Cloud Kitchen Launch Playbook

This playbook is a cloud kitchen business plan written as a sequence of decisions rather than a document for a bank. It is for someone in India with ₹4 to ₹12 lakh, a cuisine they can cook well, and a date in mind, who wants to open a delivery-only kitchen and still have money in the account in month four. By the end you will have a budget with every line filled in, a licence checklist in the order you apply, a 12-item launch menu priced from recipe cards, a delivery radius with zones and a rider decision, a launch plan that does not depend on paid placement on the delivery apps, a 30-day order target sheet, and a rule for when a second brand makes sense. The one belief it will change is that a cloud kitchen is a cheaper restaurant. It is a different business: the rent you save is paid back to the delivery apps as commission unless a share of orders comes to you directly, so the ordering channel is a line in the plan from the first week, not an afterthought for month six. Every rupee figure is illustrative, built from typical numbers rather than any named kitchen, and your own prices and rents replace them.

Chapter 1: The plan on one page

A cloud kitchen, also called a ghost or dark kitchen, is a food business with no dining area, where every order arrives through a channel and leaves in a bag. That sentence contains the whole economics. Rent is lower and front-of-house staff do not exist, which is the saving everyone counts. Every order also carries a channel cost and a packaging cost, which is the spend most first plans forget. The step-by-step version of the opening sequence is in how to start a cloud kitchen in India; this chapter is the arithmetic that sits under it.

Five questions the plan must answer in numbers

1. What is the average order value? A ₹180 single-item format needs volume and tight packaging; a ₹450 family-meal format needs fewer orders and better presentation. Everything from equipment to radius follows from this number.
2. How many households sit within a 15-minute ride? Density decides your radius, your rider cost and your ratings. Count it before signing a lease.
3. What share of orders will come through the delivery apps, and what will each cost? Commission is commonly 18 to 30 percent of order value, plus GST on the commission, processing charges and any discounts you fund.
4. What is the fixed monthly cost? Rent, salaries, gas, utilities, software. This is the number you divide by your contribution per order to find break-even.
5. How many months of fixed cost are in the bank? Three is the minimum. Kitchens that close in year one usually had equipment money and no runway.

A sample month, two ways

The table below is illustrative for a single-brand kitchen doing 60 orders a day, 1,800 a month, at ₹320 average, which is ₹5.76 lakh in sales. Food cost is 33 percent, packaging 6 percent, rent ₹25,000, three staff ₹70,000, gas and utilities ₹20,000, sundries ₹10,000. App orders carry 25 percent commission, 18 percent GST on that commission, 2 percent processing and a small funded-discount line. Direct orders carry a flat ₹10 per-order fee, a 2 percent gateway charge and ₹40 per order for your own rider. The only thing that changes between the two columns is the channel mix.

Same kitchen, same food, same staff, double the profit, because 630 orders a month moved from a channel that takes a percentage to one that takes a flat fee and a rider. That shift is the business model, and it is why the direct channel appears in Chapter 5 of this playbook rather than in a later one. The realistic margin bands by format, and the levers that move them, are in cloud kitchen profit margin: realistic numbers.

Break-even in one line

Contribution per order is price minus food, packaging and channel cost. On the app at ₹320 that is about ₹93 after ₹106 food, ₹19 packaging and ₹102 channel cost. Direct with your own rider it is about ₹139, with ₹56 of channel and rider cost. With ₹1,25,000 of fixed cost, break-even is around 1,340 app orders a month or around 900 direct orders, roughly 45 a day versus 30 a day.

Checklist: the one-page plan

• Format, cuisine and target average order value written down.
• Household count within 3 km of two or three candidate locations.
• Channel mix assumed for month one, month three and month six, with the effective app cost from Chapter 1's numbers.
• Fixed monthly cost listed line by line.
• Break-even orders per day calculated for both channels.
• Three months of fixed cost set aside before any equipment is bought.

If you want the structure as a document, the cloud kitchen business plan template follows the same sections.

