Somewhere on your menu, right now, sits a dish that costs you money every time it is ordered. It may well be the one you're proudest of. Guests photograph it, reviews mention it — and tonight it will quietly drain margin from your kitchen again. Which dish? If you can't answer within ten seconds, your menu is making decisions without you.
A menu looks like a list of dishes. In reality it is a one-page shop where the average guest spends 109 seconds — and where tiny choices move thousands of euros a year: where a dish sits on the page, whether the price wears a € sign, what the description whispers. This guide walks through all of it in six chapters, from the four-box matrix that X-rays your card to the seasonal re-tune that keeps it sharp. By the end you'll know exactly what every plate earns — and why your bestseller might be your worst dish.
The short version
- Classify every dish quarterly into Stars, Plowhorses, Puzzles and Dogs by margin × popularity — then act on each quadrant differently.
- Cost recipes to the gram: a 28–32% food cost is fine-dining health; what you don't measure drifts to 38%.
- Price the experience, not the ingredients — remove € signs, anchor with one premium item, and let prix-fixe structure do the heavy lifting.
- Drinks are half your margin engine: wine by the glass, pairings and a tight cocktail list often out-earn the kitchen per minute of labour.
- Descriptions sell — origin, technique and taste words lift a dish's orders measurably; sentimental clutter buries your Stars.
-
The matrix
Read your menu like an engineer: Stars, Plowhorses, Puzzles, Dogs
Menu engineering plots every dish on two axes — contribution margin and popularity — yielding four quadrants: Stars (high/high), Plowhorses (popular, low margin), Puzzles (profitable, unpopular) and Dogs (neither). One quarterly cycle of classify-then-act typically lifts overall menu margin by 10–15%.
You cannot manage a menu you have never measured. The method, refined since the 1980s and still unbeaten, takes one quiet afternoon per quarter: export your sales mix, cost every dish honestly (chapter 2), and place each on the matrix.
The four quadrants — and the only correct move for each Quadrant Profile The move Stars High margin, high popularity Protect: never change them casually, give them prime menu real estate, train staff to mention them first Plowhorses Popular but thin margin Re-engineer: trim portion of the costly element, renegotiate the key ingredient, or nudge price €1–2 Puzzles Profitable but ignored Promote: better description, better placement, a server story — or rename the dish entirely Dogs Low margin, low orders Remove without sentiment — every Dog steals attention from a Star The complete walkthrough with the costing spreadsheet logic is in menu engineering.
The discipline that makes it work
Two failure modes kill most attempts. First, costing by feel — "the duck is probably around 30%" — which chapter 2 cures. Second, sentiment: the chef's favourite Dog that survives every cull because of the story attached to it. The matrix is not a critique of cooking; it is a seating chart for dishes. Even a three-Michelin-star menu has Plowhorses — the craft is knowing which ones earn their place.
Do this tonightExport last month's sales by dish. Without costing anything yet, mark your gut-feel Stars and Dogs. Keep the list — after chapter 2 you'll cost them properly and discover which instincts were expensive.
Why your most expensive dish should never be your best seller
A luxury flagship — the caviar course, the whole turbot for two — earns most of its money without being ordered. It anchors the price ceiling: next to a €120 plateau, the €68 tasting menu reads as reasonable. Engineers call it the decoy effect. If the flagship ever becomes your best seller, your ceiling is too low — raise it, and watch the dishes beneath it sell easier.
-
Cost control
Control the cost under every plate — to the gram
Food cost control means a recipe card with gram-level costing for every dish, weekly tracking of the overall percentage, and systematic waste reduction. Fine dining is healthy at 28–32% food cost; unmeasured menus drift toward 38% through portion creep, price inflation and bin losses.
Margin is not made at the till; it is made on the cutting board. Three silent leaks take it away: portion creep (the 160g portion that became 180g because nobody weighs anymore), supplier drift (the cream that rose 14% across six invoices nobody compared), and the bin (trim, overproduction, the special that didn't sell).
Recipe cards are the contract
Every dish gets a card: ingredients in grams, current supplier price, yield after trim, target plate cost. This isn't bureaucracy — it is the only way the menu you engineered in chapter 1 stays engineered. When the card says the venison plate costs €9.40 and the matrix says it's a Star at €34, you have a fact. Without the card you have a mood. The full method is in controlling food costs.
