Solutions · Food & grocery

Food 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.

Whose riders, whose menu, whose money: settle it before the brief

A food delivery app is three or four products sharing one order, and the business model decides how many of them you build. Four forks change the scope more than any feature.

  • One brand or many restaurants. One brand means one menu and no onboarding or payouts, and it is the cheaper way to learn whether customers reorder. A marketplace adds restaurant onboarding, commission rules, settlement statements and a support queue for disputes between three parties.
  • Own riders or a delivery partner. Own riders give you control of timing and cost per drop, but you carry rostering, cash handling and rider support. A partner removes the rider app from your build, and you give up tracking detail and some customer experience.
  • Commission or subscription for restaurants. Commission is easy to sell and scales with orders. A monthly fee is predictable but is a harder first conversation with a kitchen that has no proof of volume yet.
  • One city or a spread. Delivery times depend on rider density. Ten kitchens in one area beat fifty across a city.

Our FullPlate concept build shows what the marketplace choice costs in screens: its payout screen has to break out commission, GST on commission and TDS, and its rider side needs offers, batching and an earnings view. A single-brand app skips all of that. The food delivery guide and the food delivery playbook cover the business side.

Write down before you brief anyone:

  • Single brand or marketplace, and which one in version 1
  • Own riders or partner, with the expected orders per rider per hour
  • Delivery radius in kilometres and the pin codes inside it
  • Who sets item prices, and who pays for discounts
  • Cash on delivery: yes or no, and who reconciles it
  • Your FSSAI licence number, or the plan to get one

Version 1: the order loop first, growth tools later

What to build first, and what to leave for the second release
ModuleFirst versionAdd laterWhy
Customer orderingBrowse, cart, UPI, order trackingGroup orders, scheduled ordersRepeat orders prove the loop. Extras come after.
Menu and add-onsHalf and full portions, required add-ons, veg filterCombos, time-based menusIndian menus need portions and add-ons from day one.
Kitchen screenAccept timer, prep time, item sold-out switch, ticket printMulti-outlet routing, inventoryA kitchen that cannot accept in seconds sinks the app.
Rider appOffers, pickup, drop, earningsOrder batching, incentivesBatching needs enough orders to be worth the rules.
PaymentsUPI and card through a licensed gateway, cash on deliveryWallet, saved cardsWallets add compliance work. Start with the gateway.
Ops dashboardLive orders, delays, refunds, riders onlineDemand heatmaps, forecastingSupport needs this on launch day.
NotificationsPush, SMS, WhatsApp statusMarketing campaignsStatus messages stop the where-is-my-order calls.
Restaurant payoutsWeekly statement, manual transferAutomated split settlementManual weekly payouts work for ten kitchens.
PromotionsOne coupon typeReferrals, loyalty tiersDiscounts hide whether the unit economics work.

Ten to twelve weeks, and what you hand us at each stage

Typical first version. Gateway and store approvals often take longer than code, so start them in week 3.
WeeksWe buildYou supply
1–2Flows, wireframes, fee and bill rules, clickable ordering prototypeMenus, delivery area, fee decisions, brand assets
3–5Customer app, menu engine, cart and billDish photos, prices, descriptions in each language, test restaurant
6–8Kitchen screen and rider app, live order stateTwo or three test riders with their own Android phones, a thermal printer if you want tickets
9–10Payments, notifications, ops dashboardPayment gateway account and KYC documents, SMS and WhatsApp sender approvals
11–12Test with real kitchens and riders, store listings, launchFSSAI licence number, Play Store developer account, privacy policy sign-off

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: what goes in, and who tests it

The beta starts around week 9, once an order can travel customer → kitchen → rider → doorstep without anyone fixing it by hand. It is a small, real service, not a demo.

  • What's in it. Ordering with UPI and cash on delivery, the kitchen accept screen with its printer, the rider app (offer, pickup, drop, earnings) and the ops board. Nothing else.
  • What stays out. Extra coupon types, group and scheduled orders, rider batching, loyalty. They make the beta harder to read and don't tell you whether the loop works.
  • Who uses it. One or two real kitchens, three riders on their own Android phones, and 8–10 customers who live inside your delivery radius: regulars, an office lunch group. You recruit them; friends who will never order again tell you nothing.
  • How they get it. Customers install it through TestFlight or Google Play testing like a normal app. Riders get the Android build. The kitchen screen runs on the actual tablet and printer it will use.
  • How long. One to two weeks, real food and real money, limited to one service (say lunch) so everyone can watch every order. A feedback button on each screen sends a screenshot and a note to the shared doc.

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 beta group towards that, or use an organisation account.

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

We don't ask "do you like it?". We hand each person a task, on their own phone, in the language they use, and watch without helping. Five sessions per role usually surface the big problems; we fix them and test again.

The kitchen test happens in a real rush, not a quiet afternoon.
WhoThe taskWhat we watch for
Customer"Order a half biryani with extra raita and pay by UPI."Do they find portions and add-ons? Do they see the full bill before paying? Does the UPI app hand back to ours on a mid-range phone?
Repeat customer"Order the same lunch as yesterday."Can they do it in under a minute without searching?
Kitchen staff"Three orders arrive at once during service. Accept them and mark one dish sold out."Accepted before the timer runs out? Readable from a metre away, with wet hands, over kitchen noise?
Rider"Pick up two orders and deliver one to a gate with no house number."Do they find the pickup counter, reach the customer without seeing their number, and mark cash collected?
Ops"A customer says the food was late and cold. Refund half."How fast do they find the order? Is the refund rule clear, and who pays for it?

