Cloud Kitchen Unit Economics in India: The Real Maths
By Hamza Ukani · · Updated
- cloud kitchen
- delivery
- unit economics
- Zomato
- Swiggy
A cloud kitchen looks like the cheapest way into the restaurant business. No dining room, no service staff, no prime-location rent. Then the first quarter’s numbers arrive and the revenue is healthy while the bank balance is not.
The reason is always the same: the dish was costed for a dining room and sold through a platform.
Work it out per order, not per month
Take one ₹400 menu item and follow it.
| Line | Amount |
|---|---|
| Menu price on the platform | ₹400 |
| Platform discount (you fund 50% of a 40% offer) | −₹80 |
| Commission (~22% of gross) | −₹88 |
| Payment gateway (~2%) | −₹8 |
| Net received | ₹224 |
| Food cost (28% of ₹400 menu price) | −₹112 |
| Packaging | −₹22 |
| Contribution | ₹90 |
That ₹90 has to cover rent, electricity, kitchen labour, marketing spend for placement, and wastage. At 40 orders a day the outlet earns ₹1,08,000 a month in contribution — and a 500 sq ft unit with three cooks will spend most of it.
The dish that shows 72% gross margin on the menu is delivering 22% after the platform has taken its share.
Your own commission and discount terms will differ — they vary by city, by contract and by how much you negotiate. Put your real numbers into the same table. The shape of the answer rarely changes.
The four numbers that decide whether it works
1. Discount funding. The single largest controllable cost, and the one owners agree to fastest because it is framed as visibility. Every percentage point of discount you fund comes out of contribution directly.
2. Effective commission. The headline rate is not the rate. Add the ad spend you buy to stay visible, and the real number is often five to eight points above the contract.
How these terms are set has been under scrutiny. The National Restaurant Association of India (NRAI) complained to the Competition Commission of India in 2022, and in November 2024 Reuters reported that the CCI’s investigation found Zomato had offered lower commissions in exchange for exclusivity and Swiggy had promised growth to restaurants that listed only with it. Both companies said no final order had been passed. Whatever the outcome, read the exclusivity and price-parity clauses in your contract before you sign — they decide whether you can sell cheaper on your own channel later.
3. Packaging. ₹18 to ₹30 for a main course done properly. It is a food cost line, not an overhead, and pretending otherwise flatters every dish on the menu.
4. Average order value. Contribution per order rises much faster than cost per order, so a menu built to push AOV from ₹320 to ₹450 through combos and sides changes the business more than any cost cutting.
Menu design for delivery is a different discipline
Price for the channel. A separate delivery price is not dishonest; the cost structure is genuinely different and every serious brand does it.
Build the menu around dishes that survive twenty-five minutes in a bag. Anything crisp, anything that steams itself soggy, anything plated to look good — those are dine-in dishes. Delivery favours saucy, robust, reheatable food and items that photograph well as a thumbnail.
Sell sides. A ₹60 side with a 75% margin on a third of your orders is a bigger profit lever than a ₹20 price rise on mains that suppresses conversion.
Build a direct channel, slowly
Every order through your own WhatsApp number or website keeps the commission. You will not replace the platforms — they are where new customers find you — but repeat customers are a different matter. A card in the bag with a direct-order number and a small reason to use it (a free side, not a discount on the whole bill) moves a share of regulars over time. Even 15% of orders going direct changes the contribution maths noticeably.
Check your platform contract first: some agreements restrict price differences between channels.
Multi-brand kitchens: where it helps and where it hurts
Running three brands out of one kitchen spreads rent and labour across more revenue — which is the entire argument for doing it. It works when the brands share a production method and most of the prep.
It fails when the brands need different equipment and different mise en place. Then peak hour arrives, three tickets land on three different stations with one fryer between them, and preparation times blow out for every brand at once. Rating falls on all three, and a shared rating problem is worse than a single-brand problem.
The metric to run the business on
Track contribution per order and orders per day, weekly, per brand. Revenue growth means nothing if contribution per order is falling — that is a business buying its own customers at a loss, and the platform is a perfectly happy partner in that arrangement.
Questions owners ask
What is a good food cost percentage for a cloud kitchen? Lower than dine-in, because commission and packaging come out of the same rupee: aim for 22–28% of the menu price. Measure contribution per order alongside it.
Is a cloud kitchen cheaper to start than a restaurant? The fit-out is cheaper. The customer is not — you pay for every order through commission, discounts and ads, where a dine-in restaurant pays once through rent and location.
How many orders a day does a cloud kitchen need to break even? Divide your monthly fixed costs (rent, salaries, utilities, software) by contribution per order and by 30. At ₹90 contribution and ₹2,50,000 fixed costs, that is about 93 orders a day.
Sources and further reading
- MediaNama, CCI finds Zomato, Swiggy violating competition laws: report (November 2024)
- Business Today, Zomato and Swiggy found guilty by CCI for violating competition laws: report
- NRAI, India Food Services Report 2024, summarised by India Retailing
- TaxGuru, GST on restaurants: all you need to know — 5% without input tax credit applies to cloud kitchens too
If you want your delivery menu costed properly against actual commission and discounts, that is what our menu engineering engagement does. For kitchens being planned now, see kitchen planning and design.