Menu Engineering 101 — Stars, Puzzles, Workhorses, and Dogs

In 1982, Michael Kasavana and Donald Smith at Michigan State University published a paper that changed how restaurants think about menus. They proposed a simple matrix: plot every menu item by two axes, popularity and profitability, and you get four quadrants. Four types of dishes. Four different strategies.

Forty years later, this framework remains the most practical tool for menu optimization. Not because it’s academic, but because it gives you a clear action for every single item on your menu. Here’s how to use it.

The two axes that matter

The vertical axis is contribution margin. How much gross profit does each dish generate after you subtract food cost? A pasta dish that sells for EUR 14 with EUR 3.50 in ingredients contributes EUR 10.50. A steak at EUR 28 with EUR 12 in ingredients contributes EUR 16. The steak looks more profitable in absolute terms, but percentage-wise the pasta wins.

For menu engineering, we use absolute margin, not percentage. Why? Because your landlord doesn’t accept percentages. EUR 10.50 in your pocket from that pasta matters more than a “75% food cost ratio” that sounds impressive but means less money per plate.

The horizontal axis is popularity. How often does this dish sell relative to other items? Kasavana and Smith defined the threshold as 70% of the expected average. If you have 20 items and sell 1,000 dishes per month, each item should sell 50 on average. The popularity threshold is 70% of that: 35 sales. Anything above 35 is popular. Below is unpopular.

The four quadrants

Stars (high profit, high popularity)

These are your best items. Guests love them, and each sale fills your margin. The strategy: protect them. Don’t change the recipe, don’t move them to a less visible menu position, don’t raise the price aggressively. Stars earn their place.

In Plattr’s Menu Intelligence panel, Stars surface automatically. The system tracks both sales volume and margin per item, so you can see which dishes belong here without running spreadsheets manually.

Puzzles (high profit, low popularity)

Puzzles make you money when they sell, but they don’t sell enough. The fix is almost never to lower the price. Instead, look at visibility. Is the dish buried on page three of the menu? Does the description sound appealing? Does the server know how to recommend it?

Try repositioning Puzzles into the high-attention zones of your menu. Add a photo. Write a better description. Train staff to mention it. If none of that works after 4-6 weeks, consider whether the dish belongs on the menu at all.

Workhorses (low profit, high popularity)

The crowd pleasers that barely make you money. Classic examples: Caesar salad, margherita pizza, house burger. Everyone orders them, but the margins are thin. The strategy: engineer the cost down or the price up, carefully.

Can you reduce portion size slightly without guests noticing? Can you swap an expensive ingredient for something that tastes the same but costs 20% less? Can you raise the price by EUR 0.50-1.00 without losing volume? Recipe engineering is your best friend here. Small changes across high-volume items add up fast.

Dogs (low profit, low popularity)

Nobody orders them, and when they do, you barely make money. The obvious move is to remove them. But think first: does this dish serve a structural purpose? Is it the only vegetarian option? The only thing for kids? If it fills a gap, keep it but hide it. If it doesn’t, cut it.

Removing Dogs frees up kitchen complexity, inventory space, and mental load for your team. A tighter menu almost always outperforms a sprawling one.

How to build the matrix for your restaurant

You need two data sets: sales mix and food cost per item. Pull your sales data from the last 8-12 weeks (shorter periods get skewed by seasonal effects). For each item, calculate:

1. Total units sold
2. Selling price
3. Food cost per portion (ingredients only, not labor)
4. Contribution margin = selling price minus food cost

Then calculate the average contribution margin across all items. Items above average go in the “high profit” row. Items below go in “low profit.”

For popularity, calculate the expected average (total items sold divided by number of menu items), multiply by 0.7, and that’s your cutoff. Above = popular. Below = unpopular.

If you’re using Plattr, the Insights Hub does this automatically. Connect your menu data and sales, and the matrix builds itself. No spreadsheets, no formulas.

Acting on the matrix

The matrix is diagnostic. It tells you what’s happening. The action plan is where operators get stuck. Here’s a practical sequence:

Start with Dogs. Remove 2-3 items that add no value. This simplifies your kitchen and your menu. Next, pick your top 2-3 Puzzles and run a 4-week experiment: reposition them, add photos, brief your servers. Track whether popularity changes.

For Workhorses, do a food cost audit on each one. Find the items where a small recipe adjustment saves the most money at volume. A EUR 0.30 reduction on a dish you sell 200 times per month saves EUR 720/year.

For Stars, do nothing except make sure they stay visible. If you’re redesigning your menu layout, Stars go in the prime spots. If you’re running a cross-selling strategy, pair add-ons with your Stars since those guests are already happy with their choice and open to extras.

Common mistakes

The biggest mistake is running the analysis once and filing it away. Menu engineering is a continuous process. Your sales mix shifts with seasons, staffing, ingredient prices, and local competition. Re-run the matrix every quarter at minimum.

Second mistake: treating the matrix as gospel. A dish might be a “Dog” on paper but be the reason a specific customer segment visits you. Use the data as input, not as a command.

Third: ignoring the connection between menu engineering and your overall revenue strategy. The matrix works best when combined with pricing strategy, menu layout, and staff training. It’s one piece of a larger system.

Getting started today

If you don’t have a POS that exports item-level sales data, start with pen and paper. Ask your kitchen which items sell the most. Estimate food costs for your top 10 items. Plot them on a 2x2 grid. Even a rough version of this exercise will reveal at least one surprise, a dish you thought was a winner that’s actually dragging margins down, or a hidden gem that nobody’s been promoting.

For restaurants already on Plattr, open your Insights Hub and look at Menu Intelligence. The quadrant view is built in, updated with every order. That’s the starting point for every menu decision you make from here.