Waste is margin in the bin
Studies put restaurant food waste at 4–10% of food purchased — at fine-dining ingredient prices, the top of that range is a full point of net margin. The countermeasures compound: trim-to-garnish thinking (the fennel fronds are the plate's garnish, not the bin's), prep-to-par sheets driven by reservation forecasts, and a weekly five-minute bin audit. Deep tactics live in reducing food waste.
- Weigh the five costliest proteins at plating for one week each quarter — portion creep dies in daylight.
- Re-quote your top ten ingredients twice a year; loyalty without comparison is charity to your supplier.
- Cost the staff meal — it is part of food cost, and the honest number changes behaviour.
Do this tonightTake your single best-selling dish and cost it to the gram against this week's invoices. If the true percentage surprises you by more than two points, you've found the first leak — and probably not the biggest one.
-
Pricing
Price the experience, not the ingredients
Menu pricing works when it manages perception: remove currency symbols, avoid price columns that invite scanning, anchor with one premium item, and use prix-fixe or tasting structures to shift the decision from price-per-dish to value-per-evening. Cost-plus pricing alone systematically underprices fine dining.
Cost × 3 is arithmetic, not strategy. Guests do not experience your costs; they experience an evening — and they judge its price against anchors you control. The craft is to set those anchors deliberately.
Five levers, all tested
- Drop the € sign. Research on menu pricing found guests spend significantly more when currency symbols are absent — "34" reads as a number; "€34,00" reads as a payment.
- Break the price column. Prices aligned in a tidy right-hand column invite top-to-bottom price shopping. Nest each price quietly at the end of the description instead.
- Anchor high. One genuinely premium item per section resets what "expensive" means (see chapter 1's secret).
- Charm endings have a class signal: 9-endings (€19,90) whisper discount; whole numbers (34) read as confidence. Fine dining prices in whole numbers.
- Mind the sweet spread: keep mains within a roughly 1.6× band — a €26-to-€68 spread makes guests price-anxious; €28-to-€44 keeps the choice about appetite.
Structure beats sticker
The strongest pricing move in fine dining isn't a number at all — it is structure. A prix-fixe alongside à la carte moves the decision from thirty prices to one, lifts average ticket predictably, and calms the kitchen. At the summit, the tasting menu turns pricing into storytelling: one number, one journey, margins engineered course by course where no guest can unbundle them.
Do this tonightPrint your current menu and circle every € sign and every price sitting in a neat right-hand column. That's tomorrow's design fix — zero cost, measurable lift, and nobody will consciously notice what changed.
-
Liquid margin
Build a drinks program that earns like a second kitchen
Drinks routinely deliver 20–30% of fine-dining revenue at margins the kitchen can't match. The engine has four parts: a wine list built around by-the-glass rotation, pairings priced as a product, a tight signature cocktail list, and aperitif/digestif moments staged into service.
No section of the menu converts attention into margin faster than the liquid one. A €14 glass of wine takes ninety seconds of labour; a €14 plate takes a brigade. Yet most rooms treat the wine list as a library and the aperitif as an accident. Treat both as products.
The wine list is for selling, not collecting
A list of 400 bins impresses sommeliers and intimidates guests into the second-cheapest bottle. A working list — profiled in wine list & beverage management — is tighter: every bottle has a job, by-the-glass selections rotate with the menu (and with what needs selling), and storage costs are counted as the working capital they are. Train the floor with confident wine advice: the question "shall I pick something for the table?" is the single highest-margin sentence in service.
Stage the liquid moments
Four moments, four products Moment Product Why it works Arrival House aperitif, offered by name "A glass of our cuvée while you read?" converts ~60% when offered, ~0% when not Ordering Pairing as default option "With or without the pairing?" reframes it from extra to choice Main course The second-glass check at the right second Refill timing, not pressure — see upselling that feels like service Dessert Digestif trolley or cheese course Theatre sells what menus can't A short signature cocktail list — five drinks that belong to your concept — completes the engine: cocktail margins beat wine, and a named house serve becomes marketing guests photograph.
