Solutions · Food & grocery

Milk & subscription delivery app: what to decide, build and test.

Planning one? Here's what to settle before you brief anyone, what goes in the first version, how the build runs week by week, how the beta is tested with real people, and the numbers that show it's working.

Wallet or monthly bill, and who gets the cutoff

A subscription app is a promise kept at 6 am, and everything behind it is a night shift. Four decisions shape the build more than any screen.

  • Prepaid wallet or monthly invoice. A wallet gives you cash up front and removes chasing, but customers need a ledger they trust and a clear refund path. A monthly bill is an easier sell and means collections work. Many dairies offer both.
  • Where the cutoff sits. A later cutoff, such as 10 pm, is friendlier to customers; an earlier one gives the dairy more time to buy and pack. Milkrun sets 10 pm and lays out the night behind it: packing at 1:30 am, loading at 4:15 am, the round from 4:50 am, delivery by 7 am.
  • Fixed route or flexible slots. Fixed routes by society and tower are cheapest to run. Slots add choice and add cost.
  • Milk only or a basket. Milk is daily and predictable. Bread, eggs and curd add stock risk and more pause and skip rules.

Then decide what happens when the wallet runs short. Milkrun's rule is that milk is kept first and extras are dropped, so a household still gets milk. Whatever you choose, it has to be written in the app. Our subscription app guide and the milk delivery playbook cover the business side.

  • Wallet, monthly bill or both
  • Cutoff time, and what a change after it does
  • Products at launch and which can be paused individually
  • Delivery areas, societies and delivery staff per route
  • Returnable bottles or crates: deposit or no deposit
  • Rule for a missed delivery: refund, credit or redeliver

Version 1: the calendar, the wallet and the route sheet

Subscription app: first release versus second
ModuleFirst versionAdd laterWhy
Subscription rulesDaily and chosen weekdays, quantity per dayAlternate-day and custom patternsMost homes start with daily. Patterns matter after launch.
Pause and skipDate range pause, skip tomorrowAuto-resume remindersTravel is the most common reason to cancel.
WalletUPI top-up, per-line ledger, low-balance alertReferral credit, bulk top-up offersA ledger that explains each deduction prevents disputes.
Cutoff engineOne cutoff, countdown, lockProduct-specific cutoffsThe lock is what lets you plan tomorrow's purchase.
Route sheetStops by society, tower and flat, delivered or not and reasonRoute optimisation, photo proof on every dropDelivery staff follow the sheet in the dark.
Procurement viewTomorrow's litres and itemsWeekly forecastBuying right reduces waste.
Owner consoleHouseholds, low balances, paused customersCohort and churn reportsThe owner decides routes from this screen.
ReturnablesCount with customer, returned, deposit lineBreakage rules, milestonesSkip it if you deliver in pouches or cartons.
MessagesWhatsApp and SMS: low balance, missed deliveryMonthly statementsAlerts cut support calls.

Six to ten weeks to a first delivery round

Shorter plans suit a single product and one route. Migrate your paper diary into the app in week 5 so the pilot starts with real data.
WeeksWe buildYou supply
1–2Subscription rules, cutoff logic, wallet ledger design, prototypeProduct list, prices, cutoff time, delivery societies and towers, current customer list
3–4Customer app: calendar, pause, walletCopy in your customers' languages, brand assets
5–6Delivery app and route sheet, owner consoleTwo delivery staff with their phones, today's paper route diary
7–8Payments, notifications, procurement viewPayment gateway account and KYC, WhatsApp and SMS sender approvals
9–10Run a live pilot on one route, fix, launchPilot households, FSSAI details, Play Store account

How we work with you each week

The same rhythm runs from week 1 to launch, so you always know what's happening and what we need from you.

  • Friday: a demo link and a 30-minute call. You open the latest build on your own phone and try it yourself, rather than watching our screen.
  • Within 48 hours: your feedback, in writing. In one shared doc, next to the screen it's about. Written feedback is clear, dated and doesn't get lost between calls.
  • Monday: triage. Every comment goes into fix now, next week or later. Anything that changes the agreed scope gets a written quote before we touch it.
  • Each milestone: two revision rounds, then your sign-off. Sign-off is against the scope we agreed, so it's clear what done means.
  • One decision-maker on your side. Most delays aren't build time. They're feedback waiting on several people who disagree.

