Inventory control: the storeroom count versus the Masterestaurant method

Which one wins for a restaurant with table service and a bar: weekly counting or the Masterestaurant method?

The Masterestaurant method wins, and the gap is not a matter of nuance:

it closes the distance between what leaves the storeroom and what gets charged on the check, which is exactly where 4% to 10% of purchased food evaporates in a table-service restaurant. A 96-seat Italian place in Bogotá closed March with a theoretical food cost of 29.4% and a real one of 36.8%, seven and a half points on food sales of 214 million pesos, nearly 16 million that nobody stole from the register. After eleven months of spreadsheets, scales and two people cross-checking, the Sunday count handed the owner the exact size of the shortfall; what it never handed him was the shift, the item or the name. Counting measures the balance. The method measures the event, and the event is the only thing you can fix on Monday. Traditional counting is a snapshot of the balance while the Masterestaurant method is a film of consumption, and that difference hits the books immediately.

Traceability of the shortfall: 0% assignable versus 87% carrying shift and name

With the snapshot you charge a 3.2 million shortfall to the monthly P&L without assigning it to any cost center: it lands as global shrinkage, zero percent traceability dressed up as rigor. With the film, 87% of that same shortfall arrives tagged with shift, product reference and the server responsible, so the owner can decide whether he is facing a training issue, a badly written recipe, or a supplier shipping 940 grams against a billed kilo. With food costs up 35% in five years according to the National Restaurant Association, arguing about a shortfall you cannot assign is arguing about the weather. The method WINS here, because a number without an owner changes nobody's behavior. The storeroom answers for what COMES IN and the dining room answers for what goes out billed, and practically every counting system audits only the first half of that sentence.

The half of the sentence nobody audits: 100% comes in, 92% goes out billed

I got this wrong for years: I believed a flawless storekeeper shielded the food cost, so I built ever finer receiving controls —double weighing, temperature seals, signed counter-orders— while the bar kept pouring by eye and no paper ever recorded the difference. At Masterestaurant we start from an idea the trade finds uncomfortable: the main leak in a table-service restaurant does not live in the storeroom, it lives in the meter and a half between the bar and the guest. That space holds the comp nobody keyed in, the drink poured by feel, the replacement for a dropped plate, the dessert given away to quiet a 5-minute complaint. Counting sees none of those four events; order versus dispatch sees all of them. Fourteen kilos of loin were missing in the Sunday count, and the traditional system left the owner with four theories and no evidence. Did they go out in fourteen badly portioned plates on Thursday, in two comps the server never rang up on Friday, in Saturday cooking loss, or through the back door the previous Monday?

March at the Bogotá Italian: 14 kilos of loin and four theories with no evidence

The number existed, the diagnosis did not, so he did what everyone does: leaned on the chef, switched meat suppliers, cut the portion from 220 to 200 grams. The hole came back in April in the same place. Once we tied every dispatch to its order ticket, 61 plates appeared served with no check attached across two Friday shifts, plus a real average portion of 247 grams against a 220-gram recipe: 27 grams per plate that no Sunday was ever going to expose. Diego F. Parra puts it dryly: a shortfall is not counted, it is traced. Counting every Sunday burns 5 to 7 man-hours per session, roughly 24 hours a month with two people and a scale, and it delivers a number up to 7 days stale: by the time you spot Monday's deviation, six services have repeated it.

Frequency and operating cost: 4 counts a month versus daily measurement in 12 minutes

The Masterestaurant method flips the allocation: it keeps one monthly physical count for verification and shifts the weight of control onto the daily close of order versus dispatch, which in a 96-seat operation takes 12 minutes at closing and is done by the same shift manager already balancing the till. Fewer total hours, yesterday's data in your hand today. With labor up 35% since 2019 according to the National Restaurant Association, spending 24 monthly hours to produce a figure you cannot act on is an odd luxury. The method wins on frequency and on cost, no argument. Traditional counting fixes what you buy and the Masterestaurant method fixes what you serve, and in a restaurant with a bar the second lever moves far more money. Renegotiating suppliers off a good count can rescue somewhere between 0.8 and 1.5 points of food cost, which is real and no owner should sneer at it.

What each system actually fixes: purchase price versus service behavior?

But the gap between theoretical and real in the Bogotá case ran 7.5 points, five times that rescue, and no price negotiation touches it because the problem never sat on the incoming invoice;

it sat in the portion, the comp and the drink poured by feel. The apparent tension resolves easily: this is not buying well versus serving well, it is that buying well has a low ceiling and serving by measure does not. If your gap clears 3 points, leave price alone and walk into the dining room. Project the case out a year and you will see why this is not a methodological debate. Seven and a half points of gap on annual food sales of 2,568 million pesos come to 192 million walking out of EBITDA with no invoice, no theft and nobody accountable, close to the annual payroll of eleven kitchen staff. With traditional counting you reach December knowing to scale-level precision how much you lost and without a single verified hypothesis about where.

The scenario that settles it: what happens if you go twelve months without measuring the floor?

With daily floor measurement, average portion gets corrected in week 2, unrecorded comps get keyed in from week 3, and a conservative recovery of 4 points leaves 102 million in the bank.

The venue's rating does not suffer either: each additional review star is worth 5% to 9% of revenue according to Harvard Business School, and even portioning shows up at the table before it shows up in the P&L. If your operation runs a bar, table service and more than two pairs of hands between the storeroom and the guest, install the Masterestaurant method and keep the physical count monthly, purely as verification; that is where 4 to 10 points of leakage justify any discipline. If you run a counter format with no bar, a low check average, a 12-item menu and one person dispatching, a well-executed weekly count is enough and building daily traceability would be over-engineering.

What to choose for your profile: bar, check average and how many hands stand between storeroom and table?

The practical breaking point is the gap: measure one month of theoretical food cost against real, and if the difference clears 3 points, your count is already failing you even when the spreadsheets balance perfectly.

Start cheap. This Sunday, instead of counting everything, take your 5 most expensive items and cross them against the order tickets from the two previous Fridays. The traditional count is a photograph of the balance; the Masterestaurant method is a film of consumption. That sounds like consultant metaphor, yet it has a direct accounting consequence: with the photograph you charge a 3.2 million shortfall to the monthly P&L and cannot assign it to any cost center, while with the film 87% of that shortfall carries a shift, a reference and a name, so the owner can decide whether the problem is training, recipe or supplier. The storeroom answers for what COMES IN and the floor answers for what GOES OUT billed, and every traditional system audits only the first half of that sentence.

Where the two methods truly split?

I got this wrong for years: I believed a sharp storekeeper armored the food cost, so I built ever finer receiving controls while the bar kept free-pouring whisky.

The storeroom was perfect. The check still did not add up. Speed of data changes what can actually be corrected. A deviation you learn about twelve hours later still has witnesses, camera footage, a ticket and memory; eight days later it is an argument between opinions. Shift-level cutoff is therefore not a technology luxury: it is the only window in which server training still functions as correction rather than as reproach. The real cost of each method is inverted relative to what most owners assume. The traditional count feels free because nobody invoices it, yet it burns 16 to 24 labor hours a month from salaried staff, plus the invisible cost of those seven gap points; the Masterestaurant method demands two uncomfortable weeks of training and afterwards cuts control OpEx to a third.

Where the two methods truly split — in practice?

One measures product, the other measures behavior. That is the line. Inventory has never stolen anything, never over-poured a drink, never handed out a dessert to calm an angry table:

people under service pressure do that, and people are not corrected by a Sunday spreadsheet, they are corrected with judgment, repeated practice and a clear rule about what may be given away and what may not.