How to start a cloud kitchen

The steps, in order:

1. Choose the format. One brand or several, and a short menu built to travel.
2. Set the budget. Roughly ₹6–15 lakh for one brand, with about 90 days of working capital held back.
3. Find the kitchen. Inside the delivery radius you want, off the expensive high street.
4. File the licences the week you sign the lease. FSSAI, the municipal trade licence, GST and Shops and Establishments.
5. Fit out and test. Equipment against a frozen layout, and every dish travel-tested in its real container.
6. List and launch. Onboard on the apps once the FSSAI number arrives, and print your direct number on every box.
7. Close every night. Tie the day’s orders, payouts and stock together.

Each step has its section below, and the 45-day plan at the end puts them on a calendar.

1Cloud kitchen or dining room?

A cloud kitchen removes the two most expensive things about a restaurant: the dining room, and the months you wait for it to fill. No high-street frontage, no interiors, no floor staff, no 60-cover fit-out. You keep the kitchen. You take on delivery economics instead, and that trade shapes the rest of this guide.

Read the last row twice. A cloud kitchen does not escape the cost of reaching guests; it turns that cost from fixed rent into a share of every order. A dining room’s rent is due whether the room fills or not; the apps take their share of every order, profitable or not. Kitchens that work treat that line as the enemy from day one.

2What does a cloud kitchen cost?

Rough ranges for one single-brand kitchen in 2026, for planning only. Quotes vary widely by city, by menu and by how much refurbished equipment you accept. Get three quotes for every line, and read the top of each range as a metro number.

Every line at its low end adds up to about ₹5.9 lakh; every line at its high end, about ₹17.2 lakh. Few kitchens hit every extreme, which is why roughly ₹6–15 lakh is the honest planning range.

Three rules survive every budget. The deposit is not the budget: a ₹40,000-a-month kitchen with a six-month deposit and a month in advance takes ₹2.8 lakh before you own a burner. Hold the working capital back: a kitchen that runs out of cash in month two closes however good the food is. And do not save on the photos. Guests cannot smell your kitchen; they can only see it.

3Which licences does a cloud kitchen need?

The list is shorter than a dining room’s: no liquor, no music licence, and usually no signage permission or eating-house licence. The core four:

1. FSSAI. Since 1 Apr 2026, registration covers turnover up to ₹1.5 crore a year and a state licence covers turnover above that, up to ₹50 crore, so many new kitchens start on registration. You apply online on FoSCoS, FSSAI’s portal at foscos.fssai.gov.in, and each premises needs its own registration or licence: a second kitchen at a new address is a new application. Delivery apps list only kitchens with a valid FSSAI licence or registration, so this sets your launch date. File it the week you sign the lease; the FSSAI guide walks through the application.
2. The municipal trade licence, from your city corporation. Timelines vary by city; start it alongside FSSAI, never after.
3. GST registration, on the GST portal, gst.gov.in. Compulsory once turnover crosses ₹20 lakh a year (₹10 lakh in the special-category states). On food ordered through the apps, the app collects and pays the 5% GST under section 9(5) of the CGST Act, as it has since 1 Jan 2022. Whether a kitchen that sells only through the apps must register below the threshold is read in more than one way, so settle it with your CA before you list.
4. Shops and Establishments registration, the state labour registration, needed once you hire and often for the current account.

A fire NOC may apply depending on the premises: commercial kitchens with a large LPG bank draw attention, so ask before the last tenant’s history becomes your problem. The full list, issuer by issuer, is in the licences guide.

Two rules apply from your first bill. Since 1 Jan 2022, FSSAI has required your licence or registration number on every receipt, invoice, cash memo and bill, so put it on your bill format before you open. And the orders you sell yourself, for pickup or by your own delivery, are restaurant service under GST: the CBIC’s Circular 164/20/2021-GST of 6 Oct 2021 says this covers cloud kitchens and central kitchens, at 5% GST without input tax credit. The GST guide has the rest.

4Where should the kitchen be?

You are not buying footfall, so stop scouting like a dining room. You are buying the homes and offices inside the apps’ delivery radius, divided by the rent it costs to reach them. In practice that means a back lane or a first floor a few hundred metres from the expensive market: the same radius at a fraction of the rent.

The checklist: enough homes and offices inside the radius; rent inside your budget, because high-street rent for a kitchen nobody visits is the classic first mistake; power and water, promised in writing; ventilation, and neighbours who will tolerate a tandoor at 11 pm; and easy access for riders, because slow pickups make your preparation times look bad to the apps, and your ranking pays for it.

One more test first-timers often skip: open the apps at your shortlisted address at 8 pm and count the kitchens in your cuisine. A crowded zone means the demand exists, and that you will need a sharper brand to rank in it.

5What will the delivery apps take from each order?

More than the commission line in the contract. Start with the base commission, which is negotiated restaurant by restaurant rather than published as one rate. Add the payment charges, the 18% GST on the app’s fees to you, the ads you buy for visibility, and your share of the discounts the app runs. The full fee anatomy, and the ways to claw some of it back, are in the commission guide.

