Ghost kitchens have the worst per-order economics of any restaurant model — and also the lowest fixed costs. There's no dining room, no front-of-house labor, no beverage program, no host stand. The kitchen runs lean; the labor is back-of-house only. That structural efficiency is what makes the model viable in spite of the 18–30% marketplace commissions that eat into every order.
The first-order math operators get wrong: 'we'll just put up more brands and capture more orders.' Adding a brand doesn't automatically add margin. A burger brand at $14 ticket with 32% food cost and a 25% marketplace commission produces $4.20 of contribution per order. A wing brand at $22 ticket with 28% food cost and a 22% commission produces $11.00 of contribution. Both brands look the same on volume reports; the wing brand makes 2.6× the margin. Operators that don't run per-brand per-channel P&L don't know which of their brands are actually profitable and end up subsidising losers with winners.
Channel mix shift is the highest-leverage strategic move. A ghost kitchen that does 80% third-party and 20% direct loses more margin to commissions than a kitchen with the inverse split. Building direct ordering volume — through SEO on your branded sites, loyalty programs that incentivise repeat-direct, in-package marketing inserts pointing customers to your direct app — shifts the channel mix over time. Operators that grow direct to 35–45% of total volume see contribution margin improvements of 4–8%, which on a $1.2M kitchen is $48K–$96K per year.
The operational risk in a ghost kitchen is overcommitment during peak. When all 7 marketplaces are firing simultaneously plus your direct channel, the kitchen queue grows faster than the line can produce. The failure mode: accept all orders, miss promise times, get flagged on the marketplaces, refund ratio climbs, marketplace algorithm punishes you on visibility, volume drops. The right move is capacity throttling at the POS — pause new orders across all channels when queue depth exceeds threshold, then unpause when the line catches up. The kitchens that survive long-term run their capacity discipline as tightly as a Michelin restaurant runs its course pacing.