing money. Inventory turns equals cost of goods sold divided by average inventory, where average inventory is the mean of the opening and closing counts. In the worked period, that is $39,400 divided by $13,800, or about 2.9 turns in 28 days. Days on hand is the period length divided by turns, so 28 divided by 2.9, about 9.8 days.

What that means in plain terms is that this kitchen holds a little under ten days of food at any moment. Higher turns mean less cash parked on shelves, fresher product and less exposure to spoilage. Lower turns mean more cushion against a supply interruption and more buying leverage on bulk orders, paid for with cash you cannot use and product that ages. Neither direction is automatically correct, and the useful discipline is watching your own trend rather than chasing a figure from another concept.

Watch what happens when turns slip. Suppose the same kitchen lets average inventory drift from $13,800 to $17,000 over a season without changing sales. Turns fall from 2.9 to about 2.3 and days on hand rise from 9.8 to about 12.1. Nothing on the profit and loss changes at all, because the food is still an asset rather than an expense, but $3,200 of cash has moved out of the bank and onto a shelf, and more of it is now old enough to be at risk. That is a real event that only the inventory count can see.

The pairing to watch is turns against your waste log. Falling turns plus rising spoilage is the classic signature of over-ordering, and it is a par level problem rather than a kitchen problem. Rising turns plus rising stockouts is the opposite failure, and it usually means somebody trimmed pars to hit a cash target and the line is now improvising.

Building theoretical usage without a full recipe system

Step seven assumes you can produce theoretical usage, and the usual objection is that building it requires a recipe system nobody has time to create. That objection is fair for a full build and wrong as a reason to skip the step, because a partial theoretical is worth most of what a complete one is worth. The trick is to build it for the items that carry the money and ignore the rest for now.

Start with the top ten menu items by sales volume and the top ten ingredients by purchase value, which overlap heavily in most kitchens. Cost the recipes for those items only, using weighed yields rather than package weights, because a 40 lb case of chicken thighs is not 40 lb of plated chicken and the difference is exactly the kind of thing this exercise is meant to find. Multiply each item’s sales count by its costed recipe, total it, and compare it against actual usage for those same ingredients pulled from your counts and invoices.

You will not have a whole-kitchen theoretical, and you do not need one. What you will have is a variance on the lines where variance is expensive, which is where the money was going to be found anyway. In the worked example the chicken line alone showed 35.5 lb and about $92 of gap on 620 plates, and that single line was found without costing a single dessert.

Build outward from there at a rate you can sustain, adding a few recipes each period until the coverage is high enough that the residual is small. Two cautions apply. Costed recipes rot: every menu change, every substitution and every supplier switch invalidates part of the file, so schedule a recipe review whenever the menu moves. And modifiers count. A dish with a free extra sauce that nobody costed will show as portioning variance forever, and the fix is in the recipe file rather than on the line. If your point-of-sale can export item-level sales counts, our comparison of restaurant POS systems covers the reporting features that make this step a download rather than a transcription job.

How often to count: weekly high value, monthly full

Cadence is where most kitchens either over-invest or give up. The pattern that holds up is two cadences running together: a full count once per accounting period, and a short weekly count of the items that carry the money and move fastest.

The full count is the one that feeds cost of goods sold and the food cost percentage, so it has to align exactly with your accounting period and cannot be skipped or shifted for convenience. A period that runs 28 days and a count that happens whenever somebody has time produce a percentage with a different denominator every month.

The weekly count is a different instrument. Its job is early warning, not accounting. In the illustrative kitchen, the twenty highest-value lines carried about $9,000 of the $13,400 closing count, roughly two thirds of the value, and counting only those takes about twenty-five minutes rather than two hours. Run that every week against expected usage from your sales, and a portioning problem or a shrinkage problem surfaces within seven days instead of within thirty. That is the difference between one bad week and one bad month.

