Menu engineering sounds like consultant jargon, but it comes down to one operation: split the menu into four groups by sales and margin, then do something different with each group.
The whole job is an hour with a sales report and a spreadsheet. Below is the spreadsheet and the decisions you have to make.
Menu engineering sheet — XLSX — columns for price, ingredient cost and sales, with food cost, margin and monthly profit calculated for you. Three worked examples at the top.
Two numbers per item
You only need two figures for each dish on the menu:
- how many units sell per month — from the POS or from the dockets,
- what the ingredients cost per portion — from the recipe, net.
The menu price is already there. From that, the sheet works out food cost, margin per unit and monthly profit.
You do not have to do the whole menu at once. Start with the 20 items that make the most turnover — they decide the outcome anyway.
The four classes and what to do with each
Compare every item against the median sales and median margin of your own menu. Not the average — one dish selling 500 times will distort it.
Stars: sell a lot, high margin
They stay and go somewhere visible. Do not raise the price casually: these are the dishes people come for, and adding €4 can cost more in sales than it gains in margin.
Decision: protect. Check they do not quietly disappear during a seasonal change.
Workhorses: sell a lot, low margin
The most common menu trap. The kitchen works, turnover grows, profit stands still.
Three ways out, in order of risk: cut the ingredient cost (different supplier, different garnish weight), raise the price by €2–3 and watch the reaction for a month, or rework the recipe so the dish stays recognisable but becomes cheaper to produce.
Decision: fix the margin, do not remove.
Puzzles: sell little, high margin
This is where the largest unused profit sits. The dish earns, but nobody orders it — usually because the guest does not notice it or does not understand it.
Order of action: rewrite the description, move it somewhere visible, add a photo, ask the floor staff to recommend it for two weeks. Only if it still does not move after a month — remove it.
Decision: give it a month, then cut.
Dogs: sell little, low margin
Off the menu. Every such item is stock in the store, space on the page and one more decision the guest has to make.
The one exception: a dish that holds a specific group — vegan on a meat menu, gluten-free, something for children. Then it stays despite the numbers, but knowingly.
Decision: remove at the next menu change.
Where on the page the guest actually looks
Order within a category has a measurable effect. Attention concentrates at the start and the end of the list — the middle disappears. In a category of 10 items, the first 3 and the last 2 are what really get read.
Practical conclusions:
- put the star and the best puzzle first and last in the category;
- put any remaining dogs in the middle;
- do not sort a category by price ascending — the guest stops at the first item, and that is usually the cheapest one.
The same applies to the wine list, where the second-from-bottom effect works on top of this.
What the spreadsheet will not show
Three things that are not in the numbers but change the decision:
The reason dish. An item somebody drives across town for can have a weak margin and average sales — and removing it costs you guests, not a menu line.
Season. Cold soup has terrible numbers in November. Count only within the window when the dish was actually on the menu.
Labour cost. The sheet counts ingredients. A dish with 40 minutes of prep and one with 5 have different real profitability at identical food cost.
Fitting this into the rhythm of the restaurant
Menu engineering is not a one-off project. A sensible cycle:
- Once a quarter — sales report, spreadsheet, four classes.
- At every menu change — decisions from the sheet instead of intuition. We described the process in seasonal menu updates.
- After every price change — check a month later whether sales fell further than the margin rose.
Point three requires price changes to be cheap. With a printed menu a test costs another print run, so nobody runs one.
On a digital menu the price changes in the panel, so the test is free: raise it by €3, watch for a month, revert or keep. See QR menu or menu examples.
Frequently asked questions
How many items do I need to analyse for this to be worth it?
The 20 best-selling dishes are enough — they make most of the turnover. A full 80-item menu gives a sharper picture, but the decisions come out the same.
What food cost is good?
It depends on the format: cafés 15–30%, fast food and street food 25–40%, bistro 30–45%, fine dining 35–50%. Consistency across items matters more than the number itself.
Should I remove a dish with a good margin that nobody orders?
Not straight away. Description, position on the page and a photo first — that usually does it. Remove only after a month with no reaction.
How often should I run this analysis?
Once a quarter and at every menu change. More often makes no sense — the data needs time to accumulate.