Chapter 2: Budget and licence checklist

Equipment is rarely the largest number. The deposit, the fit-out and the cash you need to survive the slow first weeks together outweigh the burners and the fridge. Ranges below are for a 250 to 600 square foot unit in a tier-2 city or a metro suburb designed for 60 to 120 orders a day; metro core areas push rent and deposit higher, and used equipment pulls the middle rows down.

The budget, line by line

Three budgets fall out of that table: a home-based test at around ₹1.5 lakh to prove the menu before renting anything, a standard single-brand unit at ₹6 to ₹8 lakh, which is the plan most first-time owners should make, and a multi-brand unit at ₹12 to ₹15 lakh and up, sensible only after one brand has run for a few months. The reasoning behind each row is in cloud kitchen setup cost: a line-by-line budget.

Where to cut and where not to

• Cut: new tables and racks, opening stock for a long menu, an interior nobody will see, equipment for dishes you have not validated, and printed boxes before the menu is proven.
• Do not cut: the exhaust, refrigeration quality, the deposit cushion, working capital, or the insert cards that bring repeat orders to your own number.

Licences in the order you apply

Two of these gate everything else. Without FSSAI and GST you will not be listed on the delivery apps or activated on a payment gateway, so they go in the first week, running in parallel with fit-out. A cloud kitchen has no dining area, which removes the eating house licence a restaurant needs. Local registrations vary by state and city, so check with your local office which apply to your address.

Keep clear scans of the owner's ID and address proof, the rent agreement with the landlord's written consent for commercial food use, a kitchen layout or photo, PAN, bank proof and the food category list in one folder before you apply anywhere. The delivery apps ask for FSSAI, GST, PAN, bank proof, menu and photos at onboarding. Razorpay asks for PAN, bank details, business proof and usually the GST certificate. There is no separate licence for taking orders on WhatsApp or a store page. Documents, fees and state variations are in cloud kitchen licences required in India.

Checklist: budget and licences

• Every row of the budget table filled with your own quotes, including a contractor's walk-through of the unit before signing.
• Working capital equal to three months of fixed cost, held separately from the equipment money.
• Document folder complete before the first application.
• FSSAI, GST and trade licence applied for in the same week as the lease.
• Gas agency inspection booked; extinguishers and cylinder area ready for it.
• Payment gateway activation submitted the day GST arrives.

Chapter 3: A menu of 12 items, not 60

A launch menu of 12 to 20 items, sharing most of its ingredients and with every dish surviving 30 minutes in a box, is the whole design brief. Sixty items because the cook can make them all means wastage on the slow sellers, a kitchen that stops to look things up at peak, and a listing customers scroll past because choosing is work. Twelve is the number to launch with; growth comes from variants and a weekly special, and from cutting the bottom two items every month.

Start from the ingredient matrix

List the ingredients you want to carry: proteins, bases, sauces, vegetables, dairy, dry goods. List candidate dishes as columns and mark which ingredients each uses. The rule for a place on the menu: a dish must reuse at least 70 percent of its ingredients from dishes already on the list, or it must be a hero strong enough to justify its own supply line. A fish curry that brings in fish and coconut nothing else uses is the first cut.

The structure of a 12-item launch menu

• Three or four heroes. The dishes you are known for, with the best photos and the tightest recipe cards. Chicken, mutton and veg versions of one biryani count as one dish with three variants, one container and one process.
• Four or five sides and add-ons. Raita, salad, a kebab, a drink, a dessert jar. Low food cost, built from what is already in the kitchen, and the reason an order grows from ₹280 to ₹360 without more time at the busiest station.
• Two or three combos. A hero, a side and a drink at ₹20 to ₹30 below the sum of the parts. Combos simplify the decision and steer demand to the items with the best margin.
• One weekly special. The same day each week. It gives regulars a reason to come back on a schedule and tests new dishes without committing them to the permanent list.

Price from the recipe card, illustratively

Weigh every ingredient and cost it at the current supplier price, then add packaging. A chicken biryani plate: rice ₹18, 180 grams of chicken ₹45, marinade and whole spices ₹10, onions, oil and ghee ₹8, raita ₹7, which is ₹88 on the plate. Container, lid, bag and cutlery add ₹14, so ₹102 all in. At a target food-and-packaging cost of 34 percent the price is ₹300. If you want ₹280 on the menu, the plate has to come down to ₹95. On the app menu, list it 10 to 15 percent higher, around ₹335, so that the price after commission lands near your direct price; check the partner agreement for parity clauses first. Keep direct prices equal to or below app prices, never above, because direct is where you want the repeat order.

