Menu Engineering & Yield-Adjusted Plate Cost Calculator

Why is food cost percentage the wrong number to manage?

Because it's a ratio, and you bank dollars.

The industry states it plainly: a six-dollar appetizer at 30% food cost contributes $4.20 per plate, while a twenty-eight-dollar entrée at the same 30% contributes $19.60. Identical percentage, completely different profit.

The calculator's defaults make the point sharper because the percentages differ too. The soup runs a 23.13% food cost — the best ratio on the menu, the number that would win any food cost review — and puts $6.92 in the till. The ribeye runs 40.40%, the worst ratio on the menu, and puts $30.99 in the till.

Chase the ratio and you promote the soup. Chase the money and you promote the ribeye. The gap is $24.08 per plate.

Model a modest 5% shift in covers from one to the other through menu placement — the golden triangle, a server callout, a box on the page — and across a year that's worth $114,554.

None of this means food cost percentage is useless. It's the right number for tracking purchasing discipline, spotting theft and waste, and comparing period to period. It's the wrong number for deciding which dish to feature, and the two jobs get confused constantly.

What is the yield percentage and why does it change my plate cost?

Yield is edible portion divided by as-purchased weight. A case of romaine isn't all romaine — you throw away the butt and the outer leaves. A whole ribeye loses weight to trim and again to cooking. What you paid for and what reaches the plate are different quantities, and only one of them is on the invoice.

So a $14.80 portion of ribeye at 72% yield actually costs $20.56 of purchased product, before you add anything else.

The calculator's defaults show what that does to the headline number. Cost straight from the invoice and the menu reports a 29.08% food cost — comfortably inside the 28–35% band that most full-service restaurants target, the sort of number that ends the conversation. Cost at edible portion and the real figure is 35.86%, at the very top of the range.

The difference is $2,923.73 a period, or $152,034 a year of cost that exists but never appears in the costing.

This is the specific failure that recipe software vendors point to when they argue spreadsheets can't do the job — that spreadsheets don't account for yield and prep loss at the sub-recipe level. They're right that it's the failure. They're wrong that a spreadsheet can't do it; it just has to divide by yield instead of ignoring it.

Get your yield figures from a butcher test or a prep log. Guessing them is better than assuming 100%.

How do the four menu engineering quadrants get calculated?

Two thresholds, then a 2×2 grid.

Profitability threshold is the weighted average contribution margin across the whole menu — total contribution divided by total covers, so busy items count more. The defaults give $15.12.

Popularity threshold is where most people go wrong. It isn't a flat average. Take 1 divided by the number of menu items, then multiply by 0.70. On a six-item menu that's 11.67%, not 16.67%.

That 0.70 adjustment matters: with a pure average, every item on a ten-item menu would need to sell at least 10% of the time to qualify as popular, and ordering behaviour is never distributed evenly. The factor accounts for that variation and stops the threshold becoming unrealistically strict.

Watch the salmon. It clears the profitability threshold comfortably at $20.17 but sells at 11.48% against an 11.67% popularity threshold — it misses Star status by 0.19 percentage points. Classifications near a boundary are fragile, and a Puzzle that close to the line is really a Star waiting for better placement rather than a dish with a problem.

What do I actually do with each quadrant?

The classification is a prompt, not an instruction. Each quadrant asks a different question.

• Stars — protect them. Don't raise the price casually, don't cut the portion, don't let quality drift. These carry the menu.
• Plowhorses are the interesting case, and the defaults have two of them. The burger sells 33% of covers at an $11.46 margin; the pasta sells 23% at $13.78. They're popular for a reason, which means small changes reach a lot of covers. A dollar of price or fifty cents of cost improvement on the burger alone is worth real money precisely because volume is high. But they're also the items customers notice, so move carefully.
• Puzzles are usually a merchandising problem rather than a food problem. The salmon earns well and nobody orders it. Reposition it, rename it, train servers to recommend it, move it into the golden triangle. You're trying to convert it to a Star without touching the recipe.
• Dogs are the only quadrant where removal is on the table, and even then not automatically. The soup is a Dog on these numbers, but a menu without a soup may lose covers entirely, and low-margin items sometimes exist to anchor a price point or serve a dietary need. Ask what the dish is for before deleting it.

One honest limit: this is a food cost model. Labour per plate is not included, and it should influence the decision. A dish with high food cost but almost no prep time can contribute more to fixed costs than a cheap dish that ties up a station through service.

How often should I re-run this, and what data do I need?

Quarterly at minimum, and immediately after any significant price change from a supplier.

You need four things per item, all of which you already have:

• The menu price.
• The as-purchased ingredient cost, from recent invoices rather than last year's costing.
• The yield percentage, from a butcher test or prep log.
• Units sold, straight from your POS mix report for a representative period.

Two practical notes on the data. Use a period long enough to be representative but recent enough to be true — a single week during a holiday will mislead you. And enter the complimentary cost per cover separately: bread service, garnishes and condiments are real cost that belongs in the plate but doesn't appear in any recipe.

The reason to re-run often is that both axes move. Ingredient costs shift with every delivery, which moves contribution margins and therefore the profitability threshold. And the mix moves as customers respond to whatever you changed last quarter — which is the point of doing it, but it means last quarter's classifications expire.

A menu is not engineered once. The classifications are a snapshot of a system that keeps moving, and the value comes from watching items migrate between quadrants over time rather than from any single reading.