Menu Engineering with POS Data: Finding Your Most Profitable Dishes

Most restaurant menus are built the way they were on opening day: a chef’s judgement about what belongs, a rough markup applied to food cost, and a layout that follows the order of a meal. What almost never happens afterwards is the part that decides profitability, which is going back through actual sales data to work out which dishes are earning their place and which are quietly costing money every time someone orders them.

That process is menu engineering, and the data needed to do it is already sitting inside your restaurant POS system. Every order recorded over the past six months is a record of what customers actually chose, not what the menu assumed they would.

What Menu Engineering Actually Means

Menu engineering is the practice of evaluating every dish on two axes at once, how much money it makes per plate and how often people order it, then reorganising the menu around what those two numbers reveal together. Neither figure means much alone. A dish with an excellent margin that sells twice a week contributes almost nothing. A dish that sells constantly at a thin margin keeps the kitchen busy without moving profit much either.

Looked at together, though, they tell you where the money in your menu actually comes from, which is frequently not where owners assume it does. The signature dish everyone associates with the restaurant is often not the one paying the rent.

The Two Numbers That Decide Everything

The first is contribution margin: menu price minus the cost of the ingredients that go into that specific dish. Note that this is a dirham figure, not a percentage. A dish selling at AED 60 with AED 20 of ingredients contributes AED 40 per plate.

The second is popularity, expressed as the share of total orders within its category rather than as a raw count. Comparing a starter’s sales against a main course’s tells you little, since they sell at naturally different volumes. Comparing a starter against other starters tells you something useful.

Both numbers come out of the POS, provided recipe costing has been configured. Sales counts are recorded automatically on every order; ingredient cost per dish comes from the recipe-level tracking covered in our guide to restaurant inventory management. Without that costing layer in place, you have popularity data but no margin data, which is only half the picture.

Sorting Your Menu Into Four Categories

Once each dish has both numbers attached, every item falls into one of four groups depending on whether it sits above or below the average for its category on each axis.

Category	Margin	Popularity	What it means
Stars	High	High	Your best dishes. Protect them.
Puzzles	High	Low	Profitable but overlooked. Fixable.
Workhorses	Low	High	Popular but thin. Improve margin carefully.
Dogs	Low	Low	Costing you space and stock. Usually cut.

What to Do With Each Category

Stars need protecting rather than improving. Keep the recipe consistent, keep them prominent on the menu, and resist raising the price aggressively just because they sell well, since these are usually the dishes customers return specifically for.

Puzzles are the most immediately rewarding group, because the margin is already good and only visibility is missing. Moving the dish to a better position on the menu, rewriting a flat description, or briefing staff to recommend it often lifts sales without changing anything about the dish itself.

Workhorses require the most care. They sell in volume, so any change carries risk. Small portion adjustments, a cheaper garnish, a modest price increase, or renegotiating the supplier cost of the main ingredient can all widen the margin, but a large price jump on a high-volume dish is exactly the kind of change customers notice and resent.

Dogs are usually removed, but not always. A dish that loses money and sells rarely still deserves a second look if it shares ingredients with popular items, serves a dietary requirement that keeps groups from going elsewhere, or is genuinely part of the restaurant’s identity.

Why Menu Placement Changes the Numbers

Popularity is not fixed. Where a dish appears on the page measurably affects how often it is ordered, which means a Puzzle can be converted into a Star through layout alone. Items near the top of a section and at the end of it get read more carefully than those buried in the middle, and anything visually separated, boxed, shaded, or given a photograph, draws disproportionate attention.

Category length matters as well. A section listing fifteen options spreads orders thinly and makes decisions harder, while a tighter list concentrates sales on the dishes you actually want to sell. Cutting Dogs therefore does more than remove weak items, it also increases the share of attention going to everything that remains.

How Often to Repeat the Exercise

Quarterly is a reasonable rhythm for a full review. That is long enough for menu changes to show up in the data and short enough to catch ingredient cost increases before they erode margins across a whole season. Supplier prices move, and a dish that was comfortably a Star can drift toward Workhorse territory without anything visible changing on the plate.

Between full reviews, it is worth watching contribution margin on your top few sellers monthly, since those are the dishes where a small cost increase multiplies across the largest number of covers. For groups running several branches, the same analysis run per location often reveals that a dish performing as a Star in one branch is a Dog in another, which is an argument for branch-specific menus rather than one identical list everywhere.

Conclusion

A menu is one of the few things in a restaurant that can be improved without spending money, hiring anyone, or changing how the kitchen works. It only requires knowing which dishes earn well, which sell well, and which do neither. That information already exists in the sales history your POS has been quietly collecting, and turning it into a repricing decision, a layout change, or a removal is what separates a menu that was designed once from one that is actually managed.

Frequently Asked Questions

1. How much sales data do I need before this is meaningful?

Around three months is a practical minimum, since it smooths out one-off weeks and gives each dish enough orders for the comparison to mean something.

2. Should I remove every dish that sells badly?

No. Check the margin first, because a slow seller with a strong margin is usually a placement or description problem rather than a dish problem.

3. Is a high food cost percentage always bad?

Not necessarily. A dish with a higher percentage can still contribute more dirhams per plate than a cheaper one, and dirhams are what cover fixed costs.

4. Can I do menu engineering without recipe costing set up?

Only partially. Sales counts alone show popularity, but without ingredient costs per dish you cannot calculate margin, which is half the analysis.

5. Does this work for delivery menus too?

Yes, though aggregator commission and packaging must be subtracted as well, which sometimes moves a dine-in Star into a much weaker position on delivery.

Want to know which dishes are actually carrying your menu? Explore iTudeTech‘s restaurant POS software or get in touch with our team to see what your sales data can tell you.