India is not one food delivery market. It is a set of city markets with different cuisines, price points, payment habits and competitor density. Swiggy and Zomato dominate the metros; smaller cities and niche categories still have room for a focused operator.
The winning strategy is rarely “Uber Eats for all of India”. It is one city, one cuisine or one customer segment, with dense delivery and a lower commission than the incumbents.
| Decision | Metro | Tier-2 / tier-3 |
|---|---|---|
| Competition | High; deep discounts | Lower; weaker supply |
| Customer acquisition | Paid ads, offers | Local word of mouth, campus |
| Restaurant supply | Dense, negotiated | Fewer organised vendors |
| Rider supply | Available but expensive | Cheaper, less experienced |
| Average order value | Higher | Lower; price-sensitive |
| Best entry | Niche cuisine or corporate | First-mover in one zone |
Choose a city and a wedge
A metro launch competes with two well-funded apps on selection, discounts and rider density. A niche can still work: late-night food, healthy meals, a specific regional cuisine, corporate lunches or a campus. A tier-2 city can work with a broader menu if the incumbent supply is weak.
Do not launch three cities at once. Pick one city, one dense zone and one customer promise.
Licences and compliance
Food businesses need an FSSAI registration or licence depending on turnover and activity. A marketplace may also need trade licences, GST registration and local shop-and-establishment registrations. Tax and food-safety rules vary by state. Get local professional advice; this is not legal or tax advice.
Payments: UPI, cards and COD
UPI is the default for many customers. Cards, wallets, net banking and cash on delivery still matter in different segments. Integrate a gateway that supports UPI and cards, and build a cash reconciliation process if you offer COD. Do not assume a global card gateway covers the local methods.
Unit economics
Delivery cost is the hard constraint. A 3–5 km delivery in traffic can take 20–30 minutes. If the rider cost is ₹40–₹80 per delivery and the commission on a ₹300 order is ₹45–₹60, the math is tight. You need density, batching or a delivery fee to make it work.
Model the contribution per order before launch. A cheap launch with negative contribution per order is not a business; it is a subsidy.
Restaurant supply
Start with 25–40 restaurants in one zone. Offer a lower commission than the incumbents, a clear payout cycle and a tablet/app they can use. Help with menu photography and onboarding. Restaurants are more likely to stay if the platform brings orders and pays on time.
Marketing in India
Hyperlocal Instagram, WhatsApp groups, campus ambassadors, apartment associations and local influencers work better than broad national ads. Use UPI or COD as a trust signal. Offer a first-order discount, but do not train customers to wait for deep discounts every week.
Software stack
You need customer, restaurant and rider apps, an admin panel, UPI/card/COD support, zones, commissions, offers and settlement reports. A $299 source-code licence can cover the core; budget for branding, gateway integration, app publishing and support.
See Foodomaa for India, the India cost guide, or the $299 licence.
India launch mistakes
- Launching across a metro instead of one dense zone.
- Ignoring COD reconciliation and GST from day one.
- Competing on discounts with Swiggy and Zomato instead of a niche.
Start with 25–40 restaurants in one zone, model contribution per order and fix cash handling before you spend on ads.
Tier-2 vs metro
A metro gives you restaurant density and rider supply, but also two well-funded competitors, expensive ads and discount fatigue. A tier-2 city gives you lower acquisition costs and weaker incumbent coverage, but fewer organised restaurants and lower basket sizes. The right choice depends on your supply relationships, not on city population.
| Factor | Metro | Tier-2 |
|---|---|---|
| Restaurant depth | High | Medium; needs recruitment |
| Rider supply | High but costly | Cheaper, less experienced |
| Customer acquisition | Paid ads, discounts | Local word of mouth, campus |
| Average basket | Higher | Lower |
| Competition | Swiggy, Zomato, local apps | Weaker organised delivery |
The licence and GST checklist
Food businesses need FSSAI registration or a licence depending on turnover and activity. A marketplace may also need GST registration, a trade licence and local shop-and-establishment registration. Tax rates and thresholds change; confirm with a local accountant. Build GST-compliant invoices and a reconciliation report before launch, not after the first tax notice.
Unit economics with UPI and COD
UPI is the default for many customers, but cash on delivery still matters in smaller cities and for first-time users. Model both. A UPI order has a gateway fee but cleaner reconciliation; a COD order has no gateway fee but requires a rider float, daily deposit and variance control. A 15% commission on a ₹300 order is ₹45; if the rider costs ₹45–₹70, the delivery fee and basket size decide whether the order is viable.
A 90-day launch plan
- Days 1–15: choose one city and one 3–5 km zone; interview 30 restaurants.
- Days 16–30: sign 25 restaurants; photograph menus; confirm FSSAI and GST details.
- Days 31–45: install software, configure UPI/cards/COD, zones and commissions.
- Days 46–60: recruit riders and run 30 test orders, including a refund and a failed delivery.
- Days 61–75: soft-launch with a first-order offer; monitor contribution daily.
- Days 76–90: add a second zone only if the first is stable and contribution-positive.
Mistakes that kill margin
- Launching city-wide instead of one dense zone.
- Ignoring COD reconciliation and GST invoicing.
- Matching Swiggy/Zomato discounts without their order density.
- Paying riders a fixed shift before demand exists.
- Choosing a card-only gateway that excludes UPI customers.
City shortlist: what to check
Shortlist three cities and score them on restaurant density, rider supply, payment habits, competitor coverage and your own contacts. Visit the target zone at lunch and dinner. Count the restaurants that already do takeaway, check how long a competitor takes to deliver and talk to riders. A city with a strong food culture but weak delivery coverage can be better than a metro where you have no supply relationships.