The box test

Cook every candidate dish, pack it exactly as it will ship, bag it, leave it 30 minutes, then open and eat it. Rice, gravies, breads packed separately, kebabs and most Indo-Chinese pass. Fried items, assembled chaat, crisp-topped dishes and anything with a sauce poured on top usually fail unless the components are separated. A dish that fails the box test does not go on the delivery menu however good it is hot, because one soggy order produces a review that costs more than the dish earns in a month.

Dayparts

Nothing says the whole menu has to be live all day. Breakfast items from 7 to 11, the lunch combo from 12 to 3, the full dinner list from 7. A shorter live menu per daypart means fewer "item not available" cancellations and a kitchen that preps for the next three hours rather than for everything. On Tezzu, menu scheduling by weekday and time switches items on and off by themselves, and bundles, including build-your-own bundles, handle the combos. The matrix method, the monthly review and the pricing ladder are worked through in cloud kitchen menu design: fewer items, higher margins.

Checklist: menu locked

• Ingredient matrix drawn; every dish reuses 70 percent or more of the list, or is a hero.
• 12 to 20 items: heroes, sides, combos, one weekly special.
• Every item costed from a recipe card with packaging included; food-and-packaging cost between 30 and 35 percent for heroes, lower for sides.
• App menu prices set above direct prices by a modest margin after reading the agreement.
• Box test passed by every item on the list.
• One real photo per item in daylight, in the actual packaging.
• Searchable names and one-line descriptions with portion size, spice level and veg or non-veg marked.

Chapter 4: Delivery radius and rider options

For hot food, 3 to 5 kilometres is the practical radius, with 5 to 7 workable for dishes that travel well and for pre-orders. The number that actually matters is minutes. A rider in city traffic averages somewhere around 15 to 20 kilometres an hour once signals, lanes and finding the building are included, so 5 km is 15 to 20 minutes of riding plus pickup and handover, on top of 15 minutes of preparation. The customer is eating 40 minutes after ordering, which is the edge of what most hot food tolerates. A kitchen that delivers 6 km at 3 pm may sensibly deliver 4 km at 8 pm on a Saturday.

Density before distance

Three kilometres around a kitchen in a dense residential area can hold more households than seven around a kitchen on an industrial edge. Before signing the lease, draw a 3 km circle on the map around each candidate unit and count what is inside: apartment complexes, PG clusters, office parks, colleges. A kitchen with 20,000 households inside that circle runs a small radius, short delivery times, low rider cost and high ratings. A kitchen with 4,000 has to stretch to 7 km and pays for it on every order.

Zones instead of one circle

On the delivery apps the serviceable area is set by the app and changes without notice. On your own channel the radius is yours, and the useful shape is two or three rings with different charges and minimum orders. The figures below are illustrative for a kitchen with its own rider.

Zone 3 exists for the customers who ask, at a charge that makes the trip worth it. Serving them at ₹70 with a ₹400 minimum beats refusing them, and beats serving them at ₹20 and losing money on every trip. The inner-zone threshold, set a little above your average order value, is what makes a customer add a side to cross the line. The full reasoning, including how to measure late deliveries by zone, is in cloud kitchen delivery radius: how far should you deliver.

Four ways to move the food

The usual path is a per-trip partner for the first two to four weeks while volume is unpredictable, then a salaried rider once direct orders pass about 15 a day, with the per-trip partner kept as overflow for Friday and Saturday nights. Pre-orders and scheduled delivery let you serve the far zone at off-peak times when the rider is free and the road is clear. The cost comparison between the two rider models across a month is in hyperlocal delivery: in-house vs third-party riders.