The full version, including how the beta and user testing work across every kind of project, is on our process page.

The beta: a live pilot on one route

The pilot is week 9 of the plan. It starts once the paper diary is in the app, the 10 pm lock works, and tomorrow's route sheet prints itself from tonight's subscriptions. It is a real morning round, with real milk and real money.

  • What's in it. Daily and chosen-weekday subscriptions, pause and skip, the wallet with UPI top-up and its ledger, the cutoff lock, the route sheet, the procurement view and low-balance alerts.
  • What stays out. Autopay, alternate-day patterns, bottle breakage rules, monthly statements and route optimisation. They can wait until the basic night-to-morning loop runs clean.
  • Who uses it. One route: the two delivery staff who already walk it, on their own phones, the person who packs at night, and 15–25 households on that route who already take milk from you. You pick and invite them. Neighbours doing you a favour will not pause for a trip or argue over a deduction, and those are the cases that matter.
  • How they get it. Households install through TestFlight or Google Play testing. Delivery staff get the Android build. The owner and packer use the console through a staging link on a laptop or tablet.
  • How long. Two weeks, so the pilot covers at least one weekend, a few pauses and one low-balance cycle. Keep the paper diary running in parallel for the first few mornings, in case the sheet is wrong.

A new personal Google Play developer account has to run a closed test with at least 12 testers for 14 days before it can publish. Count the pilot households towards that, or use an organisation account.

What we test with real people, and the tasks we set

We set a task and watch. The tester uses their own phone, in the language they read most easily, and we don't step in. About five people per role shows us where the app trips people up. We fix that and give the next group the same tasks.

Delivery staff are tested on the real round, before sunrise, not in the office at noon.
WhoThe taskWhat we watch for
Household"You're going to your native place from Friday to Monday. Stop the milk for those days, but keep the curd on Sunday for your parents."Do they pause a date range without help? Can they keep one product while pausing another? Do they see whether Friday is already locked?
Household"Your wallet shows ₹38 less than you expected. Find out why, then top up ₹500 by UPI."Can they read the ledger line by line? Does the UPI app return to ours? Does the new balance show straight away?
Delivery staff"Start the 4:50 am round. Tower C, flat 704 doesn't open the door. Mark it, and note the two empty bottles collected at 702."Is the sheet in walking order? Does it work in a stairwell with no signal? Is the not-delivered reason one tap?
Night packer"It's 10:05 pm. Check tomorrow's total: how many litres of full-cream, how many packets of bread."Does the procurement view match the route sheets? Do late changes after the lock stay out?
Owner"A customer calls to say milk didn't come today. Check what happened and give a credit."Can they find the stop and the staff note in under a minute? Does the credit show on the customer's ledger?

The numbers that say it's working

In week 1 you choose two or three of these. Your paper diary often gives a starting figure, such as missed deliveries last month. We add tracking before the pilot, then share the numbers at launch and 30 days after. The numbers are yours to judge. Our job is to make them easy to see.

GoalWhat we measureWhere it comes from
Every household gets its milkMissed deliveries per 100 drops, by reasonRoute sheet
The round finishes on timeTime of the last drop on each routeDelivery app timestamps
Customers manage changes themselvesPauses and skips done in the app versus by phone callApp data and owner call log
Buying matches demandLitres bought versus litres delivered each dayProcurement view
The wallet doesn't cause disputesBalance queries and refund requests per monthOwner console
Households stayHouseholds still active 30 days after joining, and pauses that never resumeSubscription data

What a subscription app costs, and what pushes it up

In the app cost guide, a customer app, delivery app and owner console is a two user types plus admin product, ballpark ₹10–25 lakh from a small studio. A single dairy with one route and a monthly invoice, with no wallet, can sit closer to the one-user-type band of ₹3–10 lakh. These are ballparks and we quote on your brief.

  • Wallet versus invoice. A wallet needs a ledger, refunds, top-up reconciliation and a policy for unused balance.
  • Autopay. Recurring debits go through the bank's mandate rules and add testing time.
  • Rule variety. Four repeat patterns plus pause plus per-day quantity is far more logic than daily only.
  • Returnable bottles. A deposit and bottle ledger is a mini-product of its own.

