Restaurant annual budget: the mistakes that drain margin vs the right method

Chapter 1 — Why does the annual restaurant budget fail?

The annual budget fails when it is built as a single sales figure minus an average 30% food cost, because that static model expires in the first quarter.

I have seen it in dozens of operations: the board approves a spreadsheet with a theoretical 12% margin, and by March protein is up 18%, avocado 40% and payroll 9% from turnover. The right budget does not project annual sales; it models Prime Cost —inputs plus labor— week by week, because that is where 60% to 68% of revenue lives. A restaurant billing 1.2 million USD a year moves roughly 23,000 USD of weekly Prime Cost; measuring it every 52 weeks, not once, is what separates the operator who corrects in time from the one who discovers the leak in December. The budget is a flight deck, not an accounting rearview mirror. Prime Cost is the sum of input cost plus total labor cost, and in a healthy restaurant it holds between 55% and 65% of sales; above 68% the operating margin disappears.

Chapter 2 — Prime Cost: the number the board must watch week by week

The mistake I see again and again is budgeting food cost and payroll separately, two silos never read together. When the chef drops food cost to 28% by piling overtime onto the line, labor climbs to 38% and real Prime Cost lands at 66%: worse than before. That is why a Prime Cost budget forces the combined weekly reading. At Masterestaurant we build the budget on 52 cells, one per week, with a Prime Cost target set at 60% and a tolerance band of ±2 points. A 3-point deviation on sales of 23,000 USD is 690 USD draining that week. That is what the management lever corrects, not the annual average. Separating CapEx from OpEx is not an accounting formality: it is the only way for the board to see real EBITDA. When a 22,000 USD combi oven purchase is diluted into monthly expense, the operating margin looks 3 to 5 points worse than it is, and an expansion decision is made on a false base.

Chapter 3 — Separating CapEx from OpEx so EBITDA does not lie

CapEx —equipment, construction, remodeling— is capitalized and depreciated over 5 or 10 years; OpEx —inputs, payroll, rent, utilities— lives in the month. A restaurant investing 180,000 USD in build-out does not charge those 180,000 to year one: it charges 18,000 to 36,000 in depreciation. Mixing them sinks the budgeted year-one EBITDA and scares off any investor who reads well. The Prime Cost budget keeps both lines apart so the projected EBITDA —ideally 15% to 20% of sales— reflects the pure operation, not the noise of investment. Theoretical vs actual cost turns the budget into a control system, because without theoretical cost there is no variance, and without variance the operator finds the leak only in the December inventory, when the year is already lost. Theoretical cost comes from recipe cards: if the recipe calls for 180 grams of tenderloin at 22 USD/kg, each plate costs 3.96 USD on that line, and multiplied by item sales it yields what you should have spent.

Chapter 4 — Theoretical vs actual cost: the margin control system

Actual cost comes from inventory. The gap is variance, and in a careless operation it runs between 3% and 6% of purchases: shrinkage, theft, over-portioning, waste. On annual purchases of 400,000 USD, a 4% variance is 16,000 USD gone without an invoice. A budget that integrates both costs lets you set a variance target of ≤1.5% and audit it every week, not every quarter. Contribution margin per dish —selling price minus input cost— is the real unit the budget must project, not percentage food cost, which deceives. A dish with 25% food cost and 8 USD of margin puts less cash in the till than one with 35% food cost and 14 USD of margin if both sell the same. Menu engineering crosses margin and popularity into four quadrants: stars (high margin, high turnover), plowhorses (low margin, high turnover), puzzles and dogs. A serious budget projects the mix: if stars carry an average margin of 15 USD and make up 40% of orders, shifting that mix two points via menu design or server suggestion changes annual EBITDA by tens of thousands.

Chapter 5 — Contribution margin per dish with menu engineering

At Masterestaurant we budget by weighted contribution margin, not average food cost, because the average hides which dish pays the rent and which one drains it. The dynamic break-even is the sales level where contribution margin covers fixed costs, and it must be recalculated every month because fixed costs change: indexed rent, base payroll, utilities. Here I load fixed payroll, rent, utilities and insurance onto break-even, never onto the plate. If monthly fixed costs are 45,000 USD and the average contribution margin is 62% of sales, break-even is 72,600 USD of monthly sales; below that figure the operation loses cash every day it opens. Presenting the board a static January break-even when rent rose 8% in July and two cooks were hired is lying to them. The right budget carries the break-even recalculated month by month and crosses it with projected sales, so the director knows how many daily covers —not how many wishes— sustain the house.

Chapter 6 — EBITDA projection across three stress scenarios

A resilient EBITDA projection demands three scenarios —base, conservative and stress—, never a single figure that reality disproves in the first month. The base scenario assumes expected sales and a 60% Prime Cost; the conservative cuts sales 12% and raises inputs 8%; the stress case models a 25% traffic drop with 15% input inflation, the blow the last bout of volatility delivered. A restaurant with a projected base EBITDA of 18% falls to 9% in the conservative and grazes 2% in stress: that number is what defends an expansion before the board, because it shows the house survives the worst quarter. Diego F. Parra builds every Masterestaurant budget with these three lanes and an action threshold: if two straight months fall below the conservative, the menu and purchasing levers activate. A budget without scenarios is not a plan, it is a wish with decimals. The traditional budget answers the accounting question 'how much did we sell and how much was left?'; the Prime Cost budget answers the managerial question 'where is margin draining this week and which lever fixes it?'.

Chapter 7 — The difference a CFO sees

The first is a rear-view mirror; the second is a flight deck. Splitting CapEx from OpEx is not an accounting formality: it is the only way for the board to see real EBITDA. When equipment purchases dilute into monthly expense, operating margin looks worse than it is and expansion decisions rest on a false base. Theoretical vs actual cost turns the budget into a control system. Without theoretical cost there is no variance, and without variance the operator only discovers capital leakage at the December inventory, when the year is already lost. With weekly variance, a 2-point deviation is corrected in 7 days.