On Tezzu, zones are set by distance with a charge and an optional free-delivery threshold per zone, customers see the charge before confirming, self-pickup points carry a collection time, and scheduled orders and pre-orders handle the far zone. For packaged food going further, Tezzu takes the order and the payment, and you book a courier service yourself.

Checklist: radius and riders

• Household count within 3 km recorded for the unit you signed.
• Two or three zones set with charges, minimum orders and an inner-zone threshold.
• Rider plan for weeks 1 to 4 and for the month after, with the trigger for hiring written down.
• A pickup point and collection time configured, even if you expect few collections.
• Late deliveries, cancellations and ratings tracked by zone from week one.
• Peak-hour radius decided separately from the off-peak radius.

Chapter 5: Launching without aggregator ads

Paid placement on the delivery apps buys visibility for a week and teaches customers to find you through the app. Launching without it is slower in week one and cheaper in every week after. The method has two channels from the first day: the apps, listed at full price with no funded discounts, for strangers; and your own WhatsApp number and store page, for everyone who has eaten your food once. The 21 ideas that make up the direct side, from society groups to Google posts, are listed in cloud kitchen marketing: 21 ideas that do not depend on aggregator ads; the sequence below picks the ones that work in the first month.

Two weeks before opening: the list and the channel

Your first 50 customers should be people you can name before the burners are lit. Write down 100 phone numbers: friends, family, colleagues, your building, the gym, the old office. Message each one personally, from your own phone, telling them what you are launching and when, and asking whether they would like a message on opening day. Record every yes. That is your first opted-in list, and those replies are the consent Meta expects before you send a campaign.

Set up the ordering channel in the same fortnight. On Tezzu that is your own WhatsApp number connected through Meta's embedded signup, the 12-item menu with photos, the free store page at tezzu.in/store/your-name, UPI and cards through Razorpay with cash on delivery as a fallback, the zones from Chapter 4, and the staff app on the kitchen phone. Setup is about 30 minutes of actual work; Razorpay activation runs for a few working days in the background, so apply early. Place five test orders and watch each one through the dashboard.

Set up your Google Business Profile with the category, hours, photos and the store page as the order link; verification takes days, so do it now. Print 500 insert cards with your WhatsApp QR, the store link and a first-order offer. Every bag that leaves the kitchen carries one from the first order, app orders included.

Week one: soft launch to people who know you

Open to the list first, with one opening message to everyone who said yes, the menu link and a launch coupon that applies itself at checkout. Ask the first 30 customers for honest feedback on WhatsApp the next morning and fix two things a day. The point of the week is to find the packaging that leaks and the dish that takes too long while the audience is forgiving.

Do not list on the delivery apps yet. Your first app ratings stick, and a 3.6 from a messy opening week follows you for months.

Week two: go live for strangers, at full price

List on the apps with good photos, the short menu and no deep discount. One modest launch offer rather than a 50 percent banner, which attracts deal hunters who never return and sets a price you have to walk back. Say no to the add-on programmes and funded promotions switched on by default; review them one by one and keep only what brings new customers. Ask every happy week-one customer for a Google review.

Then do the free things. Get added to the WhatsApp groups of three housing societies within 2 km; offer the committee a tasting box and ask to post your menu on the days vendors are allowed. Post three Instagram reels: the kitchen, the packing table, the rider leaving, each caption ending with the ordering instruction and the link in the bio. Post on WhatsApp Status three times a week with a real dish photo and the link.

Week three: the first direct campaign and a small paid test

Send your first weekly broadcast to everyone who opted in, on a weekday around 6 pm, with one offer and the order link. Launch a first-direct-order coupon worth roughly the commission you would have paid: ₹50 on a ₹300 order. If there is budget, this is the place for it rather than app placement: a WhatsApp Status ad or a Click-to-WhatsApp ad at ₹300 to ₹500 a day, aimed at a 5 km radius with one real dish photo, where the click opens a chat with your number and the customer who orders is yours from the first order. Both run from the Growth plan on Tezzu. Measure cost per first order, never clicks. The setup and the arithmetic of that test are in WhatsApp Status ads for a new cloud kitchen.

