ONDC Integration Guide for Indian E-commerce (2026)

ONDC Integration Guide for Indian E-commerce (2026)

ONDC went from government announcement to real commerce infrastructure remarkably fast. In 2026, it's no longer a "wait and see" project, food delivery, groceries, and mobility are running significant volume on it, and brands that figured out the integration early are winning channel independence.

I'm Ashish Sharma, founder of Codingclave. We've built ONDC integrations for 6 Indian brands since 2024, two restaurants, a grocery chain, a beauty D2C, a home services brand, and a B2B logistics aggregator. This guide captures what we've learned from production rollouts, not from the ONDC pitch deck.

---

What ONDC Actually Is

ONDC is a protocol, not a platform. The distinction matters.

Amazon is a platform, a single company runs the buyer and seller experience. You list on Amazon, customers buy on Amazon, Amazon takes its cut.

ONDC is a protocol, a standardized way for any seller-facing software (called a Seller Network Participant, or SNP) to talk to any buyer-facing software (a Buyer Network Participant, or BNP). You list on one SNP and your products are discoverable on every BNP connected to the network.

In practice:

• A restaurant uses Petpooja or a custom seller app as their SNP
• Their menu becomes visible on Paytm, PhonePe, MagicPin, Dunzo, and the ONDC Store
• When a customer orders from any of those, the order flows back through ONDC → SNP → restaurant
• Fulfillment happens via an LSP (Logistics Service Provider) integrated into the flow

The customer never sees "ONDC" as a brand. They use the buyer app they trust. The seller never builds a marketplace. Everyone talks to each other through the protocol.

---

The 2026 State of ONDC

Honest numbers and what they mean:

• 15M+ cumulative orders: sounds big, but spread across 3+ years. Monthly run rate varies: F&B does hundreds of thousands of orders/month; some other categories do tens or low hundreds.
• 30+ active buyer apps: Paytm, PhonePe, MagicPin, Mystore, Mythos, Ola, Namma Yatri, Costn, ONDC Store, and many niche apps.
• 400+ active seller apps: from custom enterprise deployments (Thomas Cook, Kotak Mahindra for financial services) to SME-focused SNPs (Mystore, eSamudaay).
• Geographic strength: Bangalore and Kochi lead for mobility; Delhi NCR and Mumbai lead for F&B; Kerala leads for grocery; Lucknow and smaller cities have growing D2C presence.
• Cost advantage: verified lower take rate than marketplaces (1-5% vs 8-25%), BUT only if you handle fulfillment well.

The realistic takeaway for a new brand in 2026: ONDC won't replace your Shopify store (or whichever ecommerce platform in India you build on), but it adds a cheap distribution channel that works especially well for hyperlocal categories.

---

Who Should Integrate with ONDC in 2026

ONDC works well for:

• F&B brands with real-time menu + inventory: restaurants, cloud kitchens, dessert brands
• Hyperlocal grocers and kiranas: especially in tier-2 and tier-3 cities
• D2C brands with 15%+ gross margins: skincare, fashion, home goods, health products
• Service aggregators: salon, repair, home cleaning, laundry
• B2B suppliers: logistics, wholesale goods, industrial supplies

It's less useful for:

• Very low-margin categories where 1-3% ONDC fee + 3-8% fulfillment still beats you
• Single-SKU brands without depth of catalog
• Luxury or high-ticket items where discovery happens on Google + Instagram, not ONDC buyer apps
• Software / digital goods: ONDC is built for physical commerce primarily

If your unit economics allow you to absorb 3-5% channel cost, ONDC is worth testing. If they don't, skip it and double down on direct channels.

---

The Two Integration Paths

Path 1: Partner with an existing SNP (recommended for 95% of brands)

Fastest, cheapest, lowest risk. You sign up with an SNP like Mystore, eSamudaay, ShopX, or Petpooja (for F&B). They already have ONDC certification. You list your catalog in their dashboard, and your products become discoverable across all connected buyer apps.

Timeline: 1-4 weeks for onboarding + menu/catalog creation Cost: ₹0-₹50,000 setup, then transaction fees (1-3% typically) Tradeoff: You're on the SNP's rails, their onboarding, their dashboard, their support quality. If they change pricing or go offline, you're affected.

Path 2: Build your own SNP (for brands with specific technical or scale needs)

You build your own seller network participant from scratch using ONDC's published protocol specs. You go through ONDC's certification process, register with ONDC governance, maintain your own signing certificates, handle all protocol interactions.

