Menu engineering as a food loss and waste mitigation tool: where it breaks and what works instead

When the classic Kasavana-Smith matrix falls short?

The classic matrix falls short at the exact moment your kitchen preps mise en place in batches, because the Kasavana-Smith model crosses units sold against contribution margin and neither of those two variables contains a single discarded kilo.

The telltale figure is easy to find: a dish classified as a STAR, with 68% margin and high turnover, whose base is produced in batches where 30% ends up in the bin at closing. It sold well, it margined well, and the month's cash came out worse. That blind spot is not theoretical: UNEP counted more than 1 billion meals wasted per day worldwide in 2022, and UNFCCC puts the cost of food loss and waste near USD 1 trillion a year. If your matrix cannot see those kilos, you are optimizing half the problem with complete confidence.

Option 1 · Menu engineering calibrated with measured waste

The option that actually moves the food-waste indicator is a three-axis matrix: popularity, margin and waste weighed in kilos across fourteen days before touching the menu. It suits the owner of one to three venues with an in-house kitchen, mid-range ticket and batch production, who already runs recipe costings and is tired of guessing. The switching cost is low in money and high in discipline: a 30 kg scale with tare, two bins labeled by origin of the discard —prep and plate returns— and fifteen minutes of logging at closing. Fourteen days of baseline and you will know which of your stars is really a workhorse in disguise. The outlay runs around USD 120 in equipment and roughly 3.5 hours a week from a head chef. In return you get what no narrative redesign gives you: a starting figure against which to measure anything you do afterwards.

Option 2 · Plate costing with waste folded into net margin

If you would rather not build a new matrix, fold waste into the recipe costing and work with NET margin per dish, which is the poor but honest version of the calibrated method. Arithmetic rules here: a USD 12 dish at 70% margin that discards 20% of its production leaves USD 6.72 net per unit sold; a USD 9 dish at 58% with no discard leaves USD 5.22. The gap almost closes entirely, and at 25% discard the expensive dish drops below. It fits the owner who has an accountant or an admin assistant able to maintain recipe cards, and it costs practically nothing in equipment: two new columns in the cost sheet and a waste factor per recipe reviewed every quarter. Diego F. Parra insists at Masterestaurant that food cost per dish caps at 32% —a maximum, not a recommendation— and that rule only makes sense if the 32% already includes what got thrown out.

Option 3 · Waste-tracking software with a connected scale

For operations of four venues or more there is the technology route: connected scales that photograph and weigh every discard, sort it by category and feed a live dashboard. The profile is clear —a chain with an operations manager, centralized production and the ability to sustain an annual contract— and so is the switching cost: between USD 3,000 and USD 9,000 per year per site depending on vendor, plus two weeks of adoption with the kitchen team. ReFED reported that food surplus in the United States fell 2.2% in 2024, to roughly 70 million tons, and a good share of that reduction comes from operators who stopped estimating and started weighing. The risk here is cultural rather than technical: if the cook reads the scale as surveillance, the data goes dirty within three weeks. Buy the scale only after sustaining manual logging for a full quarter.

Option 4 · Menu redesign by cutting references

Trimming the menu is the cheapest option and the most underrated, and it works because every reference you remove takes its own inventory, its own mise en place and its own waste with it. A 48-dish menu brought down to 28 usually frees between six and nine exclusive inputs, the ones that go into a single low-selling dish and get bought anyway at the supplier's minimum unit. The profile that gains most is the independent restaurant with a menu inflated by historical accumulation, where nobody has removed anything in four years. It costs zero in equipment and plenty in uncomfortable conversations with the chef. The warning is serious: cutting without per-reference sales data is amputating blind, and I have fixed menus where the dish removed was the one bringing in the table of six. Measure ninety days of item-level sales first, then cut.

Sequence matters more than the sophistication of the method

The decisive difference between these paths lies not in sophistication but in ORDER. The classic method redesigns first and measures later, if it measures at all; the calibrated one measures fourteen days and redesigns with evidence in hand. That inversion of sequence explains why two restaurants with the same menu and the same consultant end up with food costs four points apart, which on monthly revenue of USD 60,000 amounts to USD 2,400 a month, nearly a cook's salary. In public food-waste mitigation programs the consequence gets worse still: with no baseline in kilos, the operator can only report activities —workshops delivered, menus redesigned, restaurants visited—, never tons avoided. Monitoring turns narrative. When an evaluator asks about impact and the answer is a list of workshops, the instrument loses credibility and the next budget goes to another sector.

Why this is development policy and not just kitchen management?

In Latin America and the Caribbean, food service concentrates a high share of waste at the consumption stage, which makes the matter considerably bigger than an argument about recipe costings.

SATE Institute works this front with Masterestaurant S.A.S. as technology partner, and the logic is local economic development: if 8% of a restaurant's revenue goes into the bin, that business does not capitalize, does not formalize payroll and does not qualify for credit, and the municipality loses its tax base. Regional context sharpens the bill —Acodrés reported more than 2,000 restaurants closed in one year in Colombia, and the ILO measured female informal employment growing 22.8% in 2024 against 15.7% for men—. A program that finances the wrong tool burns public budget and discredits the instrument for three years.

When NOT to switch methods: the honest answer?

Stay with the classic two-axis matrix if your kitchen cooks to order, with no batch production, protein portioned by the supplier and sides plated one by one:

prep waste there is marginal and building a weighing system will cost you more in hours than it returns in kilos. Stay as well if your food cost has run below 30% for six months and your inventory closes with variance under 2%, because your problem is not waste, it is something else, most likely labor cost, which the Bureau of Labor Statistics places between 25% and 35% of revenue and which tends to be the real leak once the kitchen is already in order. And never switch during peak season or a change of chef. Start on a Monday in February, with the scale on the prep table and a notebook, and decide in March with fourteen days of your own numbers.

The four differences that decide the outcome

The biggest difference is sequence, not sophistication: the classic method redesigns first and measures later, if it measures at all; the calibrated one measures fourteen days and redesigns on evidence. That inversion explains why two restaurants with the same menu and the same consultant end up four points apart on food cost. The classic method optimizes unit margin; the calibrated one optimizes margin NET of waste. A USD 12 dish at 70% margin that discards 20% of its production yields less than a USD 9 dish at 58% with no discard, and the two-axis matrix will recommend pushing the first one with its logic fully intact.

The four differences that decide the outcome — in practice

In monitoring and evaluation terms, one is narrative and the other metric. Without a kilogram baseline, a food loss and waste mitigation program can only report activities, workshops delivered and menus redesigned, which is precisely the kind of indicator multilateral development banking stopped accepting as evidence of impact. And there is a risk difference: waste sustained above 6% of purchases is an early predictor of cash strain. The restaurant still pays payroll and still invoices, yet its working capital erodes quietly. A credit analyst holding that series sees the deterioration a quarter earlier than through the income statement.