What app ads actually buy, so you know when to use them

Promoted placement on the apps is useful on three occasions: a new area, a new brand and a slow season. It is wasteful when it shows your listing to people already searching for your name, and it adds its own cost on top of the commission on every order it produces. If you ever run it, track the before-and-after order count and switch it off the week the extra orders stop covering the spend plus the commission.

Checklist: launch without placement spend

• 100 personal messages sent; yes replies recorded with dates.
• Ordering number, menu, store page, payments, zones and staff app live and tested with five orders.
• Google Business Profile verified with the store page as the order link.
• 500 insert cards at the packing station; a card in every bag from order one.
• Week-one feedback from 30 customers, two fixes a day.
• App listing live in week two at full price, default promotions switched off.
• Three society groups, three reels, Status three times a week.
• First broadcast sent in week three with an auto-apply coupon; any paid test measured by cost per first order.

Chapter 6: The first 30 days order plan

The first month works in four stages: people who know you, then strangers from the apps and Google, then the first direct campaign, then repeat and referral. Targets below are for a popular cuisine in a dense urban area; halve them for a niche cuisine or a quieter suburb and keep the sequence. The week-by-week narrative with the actions behind each target is in how to get orders for a new cloud kitchen in the first 30 days.

Direct share dips in week two when the apps switch on and then climbs, which is expected. The insert in every app bag is what makes it climb; if it is under 10 percent by the end of week four, the card or the coupon is not working and needs a rewrite before anything else.

The five numbers for the first month

• Orders per day by source, so you know what each channel contributes.
• Direct-order share, because every point it rises cuts commission.
• Second-order rate within 14 days, which predicts whether the kitchen survives month three. One in three first-time customers ordering again is healthy; one in ten means you will spend forever buying new ones.
• Average rating on the apps and on your store page, checked daily in the first month, with the reason behind every rating under four fixed the same day.
• Cost per first order on any paid spend, compared with the commission you would have paid for the same order through an app.

Operations that hold at 60 orders a day

Order flow breaks before the kitchen does. New-order alerts have to reach a phone in the kitchen, with a new-order alert on if the exhaust is loud; a thermal KOT goes to the cook; statuses get changed as the order moves so the customer gets "out for delivery" and the rating request goes out after "delivered". Give the dashboard to a named person per shift. The mistakes that close kitchens in year one are mostly operational, and why cloud kitchens fail lists them from owners who shut down.

Checklist: end of month one

• Real food cost and real packaging cost known from a month of purchases.
• The three dishes that make up most orders identified; the menu shortened around them.
• Direct share at 20 percent or more; opted-in list growing with every first order.
• Second-order rate measured and written down.
• Rider plan reviewed against actual deliveries per day.
• App promotions reviewed; anything that did not bring new customers switched off.

Chapter 7: Multi-brand expansion, when and how

A second brand from the same kitchen is a good idea on a specific day: the day the first brand has run for 90 days, holds a steady order count, has a known food cost and a repeat rate you can state. Before that, a second brand doubles the listings to maintain, the photos to shoot and the packaging stock to hold while adding a fraction of the revenue. Three is a sensible ceiling for one kitchen and one owner.

Why a second brand works at all

Discovery: three entries in a scrolling list get seen more than one. Utilisation: a biryani kitchen is busy from 7 to 10 pm and quiet the rest of the day, so a lunch roll brand uses the same cooks and gas in the empty hours.

Choosing brands that share a pantry

The rule is overlap in ingredients and equipment, with no overlap in what the customer sees. Two brands selling butter chicken are one brand with two names and twice the sticker stock. One brand needing a pizza oven next to two needing a tandoor is two kitchens' worth of equipment.

Each brand gets its own 8 to 15 item menu, no item appears on two brands, and prep runs from one list written by ingredient: 12 kg of chicken marinated at 9 am, 8 for biryani and 4 for rolls. Base containers, bags and cutlery are shared across brands and differentiated with a ₹1 sticker and an insert card, so packaging stock stays at a dozen SKUs. Printed boxes wait until a brand has survived 90 days.