The numbers that say it's working

In week 1 you pick two or three of these, with today's figure if you have one (from phone or WhatsApp orders, say). We add the tracking before the beta, report at launch and again 30 days later. We don't promise the numbers; we make them visible so you can decide what to change.

GoalWhat we measureWhere it comes from
Customers order without helpShare of test tasks finished unaidedTesting sessions
The kitchen keeps upTime to accept an order; share of orders rejectedKitchen screen
Food arrives when promisedPromised versus actual delivery timeRider app timestamps
Checkout doesn't leakDrop-off from cart to paid, by payment methodApp analytics
People come backFirst-time customers who order again within 14 daysOrder data
Fewer things go wrongRefunds and "where is my order" calls per 100 ordersOps dashboard

What a food delivery app costs, and what moves it

Using the app cost guide, a single-brand app with a customer app, kitchen screen and rider app sits in the two user types plus admin band, ballpark ₹10–25 lakh from a small studio. A multi-restaurant marketplace with live maps and payments sits in the ₹25–60 lakh band. Those are ballparks, and we quote against your brief.

  • Marketplace versus single brand. Onboarding, commission and payouts are most of the gap between the two bands.
  • Live map tracking. Status-only updates cost less than a moving rider on a map, and cost far less to run.
  • Own riders versus partner. Your own riders need an app; a partner needs an integration instead.
  • Languages. Each language is content to translate and test, not just a switch. See the regional language guide.

Running costs are separate from the build: maps, SMS, gateway fees and hosting all grow with orders.

Food, tax and payment rules that shape the design

  • FSSAI. The 14-digit licence or registration number has to appear on bills and receipts, and FSSAI has told platforms not to list a food business that does not show a valid licence. Build a field for it at onboarding and print it on every invoice. Which licence category you need depends on your size, so confirm that on the FSSAI portal.
  • GST on marketplace orders. For restaurant service supplied through an e-commerce operator, the operator pays the 5% GST under section 9(5), in cash with no input tax credit. If you are the operator, your invoicing and returns have to reflect that. If you run a single-brand app selling your own food, your position is different. Ask a CA.
  • Collecting money for others. If your platform collects customer payments and settles them to many restaurants, RBI's Payment Aggregator Directions of September 2025 apply to whoever does that. Most startups route this through a licensed gateway instead of holding funds themselves.
  • Complaints. The Consumer Protection (E-Commerce) Rules expect a named grievance officer who acknowledges within 48 hours and resolves within a month. The rules were amended in 2026, so check the current text.
  • Personal data. Phone numbers, addresses and order history are personal data. The DPDP Rules roll out in phases over 18 months from November 2025, so build consent and deletion in now. See the DPDP checklist.
  • Rider work. Social security rules for platform workers have been changing. Ask a lawyer how they apply to your rider model before you set pay.

This is general information for planning, not legal or tax advice. Rates and rules change; confirm with a CA or lawyer before you launch.

Seven mistakes that sink delivery apps, and what to do instead

  • Launching in the whole city. Thin rider coverage means slow orders and bad reviews. Pick one area and fill it.
  • Skipping cash on delivery. Many first orders are cash. Offer it with a rider confirmation step.
  • Hiding fees until checkout. Customers abandon at surprise charges. Show the bill early, line by line.
  • Treating the kitchen screen as an afterthought. Kitchens decide acceptance speed. Test the screen in a real dinner rush.
  • Testing only on new phones. Riders use older Android handsets. Test there, on weak data.
  • Rider pay that ignores waiting. Unpaid waiting at the counter drives riders off. Pay for it or limit it.
  • No refund and dispute rules. Decide who pays when food is late or wrong before the first complaint, and show it in the ops dashboard.

Instead, start with one cuisine or one neighbourhood, a real test kitchen and three real riders, and watch the order loop for two weeks before adding a second area. If you want to see the flow first, the two-week concept sprint gives a clickable prototype. A cloud-kitchen founder should also read the cloud kitchen playbook.

Cost and timeline

10–12 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.

  • One restaurant or many: a marketplace needs onboarding, payouts and commission settings
  • Whether you run your own riders or use a delivery partner
  • Live map tracking versus status-only updates
  • Number of languages in the customer and rider apps
  • Integrations with your POS, printer or accounting software

How app costs work in India →

Questions

About your food delivery app.

How much does a food delivery app cost to build in India?

It depends mostly on how many sides you need and whether you run a single kitchen or a marketplace. We give a fixed quote within 48 hours of your brief, with scope and timeline in writing. Our app cost guide explains the drivers, and the food delivery guide covers the rest.

How long does it take to launch?

A full first version with the customer app, kitchen screen, rider app and ops dashboard usually takes 10 to 12 weeks. If you need to see it before you raise or commit, a two-week concept sprint gets you a clickable prototype of the ordering flow.

Will it work on cheap Android phones and slow data?

Yes, we design for that from the start. Screens are light, images are compressed, and the rider app keeps working through patchy coverage and syncs when the signal returns. We test on low-end Android handsets, not just new flagships. Order history and menus are cached too.

Can a single restaurant have its own delivery app instead of a marketplace?

Yes. A single-brand app is simpler: one menu, one kitchen screen and no commission settings. Many restaurants start there to move repeat customers off aggregators, then add outlets later. Our restaurant website work covers the dine-in side too. Your customer list stays with you.

Who owns the app and the code?

You do. Source code, design files and the brand system are handed over in full, in repositories your team or another developer can keep working in. There is no licence fee to keep using what we built. We also document how the order flow works, so a new developer can pick it up quickly.

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