How to start a ghost kitchen in 2026
To start a ghost kitchen, validate that real delivery demand exists in a specific postcode radius, pick a model (your own unit, a shared cloud-kitchen pod, or a virtual brand run from a kitchen you already operate), design a tight menu that survives a 25-minute drive, equip and staff for delivery-only throughput, decide whether you go on aggregators or build your own ordering, and stand up the four tech pieces every delivery-only restaurant needs: POS, kitchen display, dispatch, and a CRM. Done in that order, a focused single-brand launch is realistic in 6 to 10 weeks, and the first profitable month usually arrives after a deliberate 90-day ramp, not on opening night.
A ghost kitchen (also called a dark, cloud, or virtual kitchen) is a food-production site with no dining room and no walk-in counter. Every order arrives through a screen. That single fact changes almost every decision you make, so this guide walks the launch in the order the work actually happens.
Step 1: Validate demand and location before signing anything
The cheapest mistake to avoid is opening in the wrong place. Because there is no foot traffic to rescue you, location is purely about delivery economics: how many households sit inside a 20-to-30-minute drive, what they already order, and how crowded the category is. Spend a week reading the aggregator apps in your target area. Note which cuisines have thin coverage, which have hour-long wait times at peak, and what average ticket sizes look like.
Validate before you commit to a lease. A short pop-up from a rented commercial kitchen, or a few weeks selling a single hero dish, tells you more than any spreadsheet. If you cannot get 15 to 25 orders a day from a soft launch in a dense area, a bigger kitchen will not fix the underlying demand problem. For a deeper look at the numbers behind a single site, the dark kitchen unit economics breakdown is worth reading before you sign.
Step 2: Choose your model
There are three common ways to launch, and they differ mostly in capital and speed.
- Your own dedicated unit. Highest control and best long-term margin, but you carry the full build-out, equipment, and lease. Best when you have a validated concept and want to own the asset.
- A shared or cloud-kitchen facility. You rent a turnkey pod inside a multi-tenant building. Fastest to open and lowest upfront cash, at the cost of ongoing rent that bites into margin. Good for testing.
- A virtual brand from an existing kitchen. If you already run a restaurant, you launch a new delivery-only menu from the same line during off-peak hours. Near-zero incremental fixed cost.
These labels get blurred constantly. If the terminology trips you up, the difference between dark, ghost, cloud, and virtual kitchens explains where each one fits.
Step 3: Design a menu built for the drive, not the plate
This is where delivery-only operators most often go wrong. A menu that photographs well in a dining room can arrive soggy, leaked, or cold. Build the menu around travel resilience first.
- Cut the item count. A focused menu of 12 to 20 SKUs cooks faster, wastes less, and is easier to forecast than a sprawling one.
- Test every dish after a 25-minute hold in real packaging. Fries, crispy coatings, and anything with separate sauce are the usual failures. Either reformulate or drop them.
- Engineer for margin. Delivery commissions and packaging eat 20 to 35 percent of a typical aggregator order, so your food cost and pricing have to absorb that before you see a profit.
- Use the same base ingredients across several dishes to keep prep and stock simple.
If you are considering multiple concepts from one kitchen, the virtual brands strategy guide covers how to run several menus off a shared line without chaos.
Step 4: Equipment, layout, and staffing
Lay the kitchen out as a one-way line: receive, prep, cook, pack, hand-off. Buy equipment that matches your hero dishes rather than a generic full kitchen — a fryer-heavy concept and a wok concept need very different lines. Reserve a dedicated, well-lit packing station near the pickup door, because packing quality is what the customer actually experiences.
Staffing for delivery-only is leaner than dine-in. A small launch often runs with one or two cooks plus a packer at peak, and no front-of-house at all. Build the schedule around your real demand curve: most delivery volume clusters into lunch and dinner spikes, so part-time peak coverage usually beats full-day shifts.
Step 5: Aggregators vs your own ordering
Most ghost kitchens launch on the big marketplaces — Bolt Food, Glovo, Wolt, Uber Eats — because they bring instant demand. The trade-off is commission, typically in the 25 to 35 percent range, plus you never own the customer relationship. The marketplaces decide what data you see.