The kill rule, written before launch

A brand that does not reach a minimum daily order count and a positive contribution after its share of rent, salaries and gas within 90 days is closed or reworked. Thirty orders a day is a common threshold for a brand that shares most of its prep; a brand with its own equipment needs more. Allocate shared costs by share of orders or share of revenue, write the rule down, and do not change it to flatter a favourite.

One ordering number or three

On the apps each brand is its own listing with its own rating, which is useful because a bad week for desserts does not drag the biryani listing down. On your own channel there are two shapes. One WhatsApp number and one store page per brand gives three identities and three menus, matching what customers see on the apps. One number and one store page with the brands as menu sections gives one cart and one delivery charge when a customer wants biryani and phirni together, which is cheaper to run and weaker on brand separation. Many kitchens start with one number and split a brand out once it has a following. Tezzu supports both shapes, with menu scheduling keeping the breakfast brand visible only in the morning, and a second kitchen later can run as its own store profile under the same account, with its own number, menu, delivery zones and staff. The operating detail, from the prep list to per-brand accounting, is in multi-brand cloud kitchen: how to run 3 brands from one kitchen.

Checklist: ready for a second brand when

• Brand one has run 90 days with a steady daily order count and a known food cost.
• Second-order rate for brand one is at one in three or better.
• Candidate brand shares 70 percent or more of its ingredients and all of its equipment with brand one.
• Candidate brand peaks in a different daypart.
• Kill rule written with a number and a date.
• Ordering channel shape decided: one number or one per brand.

Your 90-day plan

The plan starts six weeks before opening, because licences and fit-out run in parallel and the direct channel has to exist before the first customer. Order and direct-share targets are illustrative for a popular cuisine in a dense area.

The Sunday review, every week for 13 weeks

• Orders per day by source and direct share, against the target for the week.
• Second-order rate within 14 days.
• Ratings under four on any channel, with the cause fixed that week.
• Late deliveries by zone; charge or minimum raised for any zone producing a third of complaints from a tenth of orders.
• Food and packaging cost against the recipe cards.
• Cash in the bank against the three-month runway.
• One change for next week. Only one.

Frequently asked questions

How much does it cost to start a cloud kitchen in India?

A single-brand kitchen in a rented unit typically needs ₹4 to ₹12 lakh including deposit, fit-out, equipment, licences, opening stock and three months of working capital, and ₹6 to ₹8 lakh is the budget most first-time owners should plan for. A test from an existing home kitchen can be done for around ₹1.5 lakh. Shared kitchen facilities lower the upfront figure and raise the monthly rent. The line most plans miss is working capital, and it is the one that closes kitchens.

What should a cloud kitchen business plan include?

Five things in numbers: the format and target average order value, the household count within a 15-minute ride of the unit, the channel mix with the effective cost of an app order and a direct order, the fixed monthly cost line by line, and break-even orders per day for each channel. Add the budget table, the licence checklist with dates, the 12-item menu costed from recipe cards, the zone and rider plan, and the 30-day order targets. A plan that only has equipment and rent in it is half a plan.

Can a cloud kitchen succeed without delivery app ads?

Yes, and it is the cheaper way to start. List on the apps at full price for discovery, switch off the default promotions, and build the direct side from day one: a list of 100 people before opening, an insert card in every bag, your store page on Google and Instagram, society groups, and a weekly broadcast to customers who opted in. If you spend on ads, spend where the customer who orders becomes yours, such as a WhatsApp Status or Click-to-WhatsApp ad, and measure cost per first order.

How many items should a new cloud kitchen have on the menu?

Twelve to twenty, counting variants of one dish as one item: three or four heroes, four or five sides and add-ons, two or three combos, and a weekly special. Every item should reuse most of its ingredients from the others and survive 30 minutes in a box. Cut the bottom two items every month and add at most one, so the kitchen stays fast and the customer always sees a list tested against real orders.

When should a cloud kitchen add a second brand?

After the first brand has run for about 90 days with a steady order count, a known food cost and a second-order rate you can state, and only when the new brand shares most of its ingredients and all of its equipment with the first while peaking in a different daypart. Write a kill rule before launch, such as 30 orders a day and a positive contribution within 90 days, and keep to three brands at most for one kitchen and one owner.