Menu engineering, done properly

1What is menu engineering?

A sorting method, published by Michael Kasavana and Donald Smith in 1982. Plot every dish on two axes, how often it sells and how many rupees of margin each plate leaves behind, and it lands in one of four boxes:

Two lines draw the boxes. A dish counts as popular if it sells at least 70% of the average count per dish: on a 20-dish menu the average dish takes 5% of the plates sold, so the line sits at 3.5%. A dish counts as profitable if its margin per plate beats the menu’s weighted average margin, which is total margin divided by total plates sold.

Note the unit: rupees, not food-cost percentage. In the example below, dal makhani runs a 22% food cost and leaves ₹201 a plate; mutton rogan josh runs 38% and leaves ₹297. The dish with the “worse” percentage earns more on every plate. Percentages belong in reports; rupees go to the bank.

2Why does it matter on an Indian menu?

Three reasons. Indian casual-dining menus run long, and a long card hides dishes nobody has costed since it was printed. Ingredient prices move with every season, so your kitchen re-costs itself whether or not your menu does. And most Indian menus carry deliberate cross-subsidies, with breads and rice priced soft to make the mains feel fair, which only work if someone checks the arithmetic.

The gains come from demand you already have. Nothing in this guide needs a new customer: the same guests, ordering from a card that earns more per plate.

3How do you get a true cost per plate?

The sort is only as honest as its cost column, and this is where many attempts fail. Three rules:

Cost the recipe at usable yield. A kilo of raw chicken does not become a kilo of servable chicken, vegetables lose weight in trimming, and frying absorbs oil. Cost what reaches the plate, and include what menus forget: the butter brushed on the naan, the garnish, the onions and chutney that go out free.

Cost each base once, then reuse it. A makhani gravy may sit under six dishes, a fried-rice base under four. Cost the gravy per litre as a recipe of its own, and every dish built on it moves when tomato prices do. This one habit turns costing from a weekend into an hour.

Refresh purchase prices every quarter. A sort built on last year’s chicken price will point you at the wrong dishes with total confidence. Costs must follow what you paid last month, not the rate card from opening day.

The food cost guide covers yields, bases and variance in depth, and the food cost calculator does the per-dish arithmetic.

4A worked example: 20 dishes

Pull a full month of per-dish sales from your billing system, attach the cost per plate, work out the margin (price minus cost), then draw the two lines.

The table is an illustrative example: an invented North Indian casual-dining menu with plausible numbers, not data from any restaurant. Across these 20 dishes the kitchen sold 7,480 plates in the month, an average of 374 per dish, so the popularity line sits at 70% of that: about 262 plates. Total margin divided by plates sold gives a weighted average margin of about ₹157 a plate, which is the profitability line.

Read what the totals hide. This menu takes about ₹16.6 lakh a month and keeps about ₹11.8 lakh after ingredients: a blended food cost near 29%, inside the 28–32% band many operators aim for. Yet the sort still finds six plowhorses and three dogs. A menu can look fine on average and leak dish by dish; the matrix gives each leak a name.

Look at who the plowhorses are: the breads, the rice and the lassi, the highest-volume lines on the card, plus kadai chicken, whose ingredients have quietly outrun its price at a 42% food cost. The pattern is common in Indian casual dining.

5What do you do with each group?

Stars: protect them, then promote them. Leave a star’s price alone; these dishes carry the room. Protect them in the kitchen with recipe cards and portion checks, because a star that turns inconsistent becomes a complaint. Then promote them: build combos around them and put them where eyes land first on the card. Never discount a star to fill seats, or you pay guests to order what they already order.

Plowhorses: where the money usually is. High volume and a thin margin mean a small price move repeats across hundreds of plates. In the example, butter naan sells 980 plates a month at ₹70; at ₹80 it adds ₹9,800 a month of margin if the count holds, and bread demand follows the mains it goes with. Jeera rice from ₹159 to ₹169 adds ₹4,400 more. Two quiet changes make ₹14,200 a month, about 1.2% more margin than the menu earned before, from ten minutes of decisions. Where a price cannot move because guests compare it everywhere, re-cost instead: portion, yield, a cheaper garnish.

