Menu engineering is the systematic analysis of your menu along two dimensions: popularity and contribution margin. Together they create four quadrants: Stars (popular and profitable), Plowhorses (popular, low margin), Puzzles (profitable, rarely ordered) and Dogs (weak on both). The goal: make Stars visible, re-cost Plowhorses, push or cut Puzzles, take Dogs off the menu. Before you move dishes between quadrants, it pays to look at the mistakes that distort your food cost and with it every contribution margin.
The quadrant analysis is the best-known part of menu engineering, but it is only the start. A menu can be costed perfectly on paper and still leave money on the table: through weak price architecture, wasted positions and no upselling structure. This article shows you both: the quadrants and the five checks a controller uses to take a menu apart.
Contents
- The four quadrants: popularity and contribution margin
- Check 1: price architecture and price thresholds
- Check 2: price anchors and signature positioning
- Check 3: upselling architecture
- Check 4: menu structure and choice overload
- Check 5: description and naming quality
- Placement effects: where the guest's eye stops
- In practice: what basket data reveals
- FAQ
- Menu engineering in a spreadsheet
- Your next steps
The four quadrants: popularity and contribution margin
The classic menu item sales analysis combines how popular a dish is with its contribution margin per portion (net selling price minus recipe cost). Both axes together show where you should look at price, recipe or visibility. Menu engineering adds the design and placement of the menu on top of this classification.
| High contribution margin | Low contribution margin | |
|---|---|---|
| Sells often | Stars: protect and show | Plowhorses: re-cost |
| Sells rarely | Puzzles: push or cut | Dogs: cut |
What you do in each quadrant:
- Stars: Leave them alone, except to highlight them. Best position on the menu, recommended by the floor team, never cut the price lightly.
- Plowhorses: Popular, but low margin. This is the biggest quiet lever: check the recipe (can food cost come down without the guest noticing?), raise the price moderately, unbundle the sides. A Plowhorse that sells 30 times a day turns 50 cents more contribution margin into more than €5,000 a year.
- Puzzles: Profitable, but invisible. Before you cut: a better position, a name that sparks curiosity, a description that sells, plus a recommendation from the team. If that changes nothing after two or three months, it wasn't a Puzzle but a Dog with good costing.
- Dogs: Off the menu. Every Dog costs storage space, prep, spoilage and, the underrated part, guest attention that a better dish misses out on.
The data comes from two sources: sales per dish from the POS and contribution margins from your costing (how to build that cleanly is in the article on restaurant menu pricing). Combining both is an afternoon's spreadsheet exercise, per location, per quarter. Or a report that falls straight out of your POS data.
So much for the quadrants. Now the part most menu engineering articles leave out: the menu itself as a sales tool. Five checks a good controller would run on their first day.
Check 1: price architecture and price thresholds
What to look for: Are the price endings consistent or have they grown historically? A menu with €12.90, €13.50 and €14.00 side by side tells the story of many small one-off decisions, but not of a pricing strategy. Second: is there enough price spread within each category? If every pasta dish sits between €12.90 and €14.90, you give the guest no way to trade up, and yourself no room for high-margin premium options. Third: do the prices sit at live price thresholds, or are you clinging to limits that inflation moved long ago?
Example: A main course category where the most expensive dish costs €16.90. It has done for three years, because "€17 feels like too much". The most expensive dish caps the perceived price range of the whole category: it tells the guest what the house is worth at most, and so also lowers acceptance of the prices below it. Compression at the top costs you twice: on the top dish itself and on everything priced beneath it.
Check 2: price anchors and signature positioning
What to look for: Does the menu have a price anchor, a deliberately high-priced dish that opens up willingness to pay and makes the rest of the menu look reasonable? And: is the most popular dish in the house treated as a signature dish, with the best position, a confident price and a visible highlight? Or is the secret star sitting quietly on line seven at the average price?
