Case Study 2: Price It or Shrink It

The contested decision every operator faces, and the arithmetic that makes the wrong answer so attractive

Note on sources. The industry backdrop in this case is public record. The 120-seat restaurant whose numbers are worked below is a clearly labeled composite, built from patterns that are common across independent full-service operations. Its figures are internally consistent and illustrative; they are not any real business's records.

Background: what happened to menu prices

Between 2021 and 2024, restaurant menu prices in the United States rose at a pace not seen in decades. The Bureau of Labor Statistics tracks this as "food away from home," and for a sustained stretch it outran grocery inflation — meaning that eating out became more expensive relative to cooking at home, which is the comparison guests actually make.

Operators had reasons. Commodity costs rose. Wages rose sharply in the post-2020 labor shortage. Delivery commissions, packaging, insurance, and card fees all rose. Restaurants raised prices because the alternative was to run a business with no margin.

By 2024, the guest response was visible in the numbers, and it was visible first at the bottom of the market. Several large quick-service chains publicly reported softening traffic, particularly among lower-income guests, and reversed course with value offers — most prominently McDonald's, which launched a widely reported \$5 Meal Deal in June 2024 and extended it as it proved popular. At the same time "shrinkflation" entered the mainstream consumer-affairs conversation. It began as a packaged-goods complaint, but the concept transferred instantly and unkindly to restaurants: the suspicion that the plate had quietly gotten smaller.

That is the environment in which an independent operator has to make the decision this case is about.

The operating issue

Here is the composite. A 120-seat independent full-service restaurant, dinner six nights, with \$1,400,000 of annual food sales and a food cost held, honestly and with weekly counts, at 30.0% — \$420,000.

Proteins account for about 55% of food cost (\$231,000). Over four months, the operator's protein basket rises 14%.

🧮 The damage

Protein cost rises \$231,000 × 0.14 = **\$32,340**.

New food cost: \$420,000 + \$32,340 = \$452,340**, which on unchanged sales of \$1,400,000 is 32.3% — a 2.3-point** deterioration, worth \$32,340 a year straight off the bottom line.

For scale: on a restaurant of this size, 2.3 points of food cost is roughly the entire annual operating profit of a business running a 4% margin.

The operator has three real options and takes advice from three people who each recommend a different one.

Option A — price it

Restore the 30% by raising menu prices. To carry \$452,340 of food cost at 30%, food sales must reach \$452,340 ÷ 0.30 = **\$1,507,800 — a 7.7% price increase**, assuming covers hold.

Nobody believes covers hold. So the real question is: how many covers can you lose and still be better off than doing nothing?

• Do nothing: \$1,400,000 − \$452,340 = \$947,660 of gross food contribution.
• Raise prices, covers hold: \$1,507,800 − \$452,340 = \$1,055,460.
• Raise prices, lose 5% of covers: \$1,432,410 − \$429,723 = \$1,002,687.

Setting the raised-price case equal to the do-nothing case gives \$947,660 ÷ \$1,055,460 = 0.898 — the price increase is the better decision unless it costs more than about 10.2% of covers.

That number is worth pausing on, because it is far more tolerant than most operators assume. The fear of a price increase is usually larger than the arithmetic justifies.

Option B — shrink it

The chef's proposal: hold every price, and reduce protein portions by about 12%. A ten-ounce chop becomes just under nine. Nobody announces anything.

The arithmetic is almost perfect. Protein cost becomes \$231,000 × 1.14 × 0.88 = **\$231,739 — within \$739 of where it started. Total food cost becomes \$420,739, or 30.1%** of unchanged sales. The problem is solved on paper in a single afternoon, with no menu reprint, no guest conversation, and no risk of a competitor undercutting the new prices.

This is why the option is dangerous. It is not stupid. It works.

What it costs does not appear in this arithmetic at all. A guest who has ordered the chop six times notices the seventh. What they conclude is not "protein costs have risen 14%." What they conclude is that the restaurant is getting away with something — and, per Chapter 1's §1.5 and Chapter 23's subject, the second visit is where the entire profitability of a restaurant lives. A 12% portion reduction that costs 3% of repeat visits has bought \$32,340 of margin and sold something worth considerably more, on a timescale no report will connect back to the decision.

