How to make a restaurant profitable: the six mistakes eating your margin and the method that gives it back

Start with the standardized recipe, not with the price list

Before you touch a single price, write the spec sheet for every dish with exact gram weight, declared trim loss and cost per portion, because without that document any menu adjustment is an expensive hunch. The deliverable is measurable: one sheet per recipe showing unit cost, food cost percentage and the date the ingredients were last valued, and you verify it by pulling three dishes at random, weighing what leaves the line and comparing it against the sheet; if gram weight drifts more than 5%, the sheet is decorative. The food cost ceiling per dish in the Masterestaurant method is 32%, and that 32% is a MAXIMUM, never a target. With net profit for a full-service restaurant sitting between 3% and 5% according to Statista, four points of drift in food cost do not bite into your margin: they take all of it. Contribution margin in dollars, not food cost percentage, is the number that decides what stays on the menu.

Work out contribution margin per dish and rank the menu by dollars, not by percentage

A dish running 38% cost that leaves 14 dollars per check beats one at 24% leaving 5, and in the 92-seat corporate restaurant that opened this guide, the fourteen best sellers were precisely the lowest contributors: servers pushed them because they were cheap. Build a four-column table —dish, units sold over 90 days, unit contribution, total contribution— and sort by the last one. You will see that 20% of the menu usually carries more than half the money. The deliverable: that table, signed, with dishes classified as star, cash cow, question mark and dog. You verify it by adding total contribution and setting it against fixed expense for the same quarter. Prime cost —food plus beverage plus fully loaded payroll— belongs between 55% and 60% of net sales, and you read that number every Monday, not when the accountant shows up.

Set prime cost at 55-60% and watch it weekly, never monthly

Payroll already accounts for more than 25% of restaurant expenses in 2024, up from 23% in 2021 according to Toast, and profitable operators run payroll at 34,2% of sales against 36,5% for the full-service average, per the National Restaurant Association. Those 2,3 points of difference are, in a room billing 218.000 dollars a quarter, roughly 5.000 dollars that appear or vanish without anyone noticing. The deliverable is a four-line weekly dashboard: net sales, food cost, beverage cost, loaded payroll. When prime cost breaks 62% two weeks running, stop buying and rework the schedule. Dropping the new hood, the convection oven or the restroom remodel into this month's expense is the most common way to believe a restaurant loses money when it actually earns. CapEx —anything lasting more than a year— gets capitalized and depreciated; OpEx —ingredients, payroll, energy, routine maintenance— lives in the period's income statement.

Split CapEx from OpEx before you look at any profit figure

Mixing them warps EBITDA, and EBITDA is the figure they will value your business with: a single-unit independent sells between 1,5x and 3x SDE, and EBITDA multiples run around 2,80x to 3,65x according to the Sofer Advisors valuation guide, with fast-casual between 4x and 7x. The deliverable: an asset schedule listing purchase date, value and useful life. You verify it by reconciling monthly depreciation against that schedule. Here sits the tension almost nobody resolves: cutting food cost and raising guest satisfaction look like opposing forces, because everyone assumes low cost means small portions or bad product. The bridge is YIELD. A whole beef primal broken down in house delivers between 12% and 18% more sellable plate than the same cut portioned by the supplier, and the trim feeds the stock, the ragù and the staff meal. For years I attacked the problem through purchasing, haggling over the price per kilo, celebrating 6% discounts; it helps, though the real money was never there.

Buy yield, not price per kilo

It sat in the mix of what gets sold and in the fixed structure nobody had reviewed since opening day. The deliverable: a documented yield test per protein, recording gross weight, clean weight and the true cost of the sellable portion. The most expensive mistake is not buying dear: it is raising prices before standardizing recipes, because you hand your own disorder to the guest and the guest notices by the third month. Four others follow, and they repeat with a consistency that stopped surprising me. Second, loading payroll and rent onto plate cost, when those belong to the break-even calculation and not to the recipe. Third, counting inventory once a month, which makes it impossible to trace shrinkage to its cause. Fourth, deciding the menu on food cost percentage instead of dollar contribution. Fifth, treating discounts as marketing without calculating how many extra checks recover the margin point you gave away; with net profit at 3% to 5%, a 10% discount demands nearly double the volume of that dish.

The five mistakes that will cost you the quarter

The deliverable here is a signed list of the ones that apply to your room. With the contribution table in hand, the intervention turns surgical: pull the two dishes with negative contribution, reformulate the two with low contribution and high turnover by changing the garnish or the cut, and lift the stars to better visual real estate. No redesigning the whole menu. As Diego F. Parra, restaurant consultant and founder of Masterestaurant, puts it, a restaurant does not fail from buying expensive but from selling badly what it bought well, and that sentence becomes operational right here. If your average check is 26 dollars and you shift 3 dollars of contribution across the four dishes covering 40% of checks, across 8.000 quarterly checks that is nearly 4.000 dollars landing straight on the bottom line. The deliverable: menu version 2, dated, with total contribution measured again at day 45 to confirm the move.

Closing checklist: how you know everything landed

You will know the guide got executed when you can answer six questions with a document instead of an opinion. One: does a spec sheet exist for 100% of active dishes, valued with ingredient prices from the last 30 days? Two: does any dish break 32% food cost, and if so, is it justified by dollar contribution? Three: did prime cost over the last four weeks land inside 55-60%? Four: does the asset schedule reconcile with depreciation, leaving EBITDA clean of CapEx? Five: did payroll move toward the 34,2% that profitable operators run, per the National Restaurant Association? Six: does quarterly total contribution cover fixed expense with room to spare? If a single one fails, go back to that step before touching prices. And start tomorrow with the simplest: weigh three portions and compare them against the sheet. The underlying difference is not accounting, it is sequence.

Where the two routes split?

The mistake route starts with price because price is the one thing an owner can change in an afternoon; the method starts with cost structure because that is the only thing explaining why the price falls short.

A restaurant that raises the menu without standardized recipes is passing its own disorder to the guest, and the guest notices by month three. There is a genuine tension worth resolving before we continue: cutting food cost and raising guest satisfaction look like opposing forces, because everyone assumes low cost means small portions or cheap inputs. It does not. The bridge is YIELD: a primal cut bought whole and broken down in house delivers 12% to 18% more sellable plate than the same cut bought pre-portioned, per the butchering yield ratios published by the USDA (2025). You are not shrinking the portion; you stopped throwing money into the trim bin. The second split sits in what counts as an expense.

Where the two routes split — in practice?

When an owner pays 9,400 dollars for an exhaust hood and books all of it in the month, that is not conservative accounting, it is manufactured misinformation:

that month looks terrible, the next looks brilliant, and on two lying months you make menu and staffing calls that do not fit reality. CapEx amortizes against asset life; OpEx hits the period. Confusing them is the quietest capital leak I know, because it never shows up as a line: it shows up as bad decisions. Then comes the mix, where the big money lives. Two restaurants with an identical 30% food cost can finish the year at 4% and at 12% profit, and the whole gap sits in WHAT sold: if your three star dishes return 6.20 dollars of unit margin and make up 41% of tickets, you own a business; if your three best sellers return 2.80 and make up 44%, you own a badly paid job with an apron.