Selling home cooked food online as a tiffin service

Four people in your building already eat your food twice a week and pay you in a vague, embarrassed way at the end of the month, and you're starting to wonder whether this could be twenty people instead of four.

It can, and the thing that decides whether it works is not your cooking. Take money in advance on a fixed cycle, run a fixed rotating menu with no choices, set a hard order cut-off the night before, and work out how many houses your delivery window can physically reach before you work out how many dabbas your kitchen can fill. At ₹90 a meal and 26 meals a month, one subscriber is ₹2,340 of predictable revenue, and predictable is the entire product.

A tiffin service is a subscription business that happens to involve food. Almost everything that goes wrong in the first year goes wrong because someone treated it as a restaurant instead.

The map is your constraint, not the stove

Your kitchen can probably cook for sixty. It cannot deliver to sixty.

Lunch has to land between about 12:30 and 1:30, and that's not negotiable, because your customer has a fixed break. Work backwards. If a rider takes four minutes per drop in a dense apartment cluster and eleven minutes per drop when the addresses are spread across three areas, then one hour of delivery window is fifteen dabbas in the first case and five in the second. The last person on the route gets food that left your kitchen fifty minutes before it reached them, so it has to have been packed hot enough to survive that.

Which means your first business decision is a circle on a map, not a menu. Pick a radius, and inside it pick the clusters: one office park, two apartment societies, one hostel. Fill those before you accept a customer four kilometres away who will single-handedly cost you twenty minutes a day forever.

Say no to that customer. Say it politely, say "I don't deliver to that area yet", and put them on a list. The one out-of-zone subscriber who convinced you to make an exception is the reason the other fourteen get cold food on Thursdays.

The mechanics of running local delivery without a courier company, including how to work with a rider you pay by the day rather than by the drop, are in delivering locally without a courier.

Prepaid cycles, and why they are non-negotiable

Do not run a tab. The seller who lets a customer "settle at the end of the month" is running an unsecured credit business with the worst possible collections process, which is knocking on a neighbour's door.

Three cycles that work:

Cycle	Price shape	Best for
Trial, 3 days	Full price per meal, ₹100 x 3 = ₹300	New customers, paid, never free
Weekly, 6 meals	₹95 x 6 = ₹570	Students, people who travel
Monthly, 26 meals	₹90 x 26 = ₹2,340	Your core, and the one you push

Collect before the first meal of the cycle. Not on day two. The weekly customer pays on Sunday evening for Monday to Saturday; the monthly customer pays on the last working day of the previous month.

Price the month slightly below the weekly rate per meal, so the monthly plan is obviously the better deal. That discount is buying you cash in hand and a customer who has already committed, which is worth far more than ₹5 a meal.

Never do a free trial. Charge full price for the trial, and if they don't continue, you've still been paid for three lunches. Free trials attract people who want free lunch, and a startling number of them will want a second free trial "for my friend".

The pause day, and the rules around it

Every subscriber travels. Every subscriber gets ill. If you have no pause policy, you'll invent one badly at 9pm on a Sunday over WhatsApp, differently each time, and within two months your billing will be a mess nobody can audit.

Write it once:

• Pause with notice by 8pm the night before. Paused days are added to the end of the cycle.
• Up to 6 paused days in a monthly cycle. More than that, the cycle is credited but not extended.
• No same-day pauses. The food is already cooked. This is the rule people push on and the one you must hold, because same-day pausing means you're guessing your quantities every morning.

Keep the pause record in the same place as the payment record. A single sheet with one row per subscriber, one column per day of the cycle, ticked when delivered and marked P when paused. It takes ninety seconds a day and it settles every dispute you will ever have.

The cut-off is an order of operations, not a preference

Orders for tomorrow close at 9pm today, because you buy vegetables in the morning and you buy them for a known number.

This sounds obvious and it's the thing new tiffin sellers break first, because a 10am "can I get one today?" feels like free money. It isn't. It's either an extra portion you had to stretch across everyone else's dabbas, or it's a second trip to the market that costs you forty minutes and puts your entire route back.

Publish the cut-off. Repeat it in the order confirmation. Then hold it for a fortnight, and by week three nobody will test it.

What a ₹90 dabba actually costs

Real numbers for a three-item vegetarian lunch: two rotis or rice, one sabzi, one dal, a small salad or pickle.

Line	Per dabba
Vegetables, dal, atta, rice, oil, spices	₹34
Gas	₹4
Disposable container and bag, if not using steel	₹11
Delivery, at a rider's day rate divided by 22 drops	₹14
Wastage, over-cooking, spoilage	₹3
Total	₹66

At ₹90 that leaves ₹24 a dabba, which is ₹528 a day on 22 subscribers, or about ₹13,700 a month, before your own labour. Four to five hours a day of cooking, packing and coordination, six days a week. That's roughly ₹115 an hour for your time, which is a real income in many Indian cities and is not a fortune anywhere.

Two levers move that number more than anything else.

Steel dabbas instead of disposables. Take a refundable deposit of ₹250 per subscriber, hand over two steel tiffins, and collect yesterday's on today's drop. That's ₹11 a meal saved, which is ₹242 a month per subscriber, which is more than the deposit inside five weeks. The cost is a washing-up load and the discipline of chasing containers back. Roughly one subscriber in fifteen will leave with a tiffin, which is what the deposit covers.