Two adjustments are worth making. Tighten the cadence temporarily whenever you are actively chasing something: a suspected loss, a new menu, a new prep cook, a new supplier. Loosen it for genuinely slow-moving dry goods, which can sit on the full-count cadence only. And resist the urge to count everything weekly out of enthusiasm, because a cadence that consumes more labor than the variance it finds is a hobby rather than a control.

Handling partial units, open containers and prep

Every count runs aground on the same three questions, and the answer to all of them is a written rule rather than a nightly judgment call. The first is partial units. A bag of flour with an unknown amount left, a jug of oil two thirds down, half a case of cans. Rule: anything sold by weight gets weighed, anything sold by volume gets recorded to the nearest quarter of the container, and anything sold by count gets counted. Write the rounding convention on the sheet so two people apply it the same way.

The second is open containers on the line. A quarter pan of diced onion in a reach-in is real inventory and real money, and in aggregate the line stations held $1,700 in the worked count, which is more than the overflow rack. Rule: count line pans at their fill fraction against a standard pan size, and count them every period rather than only when they look substantial.

The third and hardest is prepared food. A hotel pan of braised short rib is not a raw ingredient and not a sold plate. Two defensible rules exist. Value prep at the raw cost of its components, ignoring the labor, which is simple and slightly understates the asset. Or value it at a standard batch cost from the recipe file, which is more accurate and requires the recipe file to exist. Either is fine; the important part is that both counts use the same one, because valuing prep one way in the opening count and another way in the closing count invents a variance out of thin air.

The general principle behind all three: a rule you wrote once beats a decision you make at midnight. Every judgment call left to the counter is a source of variance between periods, and variance between periods is exactly what the whole procedure exists to eliminate.

Deciding what belongs in the count: food, beverage, paper

The scope question feels administrative and is not. Whether beverage, alcohol, paper goods and packaging sit inside your food inventory changes both your inventory value and your food cost percentage, and doing it differently in two periods makes them incomparable.

The most common arrangement in full-service restaurants is to run food, beverage and alcohol as three separate inventories with three separate cost percentages, and to treat paper and packaging as an operating supply rather than a cost of goods. The logic is that each category has a different margin structure and a different set of levers, so blending them hides the thing you would want to see. Alcohol in particular has both a very different cost profile and a very different shrinkage profile, and burying it inside food cost guarantees you will never notice either.

Delivery-heavy and takeout-led operations often reach a different answer on packaging, and reasonably so. When a sealed container, a lid and a bag cost real money on every single order, packaging behaves like an ingredient rather than an overhead, and some operators cost it into the dish. Our walkthrough on starting a ghost kitchen works through why that treatment makes sense when the container is part of the product.

Whatever you choose, three things follow. Write it at the top of the count sheet. Code your invoices to match it, because a purchases figure that includes paper against an inventory that excludes it produces a food cost that is simply wrong. And change it only at a period boundary with a note attached, so a future reader can see why the value stepped rather than trying to explain it as an operating result.

Counting the walk-in, the freezer and the dry store

Each storage area has its own failure mode, and knowing them makes a count faster as well as better. In the walk-in, the problem is depth. Product gets pushed to the back, the front is what gets seen, and a shelf that looks like four cases is often six. Count front to back physically rather than visually, and pull the front row forward as you go, which doubles as the stock rotation you were meant to be doing anyway. The walk-in carried the largest single share of value in the worked count, so this is where slowness pays.

In the freezer, the problem is legibility. Frozen product in unlabeled bags is genuinely hard to identify and unpleasant to handle, so counters guess. The fix is upstream: label and date everything on receipt, store by category, and keep a scale in reach so weighed items are weighed rather than estimated. If frozen product is a large share of your value, our rundown on commercial freezer cost covers sizing and configuration, since a freezer packed past the point where you can see into it is both a counting problem and a temperature problem.

In dry storage, the problem is complacency. Dry goods are cheap per unit, feel unimportant, and are the most likely to be skipped or rounded. In aggregate they carried $3,000 in the worked count, which is more than the entire line-station total. Count the aisles the same way you count the walk-in, and pay particular attention to bulk items in bins, which need weighing rather than eyeballing.

