F&B Operations

Menu Engineering for Restaurants: A Step-by-Step Guide

Menu engineering tells you which dishes earn their place on the menu and which need fixing. This guide walks through the full method, with a 10-dish worked example you can copy into a spreadsheet.

Hand pointing at dishes on a restaurant menu
  • Menu engineering sorts every dish on two measures: how often it sells (popularity) and how much money it leaves after ingredient cost (contribution margin).
  • You need two data sets: item sales for a period from your POS, and a current plate cost for every dish from recipe costing.
  • The popularity line is (1 ÷ number of items) × 70%. The margin line is the weighted average contribution margin across everything sold.
  • Stars stay as they are, plowhorses get re-costed or repriced, puzzles get repositioned, and dogs get fixed or removed.
  • Run it separately for dine-in and delivery, repeat it every quarter, and watch out for small samples, seasonality and set meals.

What menu engineering is, and where it comes from

Menu engineering is a way of analysing a menu dish by dish, using two numbers you already have: how many of each dish you sold, and how much each one contributes after its ingredient cost. The approach most operators use today is commonly credited to Michael Kasavana and Donald Smith, whose 1982 work on menu engineering introduced the four categories most people know: stars, plowhorses, puzzles and dogs.

The point of the exercise is to stop judging dishes by gut feel or by food cost percentage alone. A dish with a low food cost percentage can still earn very little per plate if its price is low, and a dish with a higher food cost percentage can be one of your best earners if it sells well at a healthy price. Menu engineering puts both factors side by side.

This guide goes deeper than the short example in our menu pricing strategy guide. It covers the full calculation, a larger worked example, channel differences and how to make the review a routine. For how to lay the menu out once you know which dishes to push, see our menu design guide.

The data you need before you start

Menu engineering is only as good as the numbers you put in. Gather these for one period, usually a month or a quarter:

  • Item sales mix. The number of each dish sold in the period, from your POS item sales report. Use quantities, not revenue.
  • Net selling price. The price you actually receive per dish, excluding GST and service charge. If a dish was sold at several prices (happy hour, promotions, staff meals), use the average net price per unit sold, or exclude discounted sales and note it.
  • Plate cost. The ingredient cost of one portion, including garnish, sauces and sides served with it. This comes from a costed recipe card. If you don’t have recipe cards yet, start with our recipe costing template.
  • Channel. Whether each sale was dine-in, takeaway, your own online orders or a delivery platform. You’ll need this later.

Two practical points. First, the plate cost must be current. If your chicken supplier raised prices last month and the recipe card still shows the old cost, every result for chicken dishes will be wrong. Our food cost guide explains how to keep costs up to date. Second, compare like with like. Analyse mains against mains, starters against starters and drinks against drinks. A $4 iced tea and an $18 main don’t belong in the same table.

If your POS handles recipe-level stock deduction and recipe costing, most of this data sits in one place. ChaChaCha deducts ingredients by recipe with each sale and holds a recipe cost per menu item, so the plate cost and the sales count come from the same system rather than two spreadsheets that drift apart.

Step-by-step calculation

Here is the full method. Set up a spreadsheet with one row per dish in the category you’re analysing.

  1. List every item and its quantity sold. Add them up to get the total number of items sold in the category.
  2. Work out each item’s menu mix percentage. Menu mix % = quantity sold ÷ total quantity sold × 100.
  3. Work out each item’s contribution margin (CM). CM = net selling price − plate cost. This is the money each plate leaves to pay for labour, rent, utilities and profit.
  4. Work out total contribution per item. Item total CM = CM × quantity sold.
  5. Work out the average contribution margin. Average CM = total CM across all items ÷ total quantity sold. This is a weighted average, so dishes that sell more count for more.
  6. Set the popularity threshold. Using the commonly applied Kasavana and Smith approach, the threshold is (1 ÷ number of items) × 70%. With 10 items, an equal share would be 10%, so the threshold is 7%. An item is “high popularity” if its menu mix % is at or above that line.
  7. Classify each item. Compare its CM to the average CM (high or low) and its menu mix % to the threshold (high or low).

The 70% factor exists because, on any menu, sales are never spread evenly. Requiring every dish to hit a full equal share would label too many reasonable dishes as unpopular. Some operators use a different factor, but 70% is the usual starting point. Whatever you choose, keep it the same each time so results are comparable.

