Inventory management: the traditional method against the Masterestaurant method

Step 1: set the perimeter before counting anything

Before the first count you must decide WHAT goes under control, and the short answer is: the 20 items that concentrate 70 % to 80 % of food and beverage spend, never the 400 sitting in the storeroom. What this step delivers is a single sheet with twenty lines, each carrying its purchase unit, its recipe unit and the conversion factor between them, because protein is bought in kilos and served in 180-gram portions, and that gap is where half of all variance errors are born. Verify it this way: ask a line cook to convert a case of tenderloin into portions without a calculator; any hesitation means the factor is written wrong. The National Restaurant Association estimates that 4 % to 10 % of purchased food is wasted, and that range gets decided in these twenty lines, not in the other three hundred eighty. Without a standardized recipe you don't have inventory, you have a storeroom with paperwork.

Step 2: the standardized recipe is your measuring instrument

Theoretical consumption —the number everything else is compared against— comes from multiplying each dish sold in the POS by the grams your recipe declares, so a recipe written by eye produces an invented variance. The deliverable here is spec sheets for the dishes that add up to 80 % of sales: ingredient, net gram weight, trim loss declared as a percentage, and cost per portion. Tenderloin carrying 18 % cleaning loss costs 22 % more per portion than the same tenderloin declared without it, and that difference eats your margin without ever appearing on an invoice. Verify by weighing three portions served during peak service, not during a quiet morning: if real gram weight drifts more than 8 % from the declared figure, fix the recipe or fix the hand, but stop counting. Cadence rules over decimal precision. Counting twenty items weekly, always the same day and the same hour —Sunday close, or before Monday's first delivery— gives you more control than an exhaustive monthly count of four hundred, because the drift shows up while you can still trace it to one specific shift.

Step 3: weekly count of the twenty, same hour, two signatures

Two people count: one calls the figure out loud, the other writes it down, and neither of them is the person who buys. The deliverable is a closed worksheet with opening stock, purchases for the period, closing stock and calculated real consumption. Done properly it takes forty minutes. Supy documents margin gains of 2 % to 10 % in operations that move to weekly audits with inventory tools, and that range is exactly the distance between a restaurant that argues and one that decides. The formula fits on one line: real consumption minus theoretical consumption, divided by theoretical consumption, and the result as a percentage is your variance. That subtraction —the one the February manager never performed across forty-three immaculate sheets— separates knowing how many cans you have from knowing how much money walked out. Diego F. Parra insists at Masterestaurant on breaking it out BY STATION rather than by restaurant: protein, bar, dairy, bakery, each with its own number.

Step 4: the subtraction that turns a census into control

A venue billing 1.2 million dollars a year moves roughly 360,000 dollars in food at a 30 % food cost, so one variance point equals 3,600 dollars a month, the salary of a full-time cook. What this step delivers is one figure per station, dated, within 45 minutes of closing the count. Take three weeks and it stops being control: it becomes archaeology. Variance between 1 % and 2 % is a healthy operation and deserves no meeting. Between 2 % and 3 % something is loosening —portioning, receiving, or waste logging— and a week of watching the flagged station usually fixes it. Above 3 % the problem is structural, and above 5 % your business is financing somebody: Sculpture Hospitality attributes 75 % of inventory shrink in the industry to internal theft, an uncomfortable figure worth facing before you shop for software. That said, don't jump to the easy conclusion. I got this wrong for years, treating every high variance as theft when most of the time it was a badly loaded recipe or a purchase unit that quietly changed size.

Step 5: read the number before you accuse anyone

Audit the paperwork first, then the station, and only at the very end the people. The most expensive mistake isn't counting badly: it's counting well and never subtracting. Four others show up in almost every kitchen. One, counting work in progress —sauce already made, marinade resting— without valuing it, which inflates protein variance and hides the bar's. Two, letting the same person buy, receive and count, which turns control into a declaration. Three, failing to log trim waste and front-of-house comps, which in a 240-cover restaurant easily add up to 1.5 % of cost. Four, switching suppliers without updating the conversion factor, where a case that dropped from 12 to 10 kilos generates a phantom 20 % drift on that line. Each of those four produces numbers that look like theft and aren't, and chasing ghosts burns the credibility of the whole system. Follow the chain all the way down.

What happens if you don't close the variance this quarter?

At a sustained 3.5 % variance in a venue billing 1.2 million dollars, you lose around 12,600 dollars a year in product you bought and never sold;

replacing that money at an 8 % operating margin requires 157,500 dollars in additional sales, which at a 25-dollar ticket means 6,300 more customers. Nobody lands 6,300 new diners in a quarter, yet any manager can pull variance from 3.5 % down to 1.8 % in eight weeks with weekly counts and corrected recipes. Here sits the paradox of the trade: the fastest route to more cash isn't selling more, it's paying less for product that never reaches a table. Your software vendor will never tell you that, because the commission lives in the module, not in your margin. You finished the implementation well when you can answer five questions without opening a file. First: what was protein variance last week, in percentage and in dollars?

Closing checklist: how to know everything landed

Second: who counted and who wrote it down, and does neither of them sign purchase orders? Third: do the spec sheets covering 80 % of sales carry declared trim loss, and were they weighed in service within the last thirty days? Fourth: does the report land within 45 minutes of the count, or does it still travel by email until Tuesday? Fifth: does the worst-performing station have an owner with a name and a review date? If all five answers arrive in under two minutes, inventory stopped being a folder and became a cash tool. Start Monday with protein and leave the other stations alone until you string together three consecutive weeks below 2 %. LATENCY. Paper hands you the drift 22 to 35 days after it happened, once the supplier has been paid and the product has been served; the Masterestaurant method hands it over in under 45 minutes. One point of food cost in a USD 1,2 million unit is worth roughly USD 3.600 a month, so thirty days of blindness cost about what a line cook costs.

The four differences that move cash

GRANULARITY. Knowing food cost climbed from 29 % to 32 % gives you nothing to act on: it flags a problem without locating it. Once variance opens by station — protein 3,1 %, bar 1,4 %, bakery 0,8 % — you stop auditing a restaurant and start auditing one station, and management time spent hunting for the cause drops from hours to minutes. INDEPENDENCE. Under the traditional method the owner validates the count; that is a ritual of trust, not a process. Board a plane for two weeks and the folder stays empty. Process standardization turns that signature into a numeric checkpoint — variance under 2 %, zero out-of-range lines — that any manager can close, and that is where owner-independent operation actually begins. TRACEABILITY. Food safety is not a separate chapter of inventory management, it is the same data read from another angle. If the receiving photo carries lot, temperature, and date, a product withdrawal is resolved in twenty minutes; if your only evidence is marker on a label, you dump the whole walk-in as a precaution and eat the difference.