Route density. Going from 15 to 25 subscribers inside the same circle costs you almost nothing extra in delivery and adds ₹240 a day of margin. Going from 15 to 25 across a wider circle adds a rider.

Menu: fewer choices, longer rotation

Do not offer a choice of sabzi. The moment you do, you're cooking three dishes for twenty people instead of one dish for twenty people, and your food cost per plate jumps because you're buying in smaller lots.

What works is a fixed four-week rotation, published in advance. Monday to Saturday, the same slot each week gets the same kind of thing, but week one's Tuesday isn't week two's Tuesday. People want variety across the month and predictability across the week. Publishing the rotation also lets a subscriber pause the day they don't like, which is far better than them quietly cancelling.

Two exceptions worth building in:

• A no-onion-no-garlic version, if enough of your market wants it. Decide once, price it the same, and cook it as a genuinely separate small batch rather than pretending.
• A jain or fasting option on the days that matter, if you know your customers. This is the kind of thing that keeps a subscriber for a year.

Handle allergies and restrictions in writing at signup, not in a chat message you'll forget. Ask the question on the order: "anything you can't eat?" Then keep the answers on the same sheet as the pauses.

Photograph the actual dabba, opened, on a table, near a window, once a week. Not a styled thali. A tiffin buyer wants to know what fills the box, and the honest photo of the real portion converts better than the beautiful one. The technique is in what to photograph when you sell food, and the relevant part for you is shooting it in daylight before it goes out rather than under a tube light at 11pm.

Registration and hygiene, checked properly

Food sold to the public is regulated nearly everywhere, and home kitchens are usually included even at small scale. In India, food businesses including small home operations are expected to be registered with the national food safety authority, and for a very small turnover the basic registration is inexpensive and done online. What exactly is required, and at what turnover the requirement steps up, changes. Check the regulator's own website or ring your local food safety office rather than relying on what a Facebook group told someone in 2023.

Whatever the paperwork says, do these anyway because they are what actually keeps people well: separate cloths for raw and cooked, food into containers within an hour of cooking, hot food packed hot rather than warm, a fridge thermometer, and no cooking at all on a day when anyone in the house has a stomach illness. That last one costs you a day of revenue and saves you the business.

Display your registration number on your shop page once you have it. In a trade built entirely on trusting a stranger's kitchen, it's the cheapest credibility you can buy.

Worked example: Sunita, 24 dabbas, Kothrud, Pune

Sunita cooks a vegetarian lunch dabba and delivers to two apartment societies and one small office, all within 2.5km. Monthly plan ₹2,340 for 26 lunches, weekly ₹570, three-day trial ₹300.

She has 24 monthly subscribers and about four weekly ones at any time. Revenue is roughly ₹58,000 a month. Ingredients, gas, containers and her rider's monthly retainer come to about ₹34,000. She takes home around ₹24,000 for six mornings a week.

Her churn runs at two or three subscribers a month, which is normal and which she plans for. She replaces them with a referral offer: an existing subscriber who brings a new monthly subscriber gets four free meals added to their cycle. That costs her about ₹264 in food to acquire a ₹2,340 customer, which is the cheapest acquisition channel available to anyone in this trade. The general version of that mechanic is in how to get repeat buyers.

Payment is UPI, every time, to her UPI ID. She sends a message on the 28th with the amount and the ID, and she chases the three who haven't paid by the 31st. Nobody eats on the 1st without having paid.

The mistake she made in month two: she accepted five subscribers from an office 6km away because it felt like growth. The route went from 35 minutes to 70, the last dabbas arrived at 1:45, and she lost four of the original subscribers before she worked out why. She dropped the far office and rebuilt.

The honest limitation

Two things Sailo will not do for you here, and you should know both before you build on it.

There's no recurring billing. Sailo sells products, so a monthly tiffin plan is a product called "26 lunches, September" that you list, sell once, and re-list next month. It won't charge anyone automatically on the 1st. You'll be sending the reminder and watching for the payment yourself.

And there's no UPI rail. The complete list of payment rails is card, WhatsApp, Telegram, Instagram, email, phone, bank transfer and cash on delivery. What you can do is put your UPI ID into the bank transfer instructions field with the exact amount and the reference you want, which is what a lot of sellers do and it works fine. But be clear about what that is: Sailo is holding your menu, your plan and your order record, and your phone is holding the money. It cannot tell you a UPI payment arrived. Only your bank app can, which means a daily check. The habit that makes that painless is in how to know a bank transfer actually arrived.

Most of your day-to-day contact will happen on WhatsApp regardless, and having the order arrive there pre-written, with the plan, the address and the total already filled in, saves the thirty messages a day that otherwise eat your afternoon. How to take orders on WhatsApp covers that handoff.

What to do next

Draw your circle on a map today and mark every cluster inside it where twelve people eat lunch at the same time. That's your customer list, and it's a shorter list than you think.

Then write your four-week menu rotation, your cut-off time, your pause rules and your monthly price on one page, and take it to the first cluster. Offer a paid three-day trial to ten people. Cook for those ten for a week before you promise anything to anybody else, because the number you need to learn is how long your route actually takes with dabbas in the bag, and there's no way to learn it except by doing it once.