The healthier long-term play is to use aggregators for discovery while steadily pushing repeat customers to your own ordering channel, where you keep the full ticket and the customer data. In practice that means running both at once. A unified order board matters here: when phone, web, your own app, Telegram, and several aggregators all land in one place, your team isn't juggling a tablet per platform. Toster consolidates those channels — including Bolt Food, Glovo, Wolt, and Uber Eats — into a single board so nothing is missed at peak, and it can power your own branded ordering and delivery alongside the marketplaces.
Step 6: Packaging and food quality in transit
Packaging is part of the recipe in a delivery-only business. Vented containers for fried items, separate cups for sauces, tamper-evident seals, and a leak test on every new dish. Run a hold test: cook the dish, seal it, leave it 25 to 30 minutes, then open it and taste. If it fails, fix it before it ships. This single discipline protects your reviews more than almost anything else, and reviews are the lifeblood of marketplace ranking.
Step 7: The tech stack you actually need
Four systems carry a delivery-only operation, and trying to glue together separate vendors for each is where margin and sanity leak away.
- POS / order management to take and route every order. In a ghost kitchen the POS is your whole operation — there is no register at a counter. Toster is its own POS, so the order board, kitchen, and dispatch all run on one system rather than bolted-together tools.
- Kitchen display system to replace paper tickets, route items to the right station, and time courses so everything is ready together. A KDS with a packing photo check also gives you a record of what actually went out the door. See how a kitchen display system handles station routing.
- Dispatch. Decide early whether you ride aggregator couriers, run your own fleet, or mix both. Owning the last mile protects margin and the experience; courier management with live GPS and route optimization makes a small fleet viable.
- CRM and loyalty to turn first orders into repeat ones — the only sustainable path to profit when commissions are high.
Newer platforms also fold in AI tools like demand forecasting and an AI voice operator for phone orders, which help a lean launch team punch above its headcount.
Step 8: Fiscal and legal basics by country
Rules differ across Europe and you should confirm specifics with a local accountant, but the shape is consistent: register the business, secure a food-hygiene approval for the premises, and meet the fiscal receipt requirements of the country you operate in. Ukraine, Poland, Czechia, Germany, and Spain each have their own fiscalization regime, and your POS has to issue compliant receipts in each. A platform with built-in multi-country fiscal support saves a painful integration; the overview of food delivery fiscalization in Europe is a useful primer.
Step 9: A realistic first-90-days ramp
Do not expect opening week to be representative. Weeks 1 to 2 are about fixing the obvious: missing items, slow tickets, packaging failures. Weeks 3 to 6 are about marketplace ranking — consistency and good reviews push you up the listings, which is when order volume starts to compound. Weeks 7 to 12 are when you tune the menu against real sales data, drop the dead SKUs, and start shifting repeat customers to your own channel. Most well-run single units reach a stable, profitable rhythm somewhere in that window, not before.
Frequently asked questions
How much does it cost to start a ghost kitchen?
Ghost kitchen startup costs vary widely by model. Renting a pod in a shared cloud-kitchen facility is the cheapest entry and can start in the low tens of thousands of euros once you add equipment and working capital, while building your own dedicated unit costs considerably more. Launching a virtual brand from a kitchen you already run is the lowest-cost route of all. Software is a smaller line item — an all-in-one platform like Toster starts around €250 per month, far less than stitching separate POS, KDS, and dispatch tools together.
How long does it take to open a dark kitchen?
A focused single-brand launch is realistic in roughly 6 to 10 weeks if your concept is validated and you use a shared facility or an existing kitchen. A ground-up build with permits and a custom fit-out takes longer. The fastest path is a virtual brand layered onto a kitchen that is already operating.
Do I need to be on aggregators to launch a delivery-only restaurant?
Aggregators give you instant demand and discovery, so most operators start there. But the commissions are high and you don't own the customer, so the smart move is to run your own ordering channel in parallel and steadily move repeat customers onto it. Toster supports both at once from a single order board.
What is the difference between a ghost kitchen and a virtual brand?
A ghost kitchen is the physical, delivery-only production site. A virtual brand is a delivery-only menu and identity that exists only on the apps — it can run from a ghost kitchen or from an ordinary restaurant's kitchen. One ghost kitchen can host several virtual brands at once.