Example: A trattoria has a dish that gets praised remarkably often in Google reviews, say the vitello tonnato. On the menu it sits unmarked in the middle of the starters, priced exactly at the category average. That is a lever wasted twice over: a dish with proven popularity can carry an above-average price without losing orders. And as a highlighted signature it gives undecided guests a safe choice, which speeds up decisions and raises satisfaction. The anchor works even if it is rarely ordered: the bistecca at €34 doesn't need to sell often; it makes the ossobuco at €24.90 the sensible middle.
Check 3: upselling architecture
What to look for: Are add-on sales built into the structure of the menu, or do they depend entirely on how the floor team feels that day? Structural means there are visible bridges that grow the check. An aperitivo block at the front of the menu. Sides and topping upgrades right next to the mains ("+ burrata €4.50"). Starter platters to share for the table. A dessert and coffee combination. Size options where they add margin.
Example: A menu with twelve mains and not a single add-on. Every burger place shows how it's done: extra toppings, sizes, sides. The check grows in steps of €1 to €4 without the guest feeling sales pressure, because they make every decision themselves. The difference between a menu with and without upselling structure doesn't show in individual checks, but in the average check across thousands of orders. The practical section below shows how big the effect can be.
Check 4: menu structure and choice overload
What to look for: How many dishes does the menu have in total, and how many per category? Beyond a certain range, the offer tips into choice overload: the guest doesn't feel courted but overwhelmed, takes longer to decide and ends up reaching for the familiar and cheap. As a rule of thumb from practice: more than 7–9 dishes per category needs a very good reason. Second question: are the categories cut from the guest's point of view or the kitchen's? Guests think in "what do I want to eat today", not in product groups.
Example: A pasta category with 16 items. For the operator that's variety, for the guest it's work, and for the kitchen it's a prep problem that drives up spoilage and food cost. The quadrant analysis gives you the cut list straight away: if you trim 16 pasta dishes to 9 and remove the Dogs in the process, experience shows you lose hardly any sales. Orders move to the remaining dishes, the kitchen gets faster, purchasing gets simpler. Cutting is the only measure in menu engineering that improves sales quality, food cost and process cost at the same time.
Check 5: description and naming quality
What to look for: Does the text sell, or does it just list ingredients? "Tagliatelle, beef ragù, Parmesan" is an inventory list. "Homemade tagliatelle with grass-fed beef ragù braised for eight hours" is a sales argument. Second: are existing quality signals made visible? Many restaurants make their own pasta, buy locally, bake their own bread, and never write it down anywhere. What the guest doesn't read doesn't exist for their willingness to pay. Third: are foreign-language dish names without explanation a barrier? If you don't know what guanciale is, you won't order it, and you won't ask either.
Example: Two identical dishes, two texts. "Ossobuco with risotto, €24.90" versus "Ossobuco alla Milanese: veal shank braised overnight with saffron risotto and gremolata, €24.90". The second text justifies the price before the guest questions it. Description quality is the cheapest lever on the whole list: it costs nothing but an hour of writing and works on every printed menu, every day.
Placement effects: where the guest's eye stops
Guests don't read menus. They scan them. And that scan follows patterns you can use:
- First and last position beat the middle. Within each category, the first and last dishes get a disproportionate share of attention. The middle is where dishes get overlooked. Check your menu: what sits in the sweet spots, and has it earned it? A common finding: out of old habit, position one holds the cheapest dish in the category. That gives the best advertising space on your menu to the dish with the lowest contribution margin.
- Highlights work, used sparingly. A box, a "house recommendation", a subtle symbol measurably steers the choice. But the effect dies of inflation: if you mark six dishes, you mark none.
- Don't line prices up in a column. Right-aligned price columns with dotted leaders invite guests to compare prices from top to bottom, so they choose by price instead of appetite. Put the price straight after the description, without visual emphasis.
The practical link to the quadrants: your Stars and your Puzzles belong in the sweet spots. Stars, because that's where they have the most effect. Puzzles, because position is the cheapest test of whether they can become Stars.
In practice: what basket data reveals
Menu engineering doesn't stop at the single-dish analysis. The next level is baskets: what gets ordered together, and what conspicuously doesn't?