Classifying: stars, plowhorses, puzzles and dogs

Category Popularity Contribution margin In plain terms
Star High High Sells well and earns well
Plowhorse High Low Sells well but earns less per plate
Puzzle Low High Earns well per plate but doesn’t sell enough
Dog Low Low Neither sells nor earns well

The labels are shorthand, not verdicts. A dog can have a reason to stay, and a star can still be improved. The categories tell you where to look first.

Worked example: 10 mains (hypothetical)

This example is hypothetical. The dishes, prices, costs and quantities are illustrative, not benchmarks. A casual restaurant exports one month of mains sales from its POS. Prices are net of GST and service charge. Plate costs come from current recipe cards.

Step 1 to 4: the raw calculation

Dish Net price Plate cost CM per plate Food cost % Sold Menu mix % Total CM
Chicken rice set $9.90 $3.40 $6.50 34.3% 520 23.5% $3,380.00
Seafood laksa $12.90 $3.60 $9.30 27.9% 410 18.5% $3,813.00
Nasi lemak $10.90 $3.20 $7.70 29.4% 300 13.5% $2,310.00
Fish and chips $14.90 $5.90 $9.00 39.6% 260 11.7% $2,340.00
Aglio olio $13.90 $2.90 $11.00 20.9% 240 10.8% $2,640.00
Mee goreng $9.50 $3.80 $5.70 40.0% 120 5.4% $684.00
Beef rendang rice $16.90 $5.20 $11.70 30.8% 110 5.0% $1,287.00
Salted egg prawns $18.90 $6.10 $12.80 32.3% 95 4.3% $1,216.00
Tom yum soup $11.90 $4.60 $7.30 38.7% 90 4.1% $657.00
Chilli crab pasta $19.90 $7.40 $12.50 37.2% 70 3.2% $875.00
Total 2,215 100% $19,202.00

Step 5 and 6: the two lines

  • Average CM = $19,202.00 ÷ 2,215 = $8.67 per plate.
  • Popularity threshold = (1 ÷ 10) × 70% = 7.0% of mains sold, which is about 155 plates this month.

Step 7: the classification

Dish Menu mix % vs 7.0% CM vs $8.67 Category
Seafood laksa 18.5% (high) $9.30 (high) Star
Fish and chips 11.7% (high) $9.00 (high) Star
Aglio olio 10.8% (high) $11.00 (high) Star
Chicken rice set 23.5% (high) $6.50 (low) Plowhorse
Nasi lemak 13.5% (high) $7.70 (low) Plowhorse
Beef rendang rice 5.0% (low) $11.70 (high) Puzzle
Salted egg prawns 4.3% (low) $12.80 (high) Puzzle
Chilli crab pasta 3.2% (low) $12.50 (high) Puzzle
Mee goreng 5.4% (low) $5.70 (low) Dog
Tom yum soup 4.1% (low) $7.30 (low) Dog

Two things stand out. Fish and chips has the second-highest food cost percentage on the list at 39.6%, yet it is a star because it leaves $9.00 a plate and sells steadily. If you judged by food cost percentage alone, you might have cut it. And the chicken rice set is the best seller by far, but at $6.50 a plate it earns less than the laksa in total despite selling 110 more portions.

What to do with each quadrant

Stars: protect them

  • Keep the recipe, portion and plating consistent. Stars are what regulars come back for.
  • Give them prime positions on the menu and in QR ordering, and make sure they are never out of stock at peak.
  • Test a small price rise only if the dish is clearly better than nearby alternatives. Our guide to raising menu prices carefully has more on this. Watch the next month’s quantity closely.

Plowhorses: raise the margin without losing the volume

  • Re-cost the recipe first. Can a garnish, side or sauce be changed without guests noticing a drop in quality?
  • Test a modest price rise. In the example, if the chicken rice set went up by $0.50 and still sold 520 plates, contribution would rise by $260 for the month.
  • Pair it with a higher-margin add-on, such as a drink or side, so the ticket earns more even if the dish itself doesn’t. Our upselling guide has scripts for staff.
  • Don’t move it to a less visible spot on the menu. Plowhorses often bring people in.

Puzzles: help guests find them

  • Rewrite the description, add a photo, or move the dish to a more visible position. For layouts to start from, see menu template examples.
  • Brief staff to recommend it, and offer tastings during a staff briefing so they can describe it honestly.
  • Check the price point. The chilli crab pasta at $19.90 may simply sit above what most guests at this restaurant spend on a main. A smaller portion at a lower price can sometimes lift volume while keeping a good margin.
  • If a puzzle doesn’t move after two review cycles, consider it a candidate for replacement. When you bring in a replacement, test, cost and train before launch.