Wallets, mandates, milk and personal data: rules to design for

  • Your wallet. An RBI prepaid instrument that can only buy from the entity that issued it, with no cash withdrawal, is classed as a closed system and does not need RBI authorisation. That fits a dairy wallet used only for your own products. A wallet that pays other businesses, or lets customers withdraw cash, is a different matter and needs advice. Have a lawyer check your wallet terms and your unused-balance refund policy.
  • Autopay. Under RBI's e-mandate framework, recurring debits up to ₹15,000 per transaction can go without extra authentication, and the customer must be notified at least 24 hours before each debit. Show the amount and date in the app and send the notice.
  • FSSAI. The licence or registration number goes on every bill and receipt. Dairy and packaged items have their own licensing categories, so confirm yours on the FSSAI portal.
  • GST. Milk, other dairy and bakery items are taxed differently,, and rates can change. Store a rate per product and ask a CA to confirm each.
  • Personal data. You hold addresses, flat numbers, phone numbers and routines such as when a house is empty. Collect only what the route needs, ask for consent, and let customers delete. DPDP obligations phase in over 18 months from November 2025. See the DPDP checklist.
  • Messages. Order alerts over WhatsApp need approved templates and opt-in. See WhatsApp for business apps.

This is general information for planning, not legal or tax advice. Confirm with a CA or lawyer before launch.

Seven subscription mistakes, and the fix

  • A vague cutoff. If a pause after 10 pm is unclear, the dispute lands on your delivery staff. State which days are locked, in the app.
  • Wallet with no ledger. Customers cannot see why money left. Show each line.
  • Silent deductions when the balance is short. Decide what is dropped first and tell the customer in advance.
  • Route sheets in the wrong order. Delivery staff lose minutes at every gate. Sort by society, tower and flat, then walk the route yourself.
  • Assuming data on delivery phones. Basements and stairwells have no signal. Make the sheet work offline.
  • No missed-delivery rule. Pick refund, credit or redelivery before launch.
  • Importing the paper diary untouched. Spellings and flat numbers differ. Clean it once before day one.

Start with one route and the households you already serve. When the missed-delivery rate is low for a few weeks, add a second route, then products.

Cost and timeline

6–10 weeks for a first version.

Fixed quote within 48 hours of your brief: scope, timeline and price in writing before we start. These are what move the number.

  • Number of products and subscription rules
  • Wallet, autopay and invoicing
  • Route planning across societies and areas
  • Bottle or crate returns
  • Integrations with your billing or accounts

How app costs work in India →

Questions

About your milk & subscription delivery app.

What does a milk delivery app cost?

It depends on the subscription rules, the delivery side and the reports you need. Every project gets a fixed quote within 48 hours of your brief. For typical ranges, read our guide to app development cost in India. A single-dairy app costs much less than a multi-vendor platform.

How does the cutoff work?

You choose a time, say 10 pm. Until then customers can change tomorrow's order; after it, the order locks and goes into the route sheet and procurement forecast. A pause that starts after the cutoff is handled by the rules, and the app says which days are already locked.

Can customers pay monthly instead of prepaid?

Yes. The app can run a prepaid wallet, a monthly bill paid by UPI, or both, with a statement that explains every charge, skip and refund. UPI autopay can collect the monthly bill automatically for customers who opt in. Customers can see each day's charge.

Will the delivery app work on cheap phones?

Yes. The route sheet is light, works offline in basements and stairwells, and syncs deliveries once the phone is back on the network. Stops are listed by society, tower and flat in route order, with large buttons that are easy to tap with one hand before dawn. Photo proof is optional.

Who owns the app and customer data?

You do. Code, design files and your customer list are handed over in full. There is no per-order or per-customer fee to us. Your subscriber list, delivery history and payments stay in your own database and cloud account. Nothing stays locked in our accounts.

Start a project

Tell us how your business runs. We'll build what it runs on.

Four taps, about a minute. Fixed quote within 48 hours.

Menu

Read

Ask an expert Ask an expertFree, real peopleHub
Ask an expert