An example from the Mama Trattoria case study: the basket analysis showed which mains and desserts were often ordered together. That led to concrete hypotheses for service and the menu. We will only publish a result figure again with definition, baseline and measurement period.
This insight wasn't in any monthly P&L or anyone's gut feeling. It was in the POS data, where it already sits today in almost every restaurant, unanalysed. The difference between a menu that "does fine" and a menu that sells systematically is rarely creativity. It is the analysis.
From analysis to action
What this is about: portfolio analysis of the menu by popularity and contribution margin.
Put it into practice: Maintain recipe costs in Kiosa Inventory
More on Food cost and inventory
- Food Cost Percentage Formula: Calculation, Example, Calculator
- Menu Item Sales Analysis: Step by Step with a Worked Example
- Contribution Margin per Dish: Calculate CM1 and CM2
FAQ
Do I need POS data for menu engineering?
For the quadrant analysis, yes. Without sales per dish you can only guess popularity, and guesses are systematically wrong (dishes that get praised a lot aren't automatically the best sellers). The five checks (price architecture, anchors, upselling, structure, descriptions) you can work through right away, without any data. A sensible order: do the checks first and build the data in parallel.
How many dishes belong on a menu?
There's no magic number, but there is a solid rule of thumb: more than 7–9 dishes per category creates choice overload and dilutes purchasing and prep. A casual dining menu with 30–40 items in total (including starters and desserts) is almost always stronger than one with 70, for the guest, for the kitchen and for food cost.
How often should I analyse my menu?
The quadrant analysis once a quarter, which fits the rhythm of menu changes and seasonal produce. The five checks once thoroughly, then with every menu revision. Prices and contribution margins should also be re-costed at least every six months, because purchase prices move independently of the menu.
Should I cut Dogs straight away?
Almost always, with two exceptions. First, dishes with a strategic role: the vegan dish or the kids' dish often decides whether the whole group comes, even if it barely sells itself. Second, new dishes that haven't had a fair chance yet (less than two or three months on the menu). Everything else blocks storage, prep and your guests' attention.
What does a price anchor actually do?
An anchor doesn't change what guests order; it changes how expensive the rest of the menu feels to them. Next to a bistecca at €34, the ossobuco at €24.90 looks like the sensible middle; without the anchor it's the most expensive dish on the menu and gets chosen less often. The anchor has to be a genuine dish you can actually deliver. Guests and staff see through a pure alibi product alike.
Menu engineering in a spreadsheet: sales mix and contribution margin
A simple spreadsheet is enough to start: it combines quantities sold, net selling prices and recipe costs. Only compare dishes within a sensible group and the same period, for example a restaurant's mains in one month. Throwing drinks and mains together distorts the classification.
Contribution margin per portion is net selling price minus recipe cost. An item's total contribution is that amount times the quantity sold. A high food cost percentage on its own therefore doesn't make a dish unattractive. Make the thresholds you choose for popularity and contribution margin visible in the sheet; the quadrants help you review, they don't replace a sales test.
How to check the recommendation: update recipe prices and portion sizes, separate voids and returns cleanly, look at promotional items separately. After a change, watch quantity and total contribution margin together. A lower food cost percentage with clearly fewer sales can make the result worse.
The worked example in the menu pricing article explains the difference between a percentage and a euro contribution; the guide to restaurant inventory counts adds the reconciliation with actual usage.
Your next steps
Food cost calculator: the foundation of every quadrant analysis. Calculate contribution margins per dish with current purchase prices.
Menu price calculator: calculate net and gross price per dish from ingredient cost, target percentage or contribution margin.
Restaurant menu pricing: from food cost to selling price, with a fully worked example and the most common costing mistakes.
30 minutes well spent. You want to see sales mix, contribution margins and baskets automatically from your POS data, instead of building them in a spreadsheet every quarter? Book a demo call and I'll show you what that looks like with real data.
Related terms
- Menu engineering: systematic menu analysis by popularity and contribution margin
- Price threshold: psychological price limits and how to use them
- Signature dish: the flagship dish and how to position it
- Choice overload: why too much choice costs sales
- Case study: Mama Trattoria: 10 locations, one dashboard