A payment mix example
Assume 60% UPI, 25% cards/wallets and 15% cash. UPI and cards settle through the gateway; cash needs a rider float and daily reconciliation. Model the fee and settlement timing for each. A payment method that converts but settles in seven days can strain cash flow in the first month.
Restaurant commission and payouts
Restaurants compare the net payout, not the commission percentage. Publish a payout calendar, send an order-level statement and pay on time. A 18% commission with reliable weekly payouts can be more attractive than a 12% commission with late settlements. Use a lower commission for self-delivery and a higher one for platform delivery; the cost difference is real.
Hiring and support
Your first hires should be operations, not more developers. A city manager who recruits restaurants and manages riders is worth more than a second engineer. Add support staff for meal peaks before you add a marketing manager. The platform is software; the business is supply, delivery and customer service.
A city-level cost table
India is not one cost base. A launch in Bengaluru, Indore and Guwahati has different rent, rider wages, restaurant expectations and customer willingness to pay. Build a city model before choosing the location; the difference can be larger than the software budget.
| Cost line | Metro | Tier-2 | Tier-3 |
|---|---|---|---|
| Rider cost per delivery | ₹70–₹110 | ₹50–₹80 | ₹40–₹65 |
| Average basket | ₹350–₹600 | ₹250–₹450 | ₹200–₹350 |
| Restaurant commission | 15–25% | 12–20% | 10–18% |
| Customer acquisition | ₹150–₹400 | ₹80–₹200 | ₹50–₹150 |
| Launch team | 6–12 people | 4–8 people | 3–6 people |
These are planning ranges, not quotes. Validate them with local riders, restaurants and a payroll provider. A metro with a ₹500 basket can absorb a ₹90 delivery cost; a tier-3 city with a ₹250 basket cannot, unless the customer pays a larger share or the order is batched.
FSSAI and GST registration sequence
Register the legal entity first, then the food business, then the tax and payment accounts. The order matters because gateways and restaurants ask for documents in sequence.
- Company or LLP: incorporation, PAN, bank account and authorised signatory.
- FSSAI: registration or licence based on turnover and activity; keep the certificate number visible.
- GST: registration if you cross the threshold or need input credit; confirm the place of supply for marketplace commissions and delivery fees.
- Shops and establishment: state-level registration where applicable.
- Payment gateway: merchant account with the entity documents and a refund policy.
- Restaurant agreements: FSSAI number, menu prices, commission and payout terms.
Do not accept a restaurant without a valid FSSAI registration. It protects customers and your platform. Keep the certificate on file with an expiry date and a reminder. This is operational compliance, not legal advice; confirm the details with a local chartered accountant.
UPI, cards and cash reconciliation
UPI is the default payment method in most Indian cities, but cash still matters for smaller orders and older customers. The reconciliation cadence should match the settlement cycle:
- Daily: gateway settlement file vs order list; flag missing, duplicate and refunded payments.
- Per shift: rider cash collected, handover amount and variance.
- Weekly: vendor payout statement, commission, refunds and adjustments.
- Monthly: GST report, gateway fees, chargebacks and unreconciled items.
Keep cash limits per rider low at launch. A rider carrying a day's collections is a fraud and safety risk. Require a handover at the end of every shift, and make the admin panel show expected cash before the rider leaves.
Language, support and local trust
English is not enough for every market. Hindi, Marathi, Tamil, Telugu, Kannada, Bengali and Malayalam may matter in the target city. The interface can be translated, but support scripts, WhatsApp replies and refund messages should be in the language the customer used. A customer who cannot explain a missing order will not reorder.
Local trust is built by visible restaurant names, a real address, a working phone number and a refund policy that is easy to find. Add the FSSAI number, grievance officer and payment options to the footer. These details do not replace marketing, but they remove hesitation at checkout.
A one-city launch budget
Put the city model on one page before you commit. The example below is a planning frame, not a quote; replace every number with local evidence.
| Line | Launch month | Monthly after launch |
|---|---|---|
| Software licence and setup | ₹25,000–₹2,00,000 | ₹5,000–₹20,000 hosting/support |
| Rider fleet (8 riders) | ₹1,20,000–₹2,00,000 | ₹1,20,000–₹2,00,000 |
| Restaurant recruitment and photography | ₹50,000–₹1,50,000 | ₹20,000–₹50,000 |
| Customer acquisition | ₹75,000–₹2,50,000 | ₹50,000–₹1,50,000 |
| Support and operations | ₹40,000–₹80,000 | ₹40,000–₹80,000 |
| Cash float and refunds | ₹30,000–₹75,000 | ₹15,000–₹40,000 |
The fleet is usually the largest fixed cost. If the zone cannot supply enough orders per rider shift, the model does not work, no matter how good the software is. Calculate orders per rider per shift at launch and at maturity. A rider who completes fewer than eight deliveries in a shift is being subsidised.
The first 30 days
- Days 1–7: register the entity, open the bank account and start FSSAI/GST applications.
- Days 8–14: sign the first 15 restaurants; collect FSSAI numbers, menus and bank details.
- Days 15–21: install and configure the platform; test UPI, cards, COD, refunds and settlements.
- Days 22–25: recruit and train riders; run 30 test orders across the zone.
- Days 26–30: soft-launch one customer segment; review every order manually and fix the top three failures before expanding.
Do not spend the first month chasing a city-wide launch. A dense zone with reliable delivery is the only proof that the model works. Everything else is a map with colours on it.
What to measure weekly
Track five numbers: orders per active restaurant, rider deliveries per shift, average delivery time, refund rate and contribution per order. A weekly review of those five is more useful than a monthly revenue report. If orders per restaurant fall while registrations rise, the launch is adding supply faster than demand; stop recruiting and fix demand.
Keep the definitions stable. Changing how you count an active restaurant or a delivery makes the trend unreadable. Write the definitions in the operating manual and use the same query every week.