Kadai chicken is the other kind of plowhorse. At a 42% food cost its recipe has drifted from its spec. Bring the portion back to ₹92 of ingredients and move the price to ₹279, and it leaves ₹187 a plate, enough to make it a star.

Reprice in small steps, one menu print at a time, and watch the weekly counts for a month after each change. To find the price a dish needs, enter its cost and your target food cost:

The calculator gives a dine-in price and a delivery-app price; the second grosses up for the app’s commission, so the same margin lands after the app takes its share.

Puzzles: reposition before you touch the price. These earn well and sell poorly, which is usually a visibility or confidence problem. Rename and describe them: one line about how the rogan josh is slow-cooked is worth trying before any price cut. Move the dish next to a star on the card. Give servers one suggestion line for tables that order breads and nothing else exciting. If a puzzle still will not move after a real quarter of effort, it may be on the wrong menu.

Dogs: cut them, with a short appeal. Every dog you remove shortens prep, simplifies stock and frees a line on the card for something that earns. Two reprieves only: a dog that completes a set (a dessert on a family menu can live inside a thali or a combo instead), or one that a loyal, high-spending table orders every time. Otherwise, two quarters running in the dog box and it comes off the card.

6Should the delivery menu get its own matrix?

Yes, and many owners skip it. On a delivery-app order the app’s commission and the packaging come off every plate, so a dish’s margin on delivery is a different number from its margin at the table. A dine-in star can be a delivery dog once the app takes its share and the box is paid for.

Run delivery as a sort of its own, with delivery-only counts and margins after commission and packaging, and set delivery prices dish by dish rather than adding one flat percentage; the menu price calculator above does the gross-up. Tax works differently too: on food ordered through an app, the app pays the GST under section 9(5) of the CGST Act, and the GST guide explains how. Some dishes should leave the delivery card for travel reasons alone: a dish that arrives soggy earns refunds and one-star ratings that no margin can pay for.

CountStand takes dine-in and takeaway orders; delivery-app orders are not open at launch, so build your delivery sort from the apps’ own sales reports.

7Which mistakes skew the matrix?

Five, and each one quietly moves dishes into the wrong box:

• One sort for the whole card. Breads, drinks and mains have different price structures, so a ₹40 roti that sells well will always look like a plowhorse beside a ₹350 curry. Run the sort across the whole card, and again within each category: mains against mains, breads against breads.
• An unusual month. A festival rush, a closure for repairs or a supplier shortage bends the counts. Use an ordinary month, or two months together.
• Combos and thalis counted twice. Either treat each combo as a dish of its own or split its sales into its parts, and do the same every time.
• Plates fired instead of plates sold. Count what was billed and paid for, not what left the kitchen. Voids and comps belong in a different report.
• Percentages instead of rupees. Sorting by food-cost percentage favours cheap dishes and punishes the ones that earn the most per plate. The profitability line is rupees.

8How often should you rerun it?

Quarterly works: often enough to catch ingredient prices moving, rarely enough to let each change show. Rerun it also before any reprint, after a real supplier price shock, when the season changes the card, and six weeks after a new dish launches, long enough to call it a star or a puzzle honestly.

Menu engineering is the first move for raising the average bill, which makes it the first chapter of any plan to grow; the guide to increasing restaurant sales ranks it against 16 other moves. The first run is the slow one. After that, a rerun takes an evening.

9Can software run the matrix for you?

A spreadsheet can: export item sales, attach recipe costs, sort twice, label the groups. It works once. By the third quarter, the export-attach-sort routine tends to lose to a busy Friday.

The lasting fix is to let the system that holds both numbers do the join. Your billing already counts every plate, and recipe-linked stock already knows what each plate costs. In CountStand, billing, recipes and stock live in one system, and the Item performance report shows what each dish sold. In every plan, menu health runs every Monday: it sorts the menu into the four groups from the last 30 days of sales and your recipe costs, and names the dishes worth a price change, a better place on the card, or a rework. It suggests; a price changes only when you change it.

See the rest of the till on the restaurant POS page, or book a demo and bring last month’s card.

10Questions owners ask