Timeline: 3-5 months for MVP, 6-12 months for production-grade Cost: ₹5,00,000 – ₹25,00,000+ build, ₹15,000-₹50,000/month infrastructure Tradeoff: Full brand control, custom workflows, your data. But you're responsible for protocol compliance updates, certificate rotation, and keeping up with ONDC spec versions.

Build your own SNP only if: you have 50+ sellers you'd onboard, you want to white-label to a network (e.g., a mall group), or you have unique fulfillment/inventory logic that existing SNPs can't accommodate.

---

The Technical Flow (Path 2 deep dive)

If you're going the build-your-own route, the protocol stack looks like this:

Core actions (Beckn Protocol)

• search: buyer asks for products matching a query/location
• on_search: your SNP responds with catalog items
• select: buyer picks items, asks for availability + cost
• on_select: your SNP confirms availability and returns quote
• init: buyer commits to an intent to pay
• on_init: your SNP generates order + payment details
• confirm: payment is made, order is confirmed
• on_confirm: your SNP acknowledges and creates fulfillment
• status + on_status: order state transitions (packed, shipped, delivered)
• track: live location tracking via LSP
• cancel / update / rating / support: lifecycle management

Message structure

Every message is a JSON payload with three blocks:

• context: protocol version, transaction ID, timestamps, bap_id/bpp_id
• message: the actual action payload (catalog, order, cart, etc.)
• signature: Ed25519 digital signature over the body

Signatures prevent tampering. Every participant maintains a keypair registered in ONDC's Registry. The Registry is the DNS of ONDC, it tells participants how to verify each other.

Registry + Gateway

ONDC publishes a Registry of all participants with their public keys and capabilities. Your seller app queries the Registry to validate incoming requests and find the right buyer app for callbacks.

The Gateway routes buyer-app searches to multiple SNPs efficiently, like a search fan-out service. For most actions after search, communication is direct between SNP and BNP.

Async vs sync

ONDC is async by design. A search triggers on_search callbacks from multiple SNPs, which can take seconds to minutes. Your architecture must:

• Accept callbacks independently of original request
• Deduplicate responses (a single search can generate 10+ on_search calls from different SNPs)
• Handle out-of-order events gracefully
• Queue and retry on transient failures

This is fundamentally different from the request-response world most web developers are used to. Designing for async is the #1 skill gap we see on inherited ONDC projects.

---

What a Production ONDC Integration Actually Includes

Beyond the minimum protocol compliance, a real production integration needs:

1. Catalog management: real-time inventory, pricing, availability across SKUs, multi-location support
2. Order management: state machine for every order (created → confirmed → packed → dispatched → delivered → rated/returned)
3. Fulfillment orchestration: either in-house rider pool or LSP integration (Shiprocket, Dunzo, Zomato Delivery Rider)
4. Payment reconciliation: daily settlement with buyer apps, mapping ONDC IDs to your internal order IDs
5. Returns & cancellations: full protocol compliance including partial returns
6. Dispute management: ONDC has a grievance framework (IGM), your app must handle it
7. Tax compliance: GST invoicing via your existing billing integration
8. Analytics: cross-buyer-app performance, hour-of-day patterns, fulfillment SLA tracking
9. Multi-tenant support: if you're building for multiple sellers, tenant isolation becomes important
10. Audit logging: ONDC requires certain audit trails for regulatory reasons

Missing any one of these creates real business problems within 3-6 months of launch.

---

Cost Breakdown for an Indian ONDC Seller App (Path 2)

A typical build-your-own SNP for an Indian brand with 100-500 SKUs, single category (e.g., grocery or F&B):

Component	Effort	Cost (INR)
Discovery + architecture	1-2 weeks	₹50,000 – ₹1,50,000
ONDC protocol layer (search, select, init, confirm, status)	4-6 weeks	₹2,50,000 – ₹4,50,000
Catalog + inventory management	2-3 weeks	₹1,00,000 – ₹2,00,000
Order management + state machine	3-4 weeks	₹1,50,000 – ₹3,00,000
Fulfillment + LSP integration	2-3 weeks	₹1,00,000 – ₹2,00,000
Payment reconciliation	1-2 weeks	₹50,000 – ₹1,00,000
Returns + IGM	2-3 weeks	₹1,00,000 – ₹2,00,000
Admin dashboard for sellers	3-4 weeks	₹1,50,000 – ₹3,00,000
Certification + ONDC approval	2-4 weeks	₹50,000 – ₹1,50,000
QA, staging, UAT, go-live	2-3 weeks	₹1,00,000 – ₹2,00,000
Total	22-34 weeks calendar	₹10,00,000 – ₹22,50,000

Add 20-30% for multi-category support, multi-language support, or enterprise features like SAP/Tally integration.

