Cost-Optimised Pricing Makes Restaurants. Reckless Pricing Quietly Unmakes Them.

Every few months an owner sits across from me with a menu in one hand and a P&L in the other, and the two documents have never met. The menu was priced by walking the street — a little under the place next door, rounded to something tidy, with a launch promotion on the apps to “build volume”. The P&L says the room is busy and the month is short. I have watched this pairing for the better part of three decades, across the UAE and wider GCC, India and Singapore, and the mechanism is the same everywhere. The price was set by looking sideways. The costs were set by the business. The gap between the two is where restaurants go to die slowly.

My argument is short and the rest of this piece proves it with arithmetic you can check. Cost-optimised pricing — a price built upward from the recipe, through the published cost bands, with VAT, commission and occupancy inside the sum — makes a restaurant that lasts. Reckless pricing, and its respectable cousin “cheap” pricing, makes a restaurant that is popular right up to the day it closes. Dubai punishes the second kind faster than most cities, and I will show you why.

What does “cost-optimised” pricing actually mean?

It means the price of a dish is an output, not an opinion. You start at the bottom — what the plate costs to put in front of a guest — and build up through the shares of revenue the rest of the business needs, until the price falls out of the arithmetic. The bands I build to are the ones behind the Restaurant Operating Index and the Profit Leak Audit: food cost 28–32% of net revenue with 32% as the ceiling, labour 25–30%, the two together — prime cost — between 55% and 62%, rent 6–12%, and delivery commission 3–6% of total revenue. Published, typical bands, not advice; your own recipe cards and P&L are the figures that count.

Take one illustrative plate — round numbers chosen so the arithmetic is easy to follow, a teaching dish, not a quotation.

Step	Basis	AED
Recipe cost of the plate	Costed from the recipe card — every ingredient, portion, garnish and trim	18.00
Net selling price	Recipe cost ÷ 0.30 (a 30% target, inside the 28–32% band)	60.00
VAT	5% on the net price, collected for the government, never yours	3.00
Menu price the guest sees		63.00

Check it: 18 ÷ 0.30 = 60.00, and 60 × 1.05 = 63.00. The price is AED 63, and it did not come from the street; it came from the plate.

Now run the sanity check downward, because a price that only clears food cost has cleared nothing. At a net AED 60, labour at 28% is AED 16.80, so prime cost is 34.80 — 58% of net, inside the 55–62% band. Rent at 10% is AED 6.00. What remains is AED 19.20, 32% of net, to carry the delivery commission share, utilities, marketing, repairs, licences, finance and the owner’s return. Hold delivery at 5% of revenue — AED 3.00 — and 27% is left for everything else, exactly the healthy-month picture in the margins piece. That is the whole method. A price that holds each band leaves a remainder that is real. A price that flatters any band takes the remainder away, and the remainder is the restaurant.

Two clarifications. “30% food cost” is a target you engineer each price to; the blended figure you run is decided by the sales mix, which is why a menu is read as a whole. And the labour and rent shares are not costs of the dish — they are the dish’s share of costs the business carries whether that plate sells or not. That distinction is the hinge of everything below.

Why is “cheap” pricing reckless in Dubai specifically?

Because in Dubai three things stack on top of the plate that a street-price never sees, and each is larger here than in most markets I have operated in.

Occupancy cost. Rent is negotiated once and lived with for years, and the arithmetic runs backwards from the lease. A unit at AED 30,000 a month needs AED 250,000 of monthly revenue to sit at the 12% ceiling and AED 500,000 to sit at the 6% floor. Price the menu below what that revenue requires and no roster, supplier or marketing plan rescues the model — you have priced the restaurant out of its own rent. The rent-versus-revenue check does that sum for any lease; run it before the menu, not after.

Aggregator commission, on top. Delivery is a large share of Dubai trade, and the platforms take a commission off the top of every order — as a teaching band, between 15% and 30% of order value depending on tier, delivery mode and the extras stacked on, as the aggregator economics read walks through. A dish priced for the table and listed at the same price on the app has given away a quarter of itself before food cost is paid. Our rescue check reads a per-order commission above 25% as structural for exactly this reason.