Across all three, temperature and organization affect the count as much as diligence does. A cooler holding steady with room for air to move is easier to count and produces less spoilage to explain later, and our rundown on commercial refrigerator cost covers what capacity actually buys you. Holding temperatures and monitoring rules are set by your local health authority, so confirm what applies to you rather than working from a figure you read somewhere.

What the count is worth beyond the food cost number

Food cost is the headline use, and it is not the only one. Once a stable, priced count exists every period, several other things become possible that were guesswork before, and they are worth knowing about because they change the cost-benefit of the two hours.

The most immediate is ordering. A count in storage order, taken the night before an order goes in, is the most accurate par-level input that exists, because it is what is actually there rather than what the system believes. Kitchens that order off a fresh count consistently carry less stock and stock out less often, which is the combination that ought to be impossible and is not.

The second is a defensible inventory value for anyone who needs one. Your accountant needs it to close a period properly, since inventory movement is the difference between purchases and cost of goods sold on the profit and loss. A lender looking at a loan application will treat a business with clean periodic inventory records differently from one whose food cost is an estimate, as our walkthrough on getting a small business loan sets out. And after a loss event, a documented, dated, priced count is far better evidence of what was on the shelf than a memory is, which our breakdown of restaurant insurance cost touches on in the context of spoilage and business interruption coverage. What any specific policy actually covers varies by policy and insurer, so read yours and ask your broker rather than assuming.

The third is equipment decisions. Consumables that live inside a machine, fryer oil most of all, only become visible as a cost line when somebody counts them. A kitchen that knows it consumes a specific number of jugs a period can price an oil filtration decision properly, and our breakdown of commercial fryer cost by type covers the equipment side of that trade. Run any such payback through the equipment ROI calculator before you commit.

When the variance points at theft rather than process

This is the part of the subject people are uncomfortable discussing, and it deserves a calm treatment rather than a dramatic one. Most variance is not theft. In the worked example, 93% of the gap was portioning, unlogged waste, stale recipes and unreported comps, all of which are process failures with process fixes. Assuming theft first is both statistically wrong and corrosive to a kitchen, and it makes the actual causes harder to find because people stop volunteering information.

That said, the count is the only instrument that can raise the question at all, and there are signatures worth recognizing. Theft-shaped variance tends to be concentrated rather than spread: it sits on specific high-value, easily portable items rather than across the sheet. It tends to be persistent across periods rather than spiky. It often survives the recipe cleanup that resolves most other causes. And it frequently correlates with specific shifts or specific access windows in a way that portioning drift does not.

If a pattern like that emerges, tighten measurement before drawing conclusions. Move the affected items to a weekly or even daily blind count. Check whether receiving is verifying deliveries against invoices, since short deliveries that nobody checks look exactly like internal loss. Check transfers, comps and voids in your point-of-sale, since an unreported comp is indistinguishable from a missing steak in the inventory arithmetic. Restrict access to high-value storage as a matter of process rather than as an accusation.

Then be careful with what you do next. Employment matters carry legal obligations that vary by jurisdiction, and acting on suspicion without evidence can create a larger problem than the one you were solving. Get the measurement clean, document what you find, and take advice from a qualified professional before any action involving a person. Our walkthrough on hiring restaurant staff covers the hiring-side controls that reduce the exposure in the first place.

Moving from clipboard to software without losing the discipline

Sooner or later somebody proposes buying inventory software, usually right after a count that went badly. It is often a good idea, and it will not fix a bad procedure, so the order of operations matters. Get the seven steps stable on paper and in a spreadsheet first, then let a system inherit a process that already works.

What software genuinely buys you is speed and the elimination of transcription errors. It holds every unit conversion permanently, so nobody recalculates $38.40 into $6.40 in a cold room. It pulls invoice prices in automatically, so the price column is never stale. It links recipes to item-level sales counts from your point-of-sale, which means theoretical usage builds itself instead of being a monthly project. And it removes the retyping step, where a correct clipboard becomes an incorrect spreadsheet.

