What is a ghost kitchen? The Zayos definition, and the three setups people mean by it
Published July 2026 · Updated July 26, 2026
A ghost kitchen is a commercial cooking space that sells food without a dining room. Every order arrives through delivery apps, a website, or the phone. Diners never walk in. The building is a production line, and the storefront is an app listing.
You will also hear dark kitchen, cloud kitchen and commissary kitchen. They describe the same idea with different accents and slightly different landlords.
Why does anyone cook without a dining room?
Because a dining room is the most expensive part of a restaurant and the delivery customer never uses it. Front retail rent, the build out, the bathrooms, the host, the servers and the bussers all exist to serve people at tables. Strip them out and you are left with the part that actually produces the food.
Every ghost kitchen decision comes back to one trade. You give up walk in discovery and walk in margin, and in exchange you get cheaper square footage and a smaller payroll. Whether that trade is good depends on one thing: how much of your demand you own before you sign the lease.
A ghost kitchen has no walk in trade to carry it through a bad delivery week. Every dollar has to be found, delivered and paid for.
What are the three kinds of ghost kitchen?
Most arguments about ghost kitchens are two people describing different buildings. There are three common setups and they carry completely different risk.
The shared commissary
A landlord builds one licensed facility and rents cooking bays by the month, usually with shared cold storage, dish and loading. You bring your menu and your staff. It is the fastest way into production because the hood, the grease trap and the health approvals already exist, and the most expensive per square foot for exactly that reason.
Best fit
New concepts testing a market without signing a ten year lease.
The operator owned dark kitchen
A restaurant company leases cheap industrial or back street space and fits it out itself, purely for delivery. No front of house, no prime retail rent, full control of the build and full exposure to the build cost and the permits. This is the version most people picture when they hear the term.
Best fit
Groups with an existing brand and enough delivery volume to fill a second production site.
The host kitchen
A normal restaurant with a dining room uses its idle kitchen capacity to cook delivery only menus, either its own or someone else licensed to it. Nothing new is built. This is the quietest and by far the most common version, and it is why the count of ghost kitchens is impossible to measure honestly.
Best fit
Existing restaurants with a slow 2pm to 5pm and staff already on the clock.
What does a ghost kitchen shift actually look like?
Prep runs the same as any kitchen. The difference starts when service does. There is no host reading the room and no server absorbing a delay with a free bread basket. The pass talks to a screen, and the screen talks to couriers who are not your employees and do not run on your schedule.
The three failure points are all timing. Food finished before a driver arrives sits and gets worse. Drivers arriving before food is finished stack up in a doorway. And a channel that keeps accepting orders when the line is twenty tickets deep turns a busy hour into a bad rating.
The practical fix is boring and it works: one screen showing every channel in one queue, hot holding sized for the worst ten minutes of the night, and a hard rule that nothing gets bagged until a courier is assigned. Operators running several channels usually consolidate them through a single tablet, which is the same setup described on the Otter integration page.
Why did so many ghost kitchens close?
The model expanded hard when dining rooms were shut, and several of the largest shared kitchen networks pulled back or closed sites in the years after. The pattern behind those closures is worth understanding because it is repeatable and avoidable.
A speculative ghost kitchen starts with no brand, no customer list and no street presence, so the only source of demand is a marketplace. Buying every order from a marketplace at a blended 25 to 35 percent is survivable as a top up on an existing business. As the sole channel, it means a quarter to a third of gross sales leaves before food cost, and the operator has no way to contact a single customer to bring them back cheaper the second time.
When a marketplace is the only channel, it is not a sales channel. It is the landlord of your demand, and the rent is 25 to 35 percent of everything.
Ghost kitchens attached to an existing restaurant brand fared better for a simple reason: they already had people who knew the name and would order direct if asked. Numbers on what the marketplace stack actually costs are on how much DoorDash charges restaurants and in the commission statistics.
What do people get wrong about ghost kitchens?
A ghost kitchen is not a business model
It is a real estate decision. Removing the dining room changes your rent, your labor mix and your marketing, but it does not by itself make anything profitable. The menu, the food cost and the channel mix decide that, exactly as in a restaurant with tables.
It is not the same as a virtual brand
A ghost kitchen is a building. A virtual brand is a name and a menu. You can run five virtual brands out of a dining room restaurant and never touch a ghost kitchen, and you can run one single brand out of a ghost kitchen for years.
Cheaper rent does not mean cheaper operating
Delivery only means every order carries packaging, a courier and a platform fee. A dine in restaurant pays more rent and then serves plenty of orders that cost nothing to deliver. Rent gets cheaper. Distribution gets much more expensive.
Nobody is walking past your sign
A dining room generates its own discovery. A ghost kitchen has no window, no patio and no smell on the sidewalk. Every diner has to be found, which means marketplace placement, search, and repeat buyers you have a way of contacting.
Where Zayos fits
Zayos gives a kitchen a direct ordering channel it owns, which is the thing a delivery only operation is usually missing. Plans are $499, $599 and $699 per location per month, month to month, no setup fee. The restaurant keeps 100 percent of food revenue and 100 percent of tips, and the only per order charge is the service fee the diner pays at checkout: $0.99 pickup, $2.99 delivery, $0 dine in, 10 percent on catering. For a delivery only kitchen, the point is having somewhere to send a repeat customer that does not charge a commission to serve them again.
The delivery only operator view is on ordering for ghost kitchens, plans are on pricing, and the setup timeline is on how it works.
Find out what your channels really cost
The free AI report runs your own numbers in about sixty seconds. No card required.
Ghost kitchen questions operators actually ask
Is a ghost kitchen cheaper to run than a normal restaurant?
What equipment does a ghost kitchen need that a restaurant does not?
Do ghost kitchens need a health inspection and a license?
Can I run a ghost kitchen out of my existing restaurant?
Do ghost kitchens still make sense in 2026?
Working out the numbers? Start with the free report at grader.zay-os.com.