Dogs: fix, reposition or remove

  • Ask whether the dish has a job. A vegetarian option, a children’s dish or a dish a group of regulars always orders may be worth keeping even if it classifies as a dog.
  • If it has no job, try re-costing and repricing once. Mee goreng at 40% food cost and $5.70 a plate needs either a cheaper recipe or a higher price.
  • If neither works, remove it. Fewer dishes usually means less prep, fewer ingredients held in stock and less waste.

Dine-in vs delivery: run the numbers by channel

A dish can be a star in the dining room and a dog on a delivery app. Delivery sales carry costs that dine-in sales don’t: platform commission, packaging and sometimes a different menu price. If you blend channels in one table, the delivery costs are hidden and your results will flatter the delivery menu.

To run a delivery version, adjust the contribution margin: delivery CM = delivery menu price (net) − plate cost − packaging − platform commission on that price. Use the commission rate in your own platform contract. The example below uses 30% purely as a placeholder so you can see the arithmetic; it is not a quoted rate.

Hypothetical, four dishes from the example above, delivery prices set higher than dine-in:

Dish Delivery price Plate cost Packaging Commission (30% placeholder) Delivery CM Dine-in CM
Seafood laksa $14.90 $3.60 $0.50 $4.47 $6.33 $9.30
Aglio olio $15.90 $2.90 $0.60 $4.77 $7.63 $11.00
Fish and chips $16.90 $5.90 $0.80 $5.07 $5.13 $9.00
Chicken rice set $11.90 $3.40 $0.40 $3.57 $4.53 $6.50

Even with higher delivery prices, every dish earns less per order than it does in the dining room, and the gap is not the same for each dish. Fish and chips loses the most, partly because of its packaging and partly because a percentage commission on a higher price takes more dollars. It also travels badly. That might make it a delivery-menu candidate for removal while it stays a dine-in star.

Your own online ordering and QR pre-orders sit somewhere in between: no platform commission, but payment processing and possibly your own delivery cost. Tag orders by channel in the POS so you can split the sales mix. If GrabFood orders flow into the POS through delivery integration and your QR and online orders are tagged, the channel split comes from one export. See our delivery operations guide for more on running a separate delivery menu.

Making it a quarterly routine

One analysis tells you where you stand. The value comes from repeating it and seeing whether your changes worked. A quarterly rhythm suits most restaurants:

  1. Week 1 of the quarter: export the previous quarter’s item sales by channel from your POS reports. Update plate costs for any ingredient whose price has moved.
  2. Week 1: run the calculation for each menu category and each main channel. Save the results next to the previous quarter’s.
  3. Week 2: agree actions with the chef: at most two or three changes per category, so you can tell which change had which effect.
  4. Weeks 2 to 3: make the changes: recipe tweaks, prices, descriptions, menu positions, staff briefings. Log the date of each change.
  5. Monthly check-in: glance at item sales for the changed dishes. You don’t need the full analysis every month, just a check that nothing has collapsed.
  6. Next quarter: run it again and compare. Did the puzzle you repositioned move up? Did the plowhorse price rise hold its volume?

Keep a simple change log alongside the spreadsheet. Six months later, nobody will remember when the rendang description was rewritten. Quarterly results also feed neatly into your wider restaurant KPI review, particularly item gross margin and food cost percentage.

Common pitfalls

  • Small samples. A dish sold 12 times in a slow month doesn’t tell you much. If your total volume per category is low, use a full quarter, or combine two months, before drawing conclusions. Avoid analysing a dish that has only been on the menu for a couple of weeks.
  • Seasonality. Soups and hot dishes may sell differently in the wetter months, and festive periods such as Chinese New Year, Hari Raya and Christmas can distort the mix. Compare a quarter with the same quarter last year where you can, and note any promotions, festive menus or events that ran in the period.
  • Bundled items and set meals. If a set includes a main, a drink and a side, the POS may record the set as one item or break it into components. Decide how you’ll treat sets and stick to it. One approach is to analyse sets as items in their own right with their full plate cost; another is to allocate the set price across the components. Mixing the two will give nonsense results.
  • Out-of-stocks. A dish that ran out every Friday night will look less popular than it is. Check for stock-outs before calling something a puzzle or a dog.
  • Stale plate costs. Using last year’s costs is the most common reason results are wrong. Re-cost before every analysis.
  • Discounts and comps. Staff meals, complimentary dishes and heavy promotions change the net price. Either exclude them or use actual average net price.
  • Treating it as the only answer. The method looks at contribution per plate, not labour or prep time. A dish with a high margin that takes one cook ten minutes at peak may not be worth pushing. Use judgement alongside the numbers.