What it does not buy you is accuracy. A system fed by a count taken in a different order, in inconsistent units, at an unfixed time, produces a wrong answer with more confidence and better graphs than a spreadsheet does. The discipline lives in steps one through four, and no product replaces it.

Practical advice if you do buy: insist that the item list is built in your storage order rather than the vendor’s default, since a mobile count screen sorted alphabetically reintroduces the exact problem step one solved. Confirm the integration with your point-of-sale actually exports item-level sales counts rather than category totals, because category totals cannot produce theoretical usage. Ask what happens when a pack size changes mid-year, which is the most common way a conversion silently breaks. And keep the printed sheet as a fallback, because the one night your tablet dies will be the last night of the period.

What counting costs you in labor

The honest objection to all of this is that counting is unpaid work in an operation with no spare hours, so it is worth pricing. In the illustrative kitchen, a full count takes two people about two hours. At an illustrative fully loaded $22 an hour, that is $88 per full count, and thirteen full counts a year is $1,144. The weekly high-value count takes two people about half an hour, roughly $22, and running it in the other thirty-nine weeks adds about $858. Total, somewhere near $2,000 a year of labor, plus pricing and entry time.

Set that against the variance the procedure exposes. The worked period’s $2,200 gap annualizes to about $28,600 at a constant rate. Recovering even half of it is roughly $14,300 against a $2,000 cost. That is a return no equipment purchase in the building will match, and unlike equipment it requires no capital, which is the argument that usually ends the discussion.

Two caveats keep this honest. The recovery is not automatic: finding the variance is step seven and fixing it is a separate program of work involving scales, retraining, recipe updates and a waste log that people actually fill in. And the first few periods produce a variance that is mostly your own paperwork rather than anything recoverable, so the payback arrives on the third or fourth period rather than the first.

Reduce the labor cost where you can without touching the discipline. A sheet in storage order is faster than one that is not. A storeroom organized to match the sheet is faster still. Labeled, dated product is faster to count than unlabeled product. And a count that starts at a fixed time with two named people beats one that starts whenever the last person finishes closing.

Common mistakes when counting restaurant inventory

These are the failures that recur, and each one is cheaper to prevent than to diagnose afterward.

• A count sheet not in storage order. An alphabetical or supplier-ordered sheet forces the counter to crisscross the room, which produces missed lines, and missed lines are the leading cause of an inventory value that jumps for no reason.
• Changing count units between periods. An item counted in cases one period and pounds the next can move a single line by a factor of forty, which is large enough to swamp every other number on the sheet.
• Counting stock and invoices into different periods. A delivery counted on the shelf whose invoice lands in the next period double-counts the food and produces a food cost percentage that is impossible in both directions.
• Counting what the sheet expects rather than what is there. Anchoring is unconscious and it is why high-value lines should be counted blind, without last period's number visible.
• Switching costing methods mid-year. Moving between last cost and weighted average creates a step change in inventory value that looks like an operating result and is not one.
• Skipping the overflow shelf and the line pans. In the illustrative count those two areas carried $2,300 between them, close to the value of the entire dry store, and they are the two most commonly ignored.
• Stopping at the food cost percentage. The percentage describes a result. Only the comparison against theoretical usage points at a cause, and a kitchen that skips step seven spends the month arguing instead of fixing.
• Valuing prep differently in the two counts. Raw component cost in the opening count and standard batch cost in the closing count invents a variance that no amount of line-level investigation will ever explain.

Troubleshooting: strange counts and impossible numbers

What if the food cost percentage came out far lower than usual and nothing changed? Suspect the closing count before you celebrate. A closing count that is $1,000 too high pulls cost of goods sold down to $38,400 and the illustrative 31.8% down to about 31.0%, and it will reverse next period when that phantom stock is not there, producing a spike nobody can explain. An unusually good number and an unusually bad number are equally worth auditing, and the second most useful habit in this whole procedure is treating a pleasant surprise with the same suspicion as an unpleasant one.