VAT, on the price. The UAE levies VAT at 5% at the point of sale, and the business collects and accounts for it on the government’s behalf. It is never revenue. Every percentage here is read against the net price; an owner who prices to a tidy VAT-inclusive number and reads costs against the gross has flattered every band by five points of denominator. Small per plate, real across a year.

Then a fourth thing, not a cost but a reflex: the discount habit on the platforms. The apps reward promotions with visibility, and a launch that opens with 20% off “to build volume” trains three parties at once — the guest to wait for the offer, the platform to expect the co-funding, and the owner to read a busy screen as a healthy one. Volume bought at negative contribution is not volume. It is a marketing spend without a budget or an end date — the per-order figure is below.

What are the three pricing mistakes that close restaurants?

The menus that close share three habits, and none of them look reckless on the day.

Mistake one: pricing to the neighbour. The place next door charges AED 55 for a similar plate, so you charge AED 55. What you have copied is their price. What you have not copied is their rent, their year of the lease, their recipe, their labour model — or whether they are losing money. In my experience the restaurant you are matching is often in its own quiet trouble and does not know it yet. You have imported a model without its margin, and you did it on a walk.

Mistake two: discount-led launches and the promotion reflex. Opening at 20% off, “buy one get one” on the apps, the weekday deal that was supposed to be temporary. Each is a pricing decision dressed as a marketing decision, taken without the post-commission arithmetic on the table. A promotion can be legitimate — briefly, with a budget, a target and a date it ends. Standing promotions are how a restaurant becomes a charity with a kitchen.

Mistake three: round-number, VAT-blind menus. The cost-built price was AED 63. That “looks odd”, so it becomes AED 60, or AED 55 because it reads better on the board — and the owner forgets that the 5% is inside that number. Every dirham rounded off comes straight out of contribution, because nothing in the kitchen got cheaper when the board got tidier.

None of the three starts at the plate. Each starts at somebody else’s number — the neighbour’s, the platform’s, the eye’s — and works backwards to a story about why it is fine.

The same dish, priced three ways — what is left after food, labour, commission and VAT?

Here is the arithmetic that ends the argument in most rooms I sit in. One dish, the AED 18 plate from above, priced three ways. Labour and packaging are held as fixed dirhams per plate — cutting the price did not cut the cook’s wage — and every line is illustrative, chosen for checkable round numbers. Rent, utilities and overheads sit below all three columns equally.

Line	Cost-built, dine-in	Priced to the neighbour, dine-in	Discount-led, on the app
What the guest pays (VAT-inclusive)	63.00	55.00	50.40 (listed at 63, 20% promotion)
VAT remitted (5%)	−3.00	−2.62	−2.40
Net price	60.00	52.38	48.00
Aggregator commission (25% of net, owner-funded promotion)	0.00	0.00	−12.00
Packaging	0.00	0.00	−3.50
Recipe cost	−18.00	−18.00	−18.00
Recipe cost ÷ net price	30%	past the 32% ceiling	past the ceiling
Labour share (28% of the cost-built net, held in dirhams)	−16.80	−16.80	−16.80
Left for rent, overheads and profit	25.20	17.58	−2.30
As a share of net	42.0%	33.6%	−4.8%

Check each column. Cost-built: 60 − 18 − 16.80 = 25.20. Neighbour’s price: 55 ÷ 1.05 = 52.38 net; 52.38 − 18 − 16.80 = 17.58. Discount-led: 63 × 0.80 = 50.40 paid; 50.40 ÷ 1.05 = 48.00 net; 48 − 12 − 3.50 − 18 − 16.80 = −2.30.

Read the middle column first, because it is the “sensible” one. The guest saved eight dirhams. The restaurant lost AED 7.62 of contribution — 30% of what the plate was leaving — and nothing about the plate changed. That is the entire cost of pricing on a walk. Then the right-hand column, the busy one. Every order the screen lights up with hands back AED 2.30 before the landlord is paid. The room is full. The month is a loss. And the owner, watching the order count climb, thinks the launch is working.

One more number, because it answers the fear that keeps owners from correcting their prices. At AED 25.20 a plate against AED 17.58, the cost-built price banks the same contribution at roughly seven plates in ten (17.58 ÷ 25.20 ≈ 0.70). You could lose three guests in ten to the correct price and be no worse off. Most menus I have re-priced from cost lost far fewer, because guests buy value and read a confident menu as a confident kitchen.

