Menu Engineering: Using the Menu Star to Fix Your Menu
The menu engineering matrix explained, how to classify your dishes, and what to do with stars, puzzles, plows and dogs.
Last reviewed .
Menu engineering classifies dishes by two numbers — popularity and profitability — and the combinations tell you what to do with each one. It is one of the oldest tools in hospitality and one of the least often applied properly, usually because the profitability measure is wrong.
The two axes
Popularity
How often the dish is ordered, as a percentage of all items ordered in the period. Use item mix, not revenue — the question is 'how many guests choose this'.
Profitability
Food cost as a percentage of the selling price. Lower is better. Many teams get this backwards by treating high margin in pounds as 'profitable' when it ignores how much of the dish is wasted.
Better still is contribution margin per plate: selling price minus ingredient cost minus the variable cost of serving it. That captures the plating, the labour step and the waste.
The four quadrants
| High popularity | Low popularity | |
|---|---|---|
| **High margin** | **STARS** — protect, feature, never let them run out | **PUZZLES** — reposition or promote; often a great dish priced or placed badly |
| **Low margin** | **PLOWHORSES** — rework the recipe; they sell, they just do not earn | |
| **DOGS** — remove, or reprice into viability |
What to do with each
Stars
Protect them. They carry the menu. Features: placement on the cover, server recommendation scripts, availability planning, and a rule that they never run out. Their cost is validated; do not cut quality on them to save a percentage point.
Puzzles
Usually an operational problem, not a product one. A genuinely good dish selling badly is typically the price, the description, the menu position, or a photograph. Reposition, rewrite the description, add an image, or change the server script. Do not remove a puzzle before you have tested positioning.
Plowhorses
Popular but barely profitable — the most dangerous quadrant, because volume disguises the problem. Re-spec the recipe: measure the portion, check the yield, substitute the expensive component, or reduce the garnish. Re-cost after thirty days.
Dogs
Low margin and low popularity. Removing a dog simplifies the kitchen, reduces stock, speeds service and improves the average contribution of everything else. Removing the bottom ten per cent of a menu typically improves overall profitability — this is counter-intuitive and reliably true.
Doing it properly
- Pull at least eight weeks of item-level sales data. A month is too short for weather and promotions to average out.
- Exclude voids, comps and staff meals from the mix calculation.
- Cost every dish accurately, including yield.
- Classify. Then sanity-check against intuition — the matrix is a prompt, not a verdict.
- Act on one quadrant at a time, and measure the effect.
The mistakes
- Popularity measured in revenue. A £60 dish sold 20 times is not popular; it is high-value. You will conclude your most popular items are your most expensive ones.
- Profitability as food cost, not margin. Understand the direction: lower food cost percentage is better. Reversing this inverts every conclusion.
- A four-week window. Seasonal variance produces permanent misclassification.
- Treating it as a one-off exercise. It is a quarterly cycle, because prices, mixes and preferences all move.
The practical value
The menu is your highest-leverage commercial asset, and most of it is unmanaged. Even a first pass usually finds that a quarter of dishes are doing nothing except slowing the kitchen down, and that the item your team is proudest of has been badly placed on the menu for years.