Dark kitchen or physical restaurant: which works by numbers (checklist 2026)

The straight answer: which model works based on your numbers

Dark kitchen makes sense if your food margin exceeds 38% and delivery already accounts for more than 40% of current sales; physical location works if you have dining experience, prime cost under 65%, and repeat customers. Most owners choose by trend, not data. Since 2023, operators ask whether dark kitchen is «the future». It is not. It is a different operational structure with cash flows that bear little resemblance to a physical restaurant. Diego Parra has audited 340 virtual brand operations across 12 countries. The errors he sees repeat: choosing wrong costs 18 to 24 months and $60,000 to $120,000 burned before realizing it was the wrong choice.

Food margin: the metric almost everyone ignores

Dark kitchen requires food margin between 38% and 42% because platforms charge 30% to 35% in commissions and delivery fees. Physical location survives on 26% to 32% because there is no platform commission. If your recipe yields 28%, physical works; dark kitchen does not. That is the first line of the checklist: measure the real margin on each dish, not «pure cost» but net cost after waste, adjustments and shrinkage. According to Statista, online food delivery revenue reached $353 billion in the U.S. in 2024, but that figure concentrates among operators with margins of 38% or higher. Dark kitchens below 36% margin do not survive past 14 months.

Customer retention: physical vs. dark kitchen (a chasm)

Dark kitchen sells novelty: new concept, high rating, appealing photo on app. The customer does not return because there is no «place» to go back to, only an app. They order and forget. Physical sells place: people return for the table, the waiter, the atmosphere, the three-year-old broken mirror that never got fixed. Between 40% and 50% of physical restaurant customers repeat; dark kitchen barely 12% to 18%. This means physical requires less marketing spend to retain; dark kitchen burns cash in constant acquisition. Rappi generates approximately $800 million annually, but its customer base repeats 1.5 to 2.3 times per quarter. An average physical restaurant repeats 4 to 6 times.

The top 5 failures and their cost in dollars

First: confusing gross margin with net margin. Failure costs 15% to 25% in real profitability. Second: failing to audit prime cost (payroll, rent, utilities). Failure breaks your breakeven. Third: raising price in dark kitchen. The platform punishes you with algorithm; you fall in rankings. Lose 20% to 35% volume in 60 days. Fourth: not measuring repeat rate. Failure means you underbud and spend 40% more on marketing than planned. Fifth: launching dark kitchen without prior delivery experience. Failure in operations, quality, timing. You close in 10 to 18 months. There is $19.3 billion annually in delivery spending in Southeast Asia (2024), but 73% of new virtual brands disappear in year one due to these five mistakes.

How to audit your current structure before deciding?

Measure four numbers today: food margin per dish (photograph the receipt, subtract cost), percentage of sales from delivery (app plus phone), prime cost as % of total sales, and repeat rate (customers who purchase 2+ times in 90 days).

Document each in a spreadsheet. If margin > 38% and delivery > 40%, dark kitchen is viable. If margin < 32% or delivery < 25%, physical is your only option. If prime cost > 68%, adjust before any move. There are 7,606 ghost kitchens active in the U.S. per OysterLink (2025); 68% meet these baseline criteria. Those that do not, have closed.

Implementing the checklist in actual operations (who, when, how often)

The owner or controller measures these numbers every Friday at week close, 15 minutes maximum, using a Google Sheets template. Enter: margin per dish (3 test recipes), % delivery (sum of receipts), prime cost (last week), customer repeats yes/no (phone numbers or app IDs). Record each Friday. Review trend each month: if margin drops 2 points two months running, adjust recipe or price. If prime cost rises, audit payroll. If repeat falls below 10%, quality or service issue. The decision between dark kitchen and physical is not made in a single meeting: it is made with 12 to 16 weeks of data. Diego Parra has seen operators decide in 3 weeks and burn $85,000 in under 8 months because of it.

Auditing compliance: what evidence to measure and when

Each checklist line has measurable evidence. Food margin: photograph the cost sheet or invoices against written recipe. % delivery: weekly report from platforms (Rappi, DoorDash, etc.) versus total cash, with screenshot. Prime cost: payroll plus rent plus utilities divided by total revenue, spreadsheet. Repeat: export customer database (purchases last 90 days) and identify duplicates by name or phone. Each Friday, before measurement, ask: is there a shift greater than 3 points from the previous week in any metric? If yes, investigate immediately. Per Momentum Works, delivery operations in Southeast Asia grew 13% in 2024, but 44% of new operations closed from ignoring these metrics in month one.

When to switch models (and how to measure if it is time)?

You switch to dark kitchen when: food margin runs 8 consecutive weeks above 38%, delivery already represents 35% to 45% of revenue, prime cost is controlled (under 66%), and kitchen staff are trained in low-contact prep.

You switch to physical when: you have dining room management experience, audited prime cost confirms 60% to 64%, loyal customer base buying 2+ times monthly. Never switch on impulse. The right choice follows these numbers; everything else is noise. Ghost kitchens in China exceed 3,200 installations, but 62% are secondary operations from brands that already had physical locations. Those that started as pure dark kitchen without prior customer data take 3 to 4 years to real profitability.

5 differences nobody mentions

**Food margin:** Dark kitchen NEEDS 38–42% because platforms take 30–35%. Physical can live on 26–32% because no commission hit. If your recipe gives 28%, physical works; dark kitchen does NOT. **Loyalty and retention:** Dark Kitchen sells NOVELTY (high rating, pretty photo, trending). Customer doesn't return because there's no «place» to return to, just an app. Physical sells PLACE: people come back for the table, server, vibe. 40–50% repeat vs. 12–18%. **Price control:** Physical = you raise price, customer pays or leaves. Dark Kitchen = platform punishes you with algorithm; customer sees competitors 2 clicks away. Less margin to move. **Real scalability:** Physical: second location = USD 40–60k new capital + 4–6 months setup. Dark Kitchen: second meal shift in same kitchen = USD 2–3k + 2 weeks. Scale is different, financing is different. **Concentrated risk:** Physical: bad location hurts for years. Dark Kitchen: Rappi ban or algorithm shift hurts in WEEKS. Risk is sharper, less forgiving.