Menu engineering is the practice of analysing every item on your menu by two numbers — how often it sells (popularity) and how much money it actually leaves behind after ingredient cost (contribution margin) — and then redesigning the menu around what those numbers tell you. Items land in one of four quadrants: Stars (popular and profitable), Plowhorses (popular but low-margin), Puzzles (profitable but slow-selling), and Dogs (neither). Each quadrant has a different playbook, and working through them is one of the highest-leverage exercises a restaurant or café can do — it changes profit without changing rent, headcount, or covers.
This guide walks through the full method: costing your recipes properly (recipe engineering), calculating contribution margins, classifying every item, and the specific actions to take in each quadrant.
The menu engineering matrix
The classic framework comes from Michigan State University researchers Michael Kasavana and Donald Smith in the early 1980s, and it has survived four decades because it is simple and it works. Plot every menu item on two axes:
High popularity | Low popularity | |
|---|---|---|
High contribution margin | ⭐ Star — protect and promote | 🧩 Puzzle — reposition or re-price |
Low contribution margin | 🐴 Plowhorse — raise margin carefully | 🐶 Dog — fix, hide, or cut |
Two things matter before you trust the quadrants:
Contribution margin, not food-cost percentage. A 20% food-cost item that contributes $2 per sale is worse for your bank account than a 40% food-cost item that contributes $6. You deposit dirhams, dollars, and rufiyaa — not percentages.
Real recipe costs, not guesses. If your ingredient costs are estimates from six months ago, the whole matrix is fiction. That's where recipe engineering comes in.
Recipe engineering: get the cost side right first
Recipe engineering is the process of documenting every menu item as a precise recipe — ingredients, quantities, yields, and preparation losses — so that each dish has an accurate, current cost per plate. It is the foundation menu engineering stands on: popularity comes from your sales data automatically, but cost only comes from disciplined recipes.
A usable engineered recipe covers:
Every ingredient with a real quantity. Not "some onion" — 80 grams of onion, including the garnish, the oil, and the sauce portion.
Yield and trim loss. A kilogram of whole chicken does not produce a kilogram of usable meat. If a recipe uses 150 g of cleaned fish, the cost must reflect the whole-fish price divided by your actual yield.
Current ingredient prices. Supplier prices move — especially where supply runs come by boat and a rough week changes the market. Recipe costs must move with them, which in practice means linking recipes to your live inventory costs rather than a spreadsheet you update "when there's time".
Batch recipes and sub-recipes. The curry base you make every morning is itself a recipe; the dishes that use it should inherit its cost per portion.
Once every item has a live plate cost, contribution margin is simple:
Contribution margin = selling price − plate cost
Do this for every item on the menu. No exceptions — the items you skip are usually the ones hiding the problem.
Classify your menu: the worked example
Popularity is measured against an even share. The common Kasavana–Smith rule: an item is "popular" if it sells at 70% of the average share or better. With 10 items on a menu, the average share is 10%, so anything above 7% of total items sold counts as popular.
Here's a worked example for a small café that sold 1,000 items last month across five dishes (numbers invented for illustration):
Item | Sold | Mix % | Price | Plate cost | Margin/plate | Quadrant |
|---|---|---|---|---|---|---|
Grilled chicken plate | 310 | 31% | $14.00 | $4.90 | $9.10 | ⭐ Star |
Fried rice | 290 | 29% | $9.00 | $4.10 | $4.90 | 🐴 Plowhorse |
Beef ribs | 90 | 9% | $22.00 | $8.60 | $13.40 | 🧩 Puzzle |
Club sandwich | 220 | 22% | $10.00 | $3.20 | $6.80 | ⭐ Star |
Veg lasagna | 90 | 9% | $12.00 | $6.90 | $5.10 | 🐶 Dog |
With five items the popularity threshold is 70% × 20% = 14%. Average margin per item sold here is about $7.10 — that's the margin line. The ribs clear the margin bar but miss the popularity bar (Puzzle); the lasagna misses both (Dog); the fried rice sells constantly but leaves the least behind per plate (Plowhorse).
What to do in each quadrant
Stars: protect and showcase
Stars pay your rent. The job is to keep them exactly as they are — same portion, same quality, same consistency — and make them impossible to miss.
Give them prime menu real estate: top-right of a physical menu, first cards with photos on a QR ordering menu.
Train servers and cashiers to lead with them.
Watch their ingredient costs like a hawk; a quiet supplier price rise can turn a Star into a Plowhorse without the menu changing at all.
Resist the urge to "improve" them. Consistency is the feature.
Plowhorses: engineer the margin up
People love these items, so you have permission to work on the economics — carefully.
Re-cost the recipe first. Plowhorses often carry invisible waste: over-portioned protein, garnish nobody eats, a side that costs more than it reads.
Nudge the price. Popular items tolerate small increases far better than operators fear, especially if the increase lands with a small presentation upgrade.
Re-plate the portion mix — slightly less of the expensive component, slightly more of the cheap one (rice, fries, salad), keeping perceived value intact.
Pair them. A combo that attaches a high-margin drink or side to a Plowhorse lifts the ticket margin even if the item margin stays modest.
Puzzles: sell the profit you already have
Puzzles make great money whenever someone orders them — the problem is that almost nobody does.
Reposition them on the menu: better placement, a photo, a one-line description that sells the dish instead of listing components.
Rename if needed. "Beef ribs" undersells what "12-hour slow-braised ribs" says for free.
Check the price barrier. Sometimes a Puzzle is just $2 above what the room will pay; a small cut that keeps a strong margin can multiply volume.
Let staff push them — a featured-dish mention at the table or counter costs nothing.
If a Puzzle stays a Puzzle for months despite real effort, accept the verdict: your guests have voted.
Dogs: fix fast or cut without sentiment
First ask whether the recipe can be re-engineered into profitability — cheaper components with equal perceived value, a smaller portion at a reset price.
If not, remove it. Every Dog occupies menu space, prep-list time, and inventory that a Star could use. Kitchens get simpler, waste falls, and stockholding shrinks.
Exception: a deliberate menu role — the kids' meal, the plain option for the non-adventurous guest in a group. Keep those, but keep them knowingly and keep them few.
Run it on a cadence, not once
Menu engineering is not a one-off project. Prices drift, seasons change what sells, and every menu change resets part of the picture. A workable rhythm:
Monthly: refresh the matrix from the latest sales mix and current recipe costs; act on anything that switched quadrant.
Quarterly: do the full pass — re-check portion specs and yields, retire persistent Dogs, decide the next Puzzle experiment.
After any supplier price shock: re-cost affected recipes immediately. This is the moment margins silently rot.
The most common failure mode is not bad analysis — it's stale inputs. The operators who win at this are the ones whose recipe costs update themselves.
Automating the numbers
Everything above can be done with a spreadsheet, a sales report, and patience — and for a small menu it's worth doing once by hand just to learn your own numbers. But the two inputs that make the matrix trustworthy, live recipe costs and per-item sales mix, are exactly what a well-integrated POS already knows.
Qvian connects the loop end to end: recipes are costed from your actual ingredient purchase prices (updated as goods are received, weighted-average, not a static list), every sale updates the item mix in real time, and Restaurant Intelligence surfaces per-item margin and popularity continuously — so the matrix is a living view rather than a quarterly chore. The same recipes drive kitchen stock deduction, which means the cost side and the inventory side can't drift apart. See how the recipe system works, or explore the full restaurant platform and POS.
If you'd rather spend your quarterly menu review deciding what to do with your Stars and Puzzles instead of rebuilding the spreadsheet that finds them — book a demo and bring your menu.
