Most menus grow by addition. A new dish goes on, an old favorite stays, and a few years later nobody is quite sure which dishes pay for the rest. Menu engineering is the method restaurants have used for decades to answer that question with evidence instead of instinct.
It needs no special software, only honest numbers, and it works just as well for a twelve-dish bistro as for a long café menu.
Two numbers per dish
Menu engineering looks at each dish through two lenses. Popularity is how many portions sold in a period. Margin is what each sale earns: the selling price, before tax, minus the cost of the ingredients in one portion.
The approach was popularized by Michael Kasavana and Donald Smith at Michigan State University in 1982. Plot every dish on those two axes and it falls into one of four groups, each calling for a different move.
Hidden gems
Classic name: Puzzles
Less ordered, high margin
Give them visibility
Stars
Classic name: Stars
Popular, high margin
Protect them
Needs attention
Classic name: Dogs
Less ordered, low margin
Rework or retire them
Crowd favorites
Classic name: Plowhorses
Popular, low margin
Lift their margin
Step 1: cost every dish
Write down the ingredients in one portion, with quantities, and price them at what you pay today. Include the parts that are easy to forget: the sauce, the garnish, the oil, the bread that comes with it. Update the costing whenever a supplier changes a price or the recipe changes.
This is the step most restaurants skip, and without it the rest is guesswork. In TruePlate you add the cost to each dish, and dishes without one are shown as needing a cost rather than being guessed at.
Step 2: count what sold
Pick a period long enough to be fair, usually a few weeks to a month, on the same menu and without unusual events like a private party or a holiday closure. For each dish, record the quantity sold and the revenue it brought in.
Multiply the margin per portion by the quantity and you have each dish’s total margin for the period, which is the figure that shows what it actually contributed.
Step 3: sort into four groups
In the classic method, a dish counts as popular when its share of sales reaches 70 percent of an equal share (with 20 dishes, an equal share is 5 percent, so the line sits at 3.5 percent), and as profitable when its margin beats the menu’s weighted average. TruePlate keeps it simpler: it compares each dish with your menu’s median quantity sold and median total margin.
- Stars: popular and profitable. The dishes that carry the menu.
- Crowd favorites (classically, plowhorses): popular, but earning less than they could.
- Hidden gems (classically, puzzles): profitable, but not ordered enough.
- Needs attention (classically, dogs): neither popular nor profitable.
Step 4: act on each group
Stars: protect them
Keep quality and portion consistent, give them the best spots on the menu and resist discounting them. They sell because guests love them as they are.
Crowd favorites: lift their margin
Guests already order them, so work on the cost side. Re-cost the recipe, look at the portion, pair them with a higher-margin side or add-on, or test a small price rise. Avoid big changes to a dish people come back for.
Hidden gems: give them visibility
They earn well when they sell, so help guests find them. Move them higher on the menu, rewrite the description, add a better photo or a 3D model guests can place on the table, ask staff to recommend them, or feature them in a promotion.
Needs attention: rework or retire
Try a new version, replace them, or remove them to shorten the menu. Keep one only when it has a job to do, such as the only vegan main or the dish a regular always orders.
What a digital menu adds
Sales show what guests chose. A digital menu also shows what they considered. Dish views and AR sessions reveal interest before the order, and the gap between the two is often where the next improvement is.
- Viewed often, rarely ordered: the dish attracts attention but something stops the order. Check the price, the description and the photo.
- Interest while sold out: TruePlate shows interest in dishes that were marked out of stock, a direct signal to prep more.
- AR interest without orders: dishes guests opened in AR but did not order are worth a closer look at portion and price.
How often to re-engineer your menu
Most restaurants review the matrix every quarter, and again after a price change, a new dish or a supplier change. Seasonal menus deserve a look at the end of each season, so the next one starts from what worked.
Menu engineering sits alongside the other weekly numbers in our guide to restaurant KPIs. Together they tell you not only how the week went, but which dishes made it go that way.
Questions restaurants ask
What is menu engineering?
Menu engineering is a method for analyzing a menu by how popular each dish is and how much margin it earns, so you can decide which dishes to promote, reprice, rework or remove.
What are stars, plowhorses, puzzles and dogs?
They are the four groups of the classic menu engineering matrix. Stars are popular and profitable, plowhorses are popular but less profitable, puzzles are profitable but less popular, and dogs are neither.
How do I calculate a dish's margin?
Take the dish's selling price before tax and subtract the cost of the ingredients in one portion. Multiply by the number sold to get the total margin the dish earned over the period.
How often should a restaurant do menu engineering?
Most restaurants review it every three months, and again after a price change, a new dish or a supplier change.