How do you price a menu that lasts?

Four disciplines. None is clever; all are rarely done.

Build a price ladder, not a price list. A menu needs a deliberate spread — an entry price that welcomes, a mid-band where most of the volume lives, a top where the anchors sit — with real gaps between the rungs. The entry rung is the one the neighbour comparison sees; the top rung is the one that makes the middle feel reasonable. A menu clustered tightly around one number has no shape, and a menu with no shape is priced by the guest’s mood.

Anchor from the top. Two or three dishes priced honestly at the upper end — signature plates with the cost to justify it — set the frame in which everything else is read. Remove them and the whole menu feels expensive; keep them and the mid-band feels fair. That is not a trick; a kitchen makes plates of different worth.

Read stars and plowhorses before you touch a price. Menu engineering puts every item on two axes — popularity and contribution margin in dirhams, not percentages — and sorts the menu into stars (protect and feature), plowhorses (popular but thin — the reprice candidates), puzzles (profitable but slow — reposition) and dogs (rework or remove). Our Menu Engineering Matrix sets the profitability line at the volume-weighted average contribution — so a high-selling thin dish cannot drag the bar down to meet itself — and the popularity line at 70% of an equal share, and caps the “recoverable” reprice on a plowhorse at 6% of price, the honest size of a single step. Price the plowhorses up in small moves, leave the stars alone, and stop subsidising the dogs.

Price the channel, not just the dish. A delivery order carries commission and packaging the dine-in price was never built for, so the delivery menu is a different menu. Work it backwards from the contribution you want. To leave the same AED 25.20 on the app that the cost-built plate leaves at the table — at 25% commission and AED 3.50 of packaging — the net price has to be AED 84.67, which is AED 88.90 on the app (0.75 × 84.67 − 3.50 − 18 − 16.80 = 25.20). You may decide not to charge that; most operators accept a lower delivery contribution for the reach. But decide it, write the number down, and read it weekly on the Delivery Margin Recovery tool — a contribution you have not chosen is one you are not measuring, and the Aggregator Commission Tracker will show it is lower than the contract says once ads, promotions and refunds stack on.

When should you raise prices, and how?

Prices should move when costs move, in the same week. That sounds obvious and almost nobody does it. A supplier letter lands, the recipe cost of the AED 18 plate becomes AED 19.50, and the menu stays at AED 63 until the next print run. The cost-built net price for that plate is now AED 65 (19.50 ÷ 0.30), AED 68.25 on the board; held at 60, the plate has drifted past the food-cost ceiling with no decision taken. A menu that is not re-costed has repriced itself downward, quietly, on your behalf.

So the first trigger is recipe cost movement, read weekly from the costed cards — the food-cost control discipline. The second is the quarterly review with real sales data, when the plowhorses are reclassified and the ladder re-checked against the mix. The third is the lease: a rent step-up is a pricing event, because the occupancy share just moved without asking you. The fourth is the platform statement: when the effective take rises — expired introductory terms, a promotion that stopped being temporary — the delivery menu moves, not the dine-in one.

As for how: dish by dish, never across the board. A flat “raise everything 5%” punishes the stars that were priced correctly and under-corrects the plowhorses that were not. Small steps — the 6%-of-price cap above is a sound discipline for a single move — on the items whose margin is thin and whose demand is strong. Change the plate when the price moves if you can, so the guest meets a deliberate dish rather than a dearer one. And never apologise on the menu. A price is a statement about what a plate is worth; a footnote about “rising costs” says you are not sure.

I will end with the two documents that never met. The menu is the most powerful pricing instrument the business owns, and most owners hand it to the street, the platform and the typesetter. Take it back. Start at the plate, build upward through the bands, add the 5% the government is owed, price the channel for what the channel costs, and read the result weekly against the four lines. That is cost-optimised pricing. It is not the dearest menu on the street and it is certainly not the cheapest. It is the one still open in five years — and if your month is already short with a full room, the audit shows in two minutes which line the pricing leaks through, and a turnaround starts from there.