What if the two counts of the same shelf disagree by a lot? Recount that section rather than averaging, and look at units first. Roughly nine times in ten a factor-of-twelve or factor-of-forty discrepancy is a case counted as an each, or pounds counted as a case, rather than anybody’s arithmetic. If the units check out, look at whether product moved between the two counts, which is what step three’s frozen window exists to prevent.

What if theoretical usage exceeds actual usage, meaning the kitchen used less than the recipes say it should have? That is not good news. It usually means recipes are overstated, portions are undersized, or the closing count is inflated, and all three are problems. Undersized portions in particular are a guest-facing issue that shows up in reviews long before it shows up here.

What if the variance simply will not close after two clean periods? Widen the search. Check receiving, since short deliveries that nobody verifies look identical to internal loss. Check that comps, voids and staff meals are being reported out of sales. Check for uncosted modifiers, which are invisible in the recipe file and permanent in the variance. And check transfers, especially if you run more than one location or do any offsite catering.

What if the count keeps getting skipped because the period ends on the busiest night of the month? Move your period boundary. Nothing requires an accounting period to end on a calendar month, and a 28-day period ending on a consistently quiet night is worth more than a calendar month ending in chaos.

Your restaurant inventory checklist

Save this and work it in order:

• Written definition at the top of the sheet: what counts as food, and whether beverage, alcohol and paper are in or out.
• Count sheet printed in the physical order you walk your storage, with numbered sections and write-in space at the bottom of each.
• One count unit per item, with the conversion from the purchase unit printed next to it.
• Price column refreshed from current invoices before the count, with any item that moved more than a few percent flagged.
• Costing method chosen, written down, and unchanged: last cost or weighted average.
• Count window fixed: after close on the last night, before the first delivery of the new period, same start time every period.
• Deliveries arriving mid-count quarantined, sealed and recorded in the next period.
• Transfers in and out logged, including catering, second locations and anything borrowed.
• Two people assigned: one counting and calling, one recording and reading ahead.
• High-value lines counted blind, without last period's figure visible.
• Rounding rules written for partial units, line pans and prepared food, applied identically at both ends of the period.
• Sheet priced and entered the same night, with question marks resolved rather than guessed.
• Cost of goods sold calculated as opening plus purchases minus closing, and food cost as a share of matched food sales.
• Inventory turns and days on hand recorded alongside the percentage, and trended.
• Theoretical usage built for at least the top ten items and top ten ingredients, and compared against actual.
• Variance broken into causes, with the residual small enough to be honest about.
• Weekly short count of high-value items running between full counts.

The bottom line

Restaurant inventory is not a hard skill, and almost nobody does it badly on purpose. It goes wrong because it is treated as a chore to finish rather than a measurement to standardize, and a measurement taken differently twice is not a measurement at all. Everything in the seven steps is in service of one idea: make this period’s count and last period’s count describe the same thing, so the difference between them means something. Sheet in storage order, one unit per item, a frozen window, two people, one costing method. None of it is clever and all of it compounds.

The payoff is that you stop at a cause rather than at a number. A 31.8% food cost is a fact you can do nothing with. A $2,200 gap between actual and theoretical usage, of which $880 is portioning drift on one station and $396 is a recipe file nobody has updated since the menu changed, is a list of jobs. That is the difference between a kitchen where food cost is discussed and a kitchen where it moves. Cost the plates with our walkthrough on calculating food cost percentage, close the waste half of the variance with our rundown on cutting restaurant food waste, and run your own opening, purchases, closing and sales through the equipment ROI calculator before the next period ends.

This walkthrough is educational material for operators building a counting procedure, not accounting, tax, legal or employment advice, and it endorses no supplier, system or product. Every dollar amount, percentage, weight and unit cost above is an illustrative sketch chosen so the arithmetic connects end to end, not a quoted price or a measured average, and your own figures will differ by concept, menu, market and supplier. The food cost band mentioned is a commonly used planning range that varies widely by format rather than a verified standard. Food storage, labeling, holding temperature and record-keeping requirements are set locally and change, so confirm what applies to you with your health authority. Where a count raises a question about a person, take advice from a qualified professional before acting.