Option C — re-engineer around it

Change the specs and the menu rather than the prices or the portions. Move two dishes to cheaper cuts cooked longer. Deepen cross-utilization so one delivery covers more dishes. Promote the two items whose contribution margin did not move. Take one high-protein item off the menu entirely and replace it with a vegetable-forward dish at a similar price.

This is the most work, the slowest, and the one that requires the kitchen to actually be capable of something new. It is also the only option that changes the underlying structure rather than redistributing the pain. Its risk is that it takes months, and the margin is bleeding now.

What it shows

The three options are not equivalent, and the industry systematically over-fears the right one. Option A survives a cover loss of ten percent. Most operators behave as though it survives two.

The most tempting option is the one whose costs are invisible to the accounting system. This is a general and slightly frightening principle. Shrinking the portion produces a clean number in the P&L and puts its entire cost in a place no P&L records: the guest's estimate of whether this restaurant is generous. That asymmetry is exactly why it gets chosen, and exactly why it should be resisted.

The public evidence points the same way. The chains that came out of the 2021–2024 price surge best were generally not the ones who held nominal prices while trimming what was on the plate. They were the ones who made an explicit value proposition — a named offer, a stated price, a thing the guest could evaluate. Guests forgave a price. They did not forgive being quietly given less.

And the limit of this whole chapter is visible here too. Every option above was analyzed with cost cards, and the cost cards were necessary. They were not sufficient. Nothing in a cost card tells you what a 12% smaller chop does to a guest's willingness to return, and no amount of costing discipline substitutes for the judgment of somebody who has stood in the dining room.

Outcome

In the composite, the operator takes a fourth path, which is the one experienced operators usually take: a bit of A, a bit of C, and none of B. A 5% price increase concentrated on the items whose costs actually moved rather than spread across the menu; two dishes re-spec'd and announced as menu changes; one item removed. Portions on the remaining items are held, and the servers are told exactly what changed and why so they can answer a guest who asks.

It recovers about two-thirds of the \$32,340 within a quarter. The remaining third is absorbed, and the plan's food-cost line is restated at 30.8% for the year, honestly, in the operator's own numbers.

That last move — restating the line rather than pretending — is the tell of a well-run business.

Lesson

When costs move, choose the response whose costs you can see.

A price increase is legible: guests see it, competitors see it, and you can measure what it did to covers within a month. A portion reduction is illegible: you will never be able to attribute the guests who stopped coming to the ounce you took off the plate.

Given two options with similar arithmetic, take the one you will be able to evaluate. In a business with a four-point margin and no contracts, the decisions that hurt you most are almost never the ones you got wrong. They are the ones you could never tell whether you got wrong.

Discussion questions

1. The composite finds that a 7.7% price increase beats doing nothing unless it costs more than about 10% of covers. Recompute this for a restaurant with a 25% food cost and again for one at 35%. Does the tolerance widen or narrow, and why?

2. Option B produces a nearly perfect food cost percentage. Construct the argument that Option B is actually the right choice in some restaurants, and specify precisely what would have to be true of the concept and the guest for that argument to hold.

3. The case claims that guests forgive a price increase but not a secret portion reduction. What evidence would you want before believing that? Design a way a single restaurant could actually test it.

4. Restating the food-cost line at 30.8% rather than defending the 30.0% is described as "the tell of a well-run business." Argue the opposite: that a plan target should be defended, not revised, and that revising it teaches an organization the wrong habit. Where do you land?

5. Every analysis in this case used cost cards, and the case still concludes that cost cards were not sufficient. Name three other decisions in this book where the numbers are necessary and insufficient in exactly the same way.

6. Concentrating a price increase on the items whose costs moved is more precise than spreading it across the menu — but it also makes the affected items look worse next to their neighbours. Under what circumstances would you spread it instead, and what does Chapter 10's material about the price ladder say about the choice?