Do this tonightCount tonight's covers, then count how many were offered — not asked for, offered — an aperitif by name. The gap between those numbers, times €11, times 300 nights, is the size of the product you haven't launched yet.
The pairing trick that doubles uptake
Houses with the highest pairing attachment never sell it as an add-on. The menu prints the tasting menu price with pairing first, the without-price second — and the server asks "with or without the wine journey?" Choice architecture does the rest: uptake roughly doubles versus "would you like to add wines?", with zero pressure and a happier table. The pairing also lets the sommelier move stock by the glass that would never sell by the bottle.
-
The card itself
Write and design a card that sells the right dishes
Guests read a menu for under two minutes, scanning predictable zones. Winning cards use that scan: Stars in the first and last positions of each section, descriptions built from origin + technique + taste, no clutter or sentimental essays, and allergen clarity that signals craft instead of compliance.
Everything engineered so far is delivered through one artefact: the card in the guest's hands. Two minutes of attention, mostly spent on the first and last items of each section and anything visually distinct. Spend those seconds deliberately.
Placement is silent salesmanship
- First and last in each section get disproportionate orders — seat your Stars there, never your Dogs.
- One box, frame or chef's mark per page maximum: highlight everything and you highlight nothing.
- Seven items per section is the comfortable ceiling; beyond it, choice anxiety pushes guests to the familiar — usually a Plowhorse.
Descriptions are the second salesperson
The formula that lifts orders, dish after dish: origin + technique + taste. "North Sea turbot, beurre blanc grilled, charred lemon" out-sells "Turbot with lemon" because every word earns its place. Research on descriptive labels shows they lift a dish's sales by up to 27% and improve post-meal taste ratings. Skip the adjective inflation — "delicious", "famous", "mouthwatering" sell nothing. The full craft is in writing menu descriptions.
Allergens: clarity is class
Fine dining handles allergens in the dialogue — "we ask every table about allergies" — backed by a kitchen matrix per dish. A card cluttered with fourteen icons reads like a pharmacy; a confident note ("allergies? tell us — almost everything can adapt") reads like craft. Both the law and the guest are better served by the second.
Do this tonightRewrite your three Puzzles (profitable, ignored) with the origin + technique + taste formula, and move one to the top of its section. Track their orders for two weeks — this is the cheapest A/B test in hospitality.
-
Rhythm
Make the seasons re-tune the machine for you
A seasonal menu cycle is the maintenance schedule of menu engineering: four times a year, ingredients get cheaper and better simultaneously, Dogs exit gracefully, Puzzles get re-staged, and the card stays newsworthy. Each rotation is also a costed re-run of every chapter in this guide.
Seasonality is the only force in hospitality that improves quality and margin at the same time: asparagus in season costs less and tastes better than asparagus out of season, with marketing built in. A seasonal menu is therefore not a creative indulgence — it is the maintenance cycle of the whole engineering machine.
The quarterly ritual
Each menu change, run the loop in order — it takes one afternoon with the numbers and one tasting session with the brigade:
- Re-run the matrix (chapter 1) on the outgoing card: which Stars survive the season change? Which Dogs finally exit, hidden gracefully behind "making room for the new season"?
- Re-cost everything (chapter 2) at the new season's prices — last quarter's recipe cards are already wrong.
- Re-anchor pricing (chapter 3): the new flagship sets the ceiling; check the spread didn't drift.
- Rotate the glasses (chapter 4): by-the-glass list and pairings follow the kitchen, and the cellar's slow movers get their by-the-glass exit.
- Re-stage the card (chapter 5): new Stars to the power positions, fresh descriptions, one new highlight.
Then let the change make noise: the new menu is a ready-made campaign for your marketing system, and the sales data it generates feeds the next quarter's matrix. The loop closes itself — margins, like sauces, are made by reduction and repetition.
Do this tonightPut the next menu-change date in the calendar now, with a two-hour "matrix + costing" block one week before it. The ritual that gets scheduled is the ritual that happens.
How engineered is your menu?
Eight yes/no checks across the six chapters. Be honest — the card doesn't care about feelings, and your browser remembers the score for next visit.