Physical restaurant vs dark kitchen: the guide that actually brings 2026 numbers

Step 1: calculate your retention rate per channel before signing anything

Before picking a format, work out how many cents of every dollar billed actually stay with you in each channel, because that number decides the model rather than the other way around. The math fits on a napkin: in a dining room you keep 100 cents of the ticket and the guest pays the tip on top; in pure delivery, with aggregator commissions running 15% to 30%, you keep between 70 and 85. The owner who wrote to me in March averaged 41 orders a day at 22 dollars with a 29% food cost, and his aggregator took 27 points plus 4 more in packaging: operating profit of 3.8% against the 11% his old dining room used to deliver. The DELIVERABLE here is one sheet with three cells per channel — retention, food cost, packaging — and a contribution margin per order. Verify it by matching last month's actual aggregator settlement against your gross sales report; if they diverge by more than two points, your sheet is the one lying.

Step 2: set the real break-even volume, not the one you were sold

A dark kitchen needs roughly 1.6 times the volume of a comparable dining room to leave the same profit, and that multiplier is the figure you should write in large letters before signing any shared-kitchen contract. It falls out of simple arithmetic: at 27% commission and 30% food cost you are left with 43 points to cover payroll, packaging, rent and profit, while the dining room starts with 70 points available. Divide 70 by 43 and you get 1.63. If your dining room billed 40,000 dollars a month, the ghost kitchen needs to hit around 65,000 to match cash, not the 40,000 the aggregator's salesperson painted for you. The deliverable is a monthly order target written on the kitchen whiteboard, with the required average ticket beside it. Verify it on day 30: if the order count sits below 80% of that goal, the model is not slow, it is badly sized.

Step 3: size the startup capital against the payback period

The dark kitchen wins on tied-up capital by a margin no dining room can argue with: setting up a station in a shared kitchen runs between 12,000 and 40,000 dollars, against the 250,000 to 900,000 of a full location with seating, according to the ranges the National Restaurant Association publishes. That much the 2020 foodtech pitch got right. Where it went wrong was in keeping quiet that cheap capital buys a thin-margin business, while expensive capital buys an ASSET. Your decision hangs on the timeline, not the amount: at 30,000 dollars and a 4% operating profit, you recover in about 19 months selling 40,000 monthly; at 400,000 dollars and 11% profit, you recover in 91 months. The deliverable is a two-column table showing investment, expected profit and months to payback. Verify it by having your accountant rebuild it with your market's numbers, not the article's.

Step 4: treat SERVICE as a revenue line, not a payroll expense

When you close the dining room you do not remove a cost, you remove the only channel where someone can raise the ticket while standing in front of the guest. Diego F. Parra hammers this in every Masterestaurant audit: a server trained in suggestive selling moves the ticket close to 18%, recovers the annoyed guest before the review gets written, and reads the table. In a dark kitchen the nearest substitute is a packaging protocol and an automated message, which help, but they do not sell dessert. Digital levers exist and they measure: a full digital offer of menu, ordering and payment lifts the ticket 20% to 30% per Sunday (QR Code Ordering 2025), and menu psychology adds 15% or more without touching prices per NeatMenu (Menu Psychology 2026) — yet both apply to either format. The deliverable is your average ticket measured by channel over four weeks. Verify it by reading the gap: if the dining room does not beat delivery by at least 12%, your service is untrained.

Step 5: quantify the hidden variable costs nobody puts in the pitch

Delivery packaging weighs between 3.5% and 6% of sales and climbs every time cardboard climbs, whereas a cloth napkin and a porcelain plate are assets rather than consumables, so they never behave that way. On 40,000 dollars a month, six points of packaging come to 2,400 dollars that stay in the till in a dining room, roughly 28,800 a year. Add the payroll pressure: base hourly pay in United States restaurants rose 4% to 14.20 dollars in 2024 according to 7shifts, and that pressure hits both models equally, though the dining room offsets it with tips and delivery does not. The deliverable is a new, separate line in your P&L called packaging, never blended into food purchases. Verify it by adding up the month's packaging supplier invoices and dividing by net sales; if the number surprises you, you had been paying it for months without seeing it.