Where your POS fits in

You can do menu engineering with a paper sales tally and a calculator, but it’s tedious and error-prone. The practical minimum is a POS that gives you item quantities by period and channel, and a reliable plate cost for each dish.

ChaChaCha covers both ends. Back-office reports give item sales for any period, and recipe costing per menu item, with ingredients deducted by recipe as each dish sells, keeps plate costs tied to what you actually use. Wastage tracking is included as well. If you need other inventory specifics, such as purchase orders or unit conversions, ask us to confirm how they work for your set-up. Multi-outlet groups can also use the head office ERP; ask us to confirm how cross-outlet item reporting is set up for your group. See our restaurant POS page, or contact us to walk through your own menu data.

Web Admin

Written by Web Admin

Frequently asked questions

What is menu engineering in a restaurant?

Menu engineering is a method of analysing each dish by popularity and contribution margin, then sorting dishes into stars, plowhorses, puzzles and dogs. It is commonly credited to Kasavana and Smith in 1982. The results show which dishes to protect, reprice, promote or remove, based on your own POS sales and recipe costing data rather than guesswork.

How do you calculate the popularity threshold in menu engineering?

The commonly used method sets the threshold at (1 divided by the number of items) multiplied by 70%. With 10 items in a category, an equal share would be 10%, so the threshold is 7%. Any dish whose share of total quantity sold is at or above 7% counts as high popularity. Keep the same factor each time so results are comparable.

Why use contribution margin instead of food cost percentage?

Food cost percentage shows cost relative to price, but not how many dollars a plate leaves to cover labour, rent and profit. A dish with a higher food cost percentage can still earn more per plate if its price is higher. Contribution margin, which is net price minus plate cost, measures what each sale actually adds to the business.

How often should I run a menu engineering analysis?

Quarterly works for most restaurants. It gives enough sales volume for reliable results and enough time to see whether changes worked. Check item sales for changed dishes monthly, and re-cost recipes whenever a key ingredient price moves. Run a fresh analysis after a major menu change once the new dishes have had a few weeks of sales.

Should I run menu engineering separately for delivery?

Yes. Delivery orders carry platform commission, packaging and often a different price, so the contribution margin per dish is different from dine-in. A dish can be a star in the dining room and weak on a delivery app. Tag orders by channel in your POS and run the calculation for each main channel separately.

What should I do with a dog on my menu?

First ask whether the dish has a job, such as being the only vegetarian option or a favourite of a group of regulars. If it does, try re-costing or repricing it. If it has no clear role and still performs poorly after one fix, remove it. A shorter menu usually means less prep, less stock held and less waste.

Is menu engineering based on the BCG matrix?

It borrows the idea. The Boston Consulting Group growth-share matrix sorts products into stars, cash cows, question marks and dogs. Menu engineering applies a similar two-by-two grid to dishes, using popularity and contribution margin instead of market growth and share. Plowhorses play the role of cash cows and puzzles the role of question marks, while stars and dogs keep their names.

Should I use menu engineering software or hire a consultant?

Start with a spreadsheet and your POS item sales report; the calculation takes an afternoon once your recipe costs are current. Software helps when it pulls item sales and plate costs automatically, so you can repeat the review every quarter without retyping. A consultant adds most value for a full menu redesign, a new concept or a group with several outlets, but ask to see their method and sample reports first.

What is the difference between menu engineering and menu design?

Menu engineering is the analysis: it uses sales and cost data to decide which dishes to keep, reprice, rework or remove. Menu design is the presentation: layout, descriptions, photos, price formatting and where each dish sits on the page or screen. Do the engineering first, then use design to give stars and puzzles the best positions, and check the results in the next quarter's numbers.

How is menu engineering used in hotels?

Hotels run the same calculation, but separately for each outlet and meal period: the all-day restaurant, room service, the bar and banquet menus each have different prices, costs and guests. Banquet packages are usually analysed as whole packages rather than individual dishes. Room service menus also need packaging and labour time in mind, because a high-margin dish can be slow to deliver to a room.

See ChaChaCha running in your business

Book a 20-minute walkthrough. We'll set up your menu or catalogue and your payments so you can see how it would run on day one.