---

Ongoing Costs

Running an SNP in production isn't free. Typical monthly costs for a mid-sized Indian SNP:

Item	Cost (INR/month)
Server infrastructure (AWS/GCP)	₹15,000 – ₹60,000
ONDC signing certificate renewal (annualized)	₹1,500 – ₹3,000
Monitoring + error tracking	₹3,000 – ₹10,000
Protocol spec updates (developer maintenance)	₹10,000 – ₹40,000
Dispute resolution staff (IGM)	₹15,000 – ₹50,000 per FTE
Total	₹45,000 – ₹1,65,000/month

Build these into your business case before committing to build-your-own.

---

Common Mistakes We See on Inherited ONDC Projects

After auditing projects built by other teams, the recurring issues:

1. Synchronous-only code paths: breaks at volume because ONDC is fundamentally async
2. Missing signature verification: allows message forgery, would fail audit
3. No idempotency: duplicate on_confirm callbacks create duplicate orders
4. Catalog hardcoded, not inventory-synced: out-of-stock orders frustrate customers
5. No protocol version abstraction: ONDC has shipped breaking spec updates; code tied to one version breaks later
6. LSP chosen blindly: some LSPs don't cover tier-3 cities or specific pin codes where your orders come from
7. IGM not implemented: disputes pile up, ONDC governance flags the participant, orders get throttled

Every one of these has cost real money on audits we've done.

---

Our Recommended Approach for Most Brands

For 95% of Indian brands I talk to:

1. Start with Path 1: partner with an established SNP (Mystore for general retail, Petpooja for restaurants, eSamudaay for hyperlocal)
2. Run for 3-6 months: learn your actual order volume, which buyer apps drive orders, what categories work
3. Evaluate Path 2 only if you find specific reasons: custom workflows, multi-brand aggregation, 100+ merchants to onboard, or ONDC is > 30% of your revenue

Jumping straight to build-your-own costs real money and delays you from learning what matters.

---

How ONDC Fits with Your Existing Channels

ONDC is a channel, not a strategy. It should sit alongside:

• Direct D2C site: your own store remains your highest-margin channel
• Marketplaces: Amazon, Flipkart, Myntra for scale and discovery
• Q-commerce: Zepto, Blinkit, Swiggy Instamart for instant delivery categories
• Social commerce: Instagram, Meesho, Glowroad
• ONDC: hyperlocal + multi-buyer-app reach

For most Indian brands, ONDC should be 5-20% of revenue. Not more, not less. Less means you're leaving free distribution on the table; more usually means your direct channel is broken.

---

Compliance and Regulatory Notes

A few things worth knowing:

• GST: every ONDC order needs a GST-compliant invoice. Our GST-compliant software guide covers what's required.
• Consumer Protection (E-Commerce) Rules: ONDC participants are subject to the same disclosure, returns, and grievance rules as any e-commerce seller.
• Data Protection (DPDP Act): customer data flowing through ONDC must comply with Indian data protection law. Your SNP must have legitimate-use basis and proper retention policies.
• Payment Aggregator rules: RBI guidelines apply to how funds flow. Most brands use their existing PA (Razorpay, Cashfree), see our API integration services guide for details.

A 30-minute compliance review at project start saves painful retrofits later.

---

Getting Started

If you're evaluating ONDC for your Indian business:

1. Free scoping call (30 min): we understand your category, volume goals, and whether ONDC is right for you. Book via free consultation.
2. SNP partner evaluation: we help you shortlist and onboard with the right existing SNP (Path 1) if that fits.
3. Custom SNP build: if you have genuine reasons for Path 2, see our custom software development services for full scoping.

For a detailed estimate on your specific ONDC scope, use get a free quote or contact us directly.

---

Summary

• ONDC is production-ready in 2026 for specific categories (F&B, grocery, hyperlocal services, D2C)
• Start with Path 1 (existing SNP) unless you have scale or customization reasons for Path 2
• Budget ₹10-25 lakh if you build your own seller app; ₹0-50K if you partner with an SNP
• Don't build your strategy around ONDC alone, treat it as one channel among several
• Protocol is async, signed, and governance-heavy, this is not a weekend integration

If you want a no-pitch conversation about whether ONDC makes sense for your business, reach out. Happy to spend 30 minutes looking at your specific case.