Step 6: decide the mix and write the channel contract

My position is firm and the numbers hold it up: the dining room wins on margin per ticket and on brand lifespan, the dark kitchen wins on speed to launch, and the expensive mistake is treating them as mutually exclusive when the mix usually outperforms either one alone. A dining room already paying rent can run delivery out of the same kitchen during off-peak hours and absorb 27% commission on incremental volume, because the fixed cost is already covered by the seated shift. The pure dark kitchen has no such cushion. Demand data belongs in this decision: 64% of households above 200,000 dollars a year eat out weekly against 42% of those below 50,000, per Morning Consult 2025, and that upper segment is what fills dining rooms. The deliverable is a written rule stating which dish sells in which channel and at what price. Verify it when your cook can recite it without looking at the page.

The five mistakes that sink the numbers while executing this change

The most repeated error is charging the same price in the dining room and on the aggregator, because whoever does that hands over 15 to 30 points of margin on every digital order without noticing; the delivery price has to absorb the commission, full stop. Second: counting the aggregator's gross sales as income and discovering the commission when the biweekly settlement lands. Third: failing to track packaging separately, which we already saw comes to 2,400 dollars a month on 40,000 in sales. Fourth: launching three virtual brands out of a one-cook kitchen, which blows up prep times and drops the rating below 4.5, the point where the aggregator's algorithm stops showing you. Fifth, and costliest, is closing the dining room to «cut costs» without ever measuring the ticket gap between channels. What happens if your rating slides to 4.2 during a peak month?

The five mistakes that sink the numbers while executing this change — in practice

You lose placement, you buy ads inside the aggregator to win it back, and you stack another five points of effective commission onto a margin already sitting at 3.8%. You will know the guide has been executed when you can answer six questions with a number instead of a hunch, and all six live on a single sheet you refresh every Monday. One: retention per channel, expressed in cents per dollar and checked against the aggregator's real settlement. Two: your volume multiplier, which in most accounts lands near 1.6. Three: months to recover the investment, calculated with your actual profit rather than the projected one. Four: dining-room average ticket against delivery ticket, with a minimum gap of 12%. Five: packaging as a percentage of sales, on its own P&L line, between 3.5% and 6%. Six: differentiated pricing by channel, written down and applied in the digital menu.

Closing checklist: how to know it came out right

If any of the six boxes sits empty or lives in somebody's head, go back to that step before you sign a shared-kitchen contract. Tomorrow, open last month's settlement and calculate the first one. REVENUE STRUCTURE. On the floor, every dollar billed arrives whole and the tip sits on top of the check; in pure delivery, 15 to 30 cents of every dollar leave as commission before food cost touches anything. At 27% commission and 30% food cost you keep 43 points to cover payroll, packaging, rent and profit — the dining room starts at 70. THE HIDDEN VARIABLE COST. Delivery packaging runs 3.5% to 6% of sales and climbs with the price of board, while a napkin and a porcelain plate behave differently because they are an asset, not a consumable. On 40,000 dollars of monthly sales, six points of packaging equal 2,400 dollars that a dining room simply keeps.

The four differences that decide the math

CHECK LEVERAGE. A trained server working with a simulator and an automated preshift moves the average check 18% within eight weeks; the best in-app upsell I have measured with serious clients reaches 7%. Applied to 900 monthly checks of 20 dollars, that gap is 1,980 dollars of extra sales per month — roughly the rent of a station in a shared kitchen. CONCENTRATION RISK. Once a single app carries 70% of your sales, you no longer own the business: the ranking algorithm does. A three-point commission increase, decided in another city and announced by email, erases half your operating profit and there is nothing you can do about it that week.