Run your own contract through this calculator before you sign the onboarding forms, not after:

With the calculator’s starting figures, a 24% commission, 5% in ads and no funded discount, the deduction comes to about 36% of each order once the 1.8% payment fee and the 18% GST on the commission, the ads and that fee are added: on a ₹450 order, about ₹286 reaches you. A discount you fund is counted as it is, with no GST on top. Your contract will differ, which is the point of running it.

The answer is channels, not complaints. ONDC, the open network, sets its commission through the buyer and seller apps on each side of an order rather than through one platform; treat it as an extra channel, not a replacement for the big apps’ volume. Direct orders, by phone for pickup or through your own delivery, carry no platform commission at all, so every regular you move there is one of the highest-margin changes a cloud kitchen can make. Print the phone number on every box from the first order: the guest already knows your food, and the box is the only advertising space you own.

6How does a multi-brand kitchen work?

One kitchen, one team, one stock room, several brands on the apps. A biryani brand, a curry-bowl brand and a rolls brand can share a hot line, much of a purchase list and one evening’s prep. Each brand gets its own listing, its own photos and its own shot at the category rankings, so the same fixed cost is spread over more orders. It is how many profitable cloud kitchens are built, and how the format reaches profit at volumes one menu would not.

The trap is accounting, not cooking. When three brands draw on one stock of chicken, rice and oil, owners lose sight of which brand makes money, and a brand that loses ₹15 an order scales its losses beautifully. The fix is costing by recipe and by brand: every dish mapped to its ingredients, every order drawing down the shared stock, every brand with its own margin. Whatever system you choose, check that it shows food cost brand by brand before you design brand number two.

7Packaging, ratings and the wrong-item tax

Your packaging is your dining room: it is the whole physical experience of your brand. Leaking dal and soggy rolls do not get second chances; they get one-star ratings, and ratings drive the ranking your break-even depends on. Travel-test every dish in its real container for 30 minutes on a bike before it goes on the menu. If it does not survive, change the dish or the container, not the review.

Then there is the wrong-item tax: missing and wrong items are refunded out of your payout, and repeat offenders slide down the rankings. The fix is dull process: a packing checklist read against the kitchen ticket, one person sealing every order, and a kitchen screen instead of paper slips, so nothing is lost between fryer and bag. Dull is the point. The kitchens that last make fewer unforced errors per hundred orders.

8When does a cloud kitchen break even?

The model: fixed costs (rent, salaries, subscriptions, the ads budget) divided by what each order really contributes after food, packaging and the apps’ share. Two things surprise first-timers: how hard the apps’ share squeezes each order, and how fast the answer improves as direct orders grow. In the calculator, enter your average order as the spend per cover, and your gross margin after the apps’ share, not before:

A worked example, with assumed figures: fixed costs of ₹1,90,000 a month, a ₹350 average order, food and packaging at 32% of the order, and the apps keeping 30%. Each app order leaves 38%, or ₹133, so you need about 1,429 orders a month, 48 a day. Move 20% of the orders to direct and the blended margin rises to 44%: 42 orders a day. At 40% direct it is 50%: 37 a day. Direct orders still cost packaging, and a rider if you deliver, but they carry no commission.

Model three cases before you commit the money: every order through the apps, 20% direct, and 40% direct. The gap between the first and the third is often the gap between a kitchen that grinds and one that grows. Put the result straight into your business plan, because a lender or partner will ask for exactly this page, and check it against what margins look like by format.

9Where CountStand fits, and where it does not yet

Honestly: a kitchen that takes nearly all of its orders through the apps will get less from CountStand at launch than a dining room or a counter will. CountStand bills the orders you take yourself, including phone orders you bill for pickup, runs the kitchen display and counts stock by recipe. Delivery-app orders, the apps’ payout reconciliation and online ordering are not open at launch.

If your plan has a pickup window, or you are building direct orders from the first day, the cloud kitchen software page has the detail.

10A 45-day launch plan

Days 1–7: paper. Lease signed and registered; FSSAI, the trade licence, GST (if it applies) and Shops and Establishments all filed; the kitchen layout frozen; the menu built for travel, not for a table: short, high-margin, every dish container-tested.

Days 8–21: build. Equipment ordered against the frozen layout; branding and photos done; packaging bought after the travel tests; billing, the kitchen screen and stock set up with every recipe costed, and your FSSAI number on the bill format; app onboarding submitted the moment the FSSAI number arrives.

Days 22–35: people and process. Cooks and a packer hired and trained on the real systems; recipes written down with yields; two suppliers for every critical ingredient; test orders run end to end, from order to kitchen to sealed bag.

Days 36–45: launch. Listings live; the launch ads budget spent on the first ratings; a card with your direct number in every bag from day one; and a close every night that ties orders, payouts and stock together. In this business, the owner who knows the numbers every night beats the one who finds out every month.

11Questions owners ask