Over-portioning: the silent leak draining your EBITDA (myth vs reality, 2026)

Chapter 1 — Why doesn't a 30% food cost mean your portions are under control?

A 30% food cost proves nothing about portions: that average has already absorbed over-portioning and hides it inside the aggregate. The P&L consolidates purchases, inventory and sales into a single late number;

by the time it flags a problem, you have already lost three months of margin. Across dozens of operations we audited at Masterestaurant, food cost closed at 29% or 31% while the variance between theoretical and actual cost ran 4 or 5 hidden points. The arithmetic is plain: if your theoretical cost from standard recipes is 26% and the real one closes at 31%, those 5 points are pure leakage —over-portioning, waste and theft blended together— yet the P&L only shows you the final 31%. Diego F. Parra puts it bluntly: the average is a rearview mirror; it tells you where you have been, not where you are bleeding right now. Each extra gram turns into lost cash, and the effect multiplies by volume until it moves EBITDA points.

Chapter 2 — How much capital does each gram of over-portioning actually leak?

Take a plate costed at 145 grams of protein that routinely leaves the pass at 172 grams: that is 27 grams of excess, an 18.6% over-portion on that input.

If the protein costs 18 USD per kilo, each plate gives away 0.49 USD; at 4,000 covers a month in that family, that is nearly 1,950 USD monthly and 23,400 USD a year on a single item. Multiply it across six or eight plates with the same deviation and you face six annual figures the 30% food cost never showed you broken down. I have seen it again and again: the owner celebrates a 'healthy' food cost while two or three signature plates bleed the contribution margin of the whole menu. Over-portioning is not a kitchen slip; it is an accounting line the operation chose not to measure. In 2026 over-portioning hurts more because the margin that used to absorb it disappeared.

Chapter 3 — Why did 2026 leave over-portioning without a margin cushion?

Input inflation and protein volatility pushed up the cost per gram, while prime cost pressure —food plus labor— compressed the contribution margin per plate.

A restaurant that tolerated a 6% portion deviation in 2022 and diluted it in a comfortable margin now sees it translate straight into lost cash, because the input costs more and every point weighs double. If your contribution margin per plate dropped from 68% to 61% in three years, those 7 lost points leave no room to give away 18% of protein. The math is cruel: the same operational error, three times the financial impact. That is why the right instrument in 2026 is not watching average food cost, but closing the variance by menu family before the monthly close turns it into consummated, irreversible loss. The variance between theoretical and actual cost isolates over-portioning from waste and theft, and turns it into its own accounting line the board can demand you close.

Chapter 4 — What does theoretical vs actual variance measure that average food cost cannot see?

Theoretical cost comes from standard recipes multiplied by real period sales: it is what you SHOULD have consumed. Actual cost comes from physical inventory:

what you truly consumed. The difference is the variance, and when you break it down by menu family and by shift, over-portioning stops being a suspicion and becomes an enforceable number. At Masterestaurant we measured operations where global variance was 4.2%, but broken down, 70% concentrated in two families and the night shift: that detail turns a diffuse problem into a concrete action. Average food cost never gives you that granularity; it consolidates everything into one digit that hides exactly the information you need to act. The most expensive difference is not how many grams are wasted, but when you find out: it is a pure timing problem. Catching over-portioning at the monthly close means losing the margin of the full 30 days, because the data arrived after the cash was already gone.

Chapter 5 — Why is the most expensive difference in over-portioning a matter of timing?

Catching it in a weekly count means losing, at most, seven days before correcting. The arithmetic is blunt:

if a plate leaks 1,900 USD monthly, monthly control lets you lose the whole sum, while the weekly count trims exposure to roughly 440 USD per detected cycle. Diego F. Parra insists on this with owners and CFOs: you do not buy a counting system for statistical precision, you buy it for reaction speed. Every week you shorten the measurement cycle is margin you rescue. Variance measured late is an autopsy; measured on time it is a tourniquet that stops the capital hemorrhage. The 90-day roadmap closes the variance in three measurable stages you can present to the board. Days 1-30: cost real standard recipes for your 15 highest-volume plates and set the theoretical cost; install line scales and define the target gram weight per plate. Days 31-60: implement weekly inventory counts by menu family and calculate the first real vs theoretical variance; expect to uncover 3 to 6 hidden points in the problem families.

Chapter 6 — How do you execute a 90-day roadmap to close the variance?

Days 61-90: set tracking KPIs —variance per family below 1.5%, gram-weight deviation below 4%, weekly count met at 100%— and build the board dashboard.

The realistic goal is recovering 2 to 4 food cost points, which on a 250,000 USD monthly sales operation means 5,000 to 10,000 USD recovered every month. This is not a kitchen project; it is a capital decision with a return measurable in the first quarter. Average food cost is a rearview mirror: it tells you where you were, not where you are bleeding. Theoretical vs actual variance is the near-real-time dashboard: it isolates over-portioning from waste and theft and puts it as its own accounting line the board can demand you close. The myth treats portioning as a kitchen matter; the reality treats it as a cost-structure matter. A plate costed at 145 grams of protein that goes out at 172 grams is not a cook's oversight: it is a control-system failure that, multiplied by thousands of covers, moves whole EBITDA points.

Chapter 7 — The differences that decide the margin

The most expensive difference is timing. Finding over-portioning at the close means losing a full month of margin; finding it at the weekly count means losing seven days at most. In a restaurant billing 120,000 USD/month with a 3% deviation, that latency difference is worth around 2,700 USD recoverable every month.