Restaurant Own Delivery vs Swiggy & Zomato: 2026 Cost Guide

A practical guide for restaurant owners in 2026 — what "own delivery" actually costs, what breaks, and the middle path most people end up taking.

If you run a restaurant in India right now, you have almost certainly had this conversation with yourself at 11pm while looking at a settlement statement.

The commission is 25–30%. On a ₹500 order you're keeping around ₹350–390. Do 25 delivery orders a day and you are handing over well over a lakh a month to a platform that also decides when your food goes on discount. Meanwhile a rider costs a fraction of that. So why not just hire two boys, buy two second-hand bikes, and keep the money?

It's a fair question. Restaurant bodies are asking a sharper version of it — in Bengaluru, a group of restaurants threatened to disengage from Swiggy and Zomato entirely in August 2026, and the deadline was pushed to the end of the month while both platforms agreed to an opt-in mechanism requiring restaurant approval before discounts run on their menu.

So the leverage is real. But before you buy the bikes, there are three things worth understanding, because most restaurants that try in-house delivery quit within six months — and it's almost never because of the riders.

Problem 1: You don't have a delivery problem. You have a demand problem.

This is the part people get wrong.

When you leave an aggregator, you don't just lose the delivery. You lose the customer who was going to find you. Those orders weren't yours — they belonged to the app. Someone opened Swiggy hungry, saw your biryani, and tapped.

Your own delivery capability doesn't generate a single order. It only saves money on orders you already have.

The honest test: how many of your delivery customers would call, WhatsApp, or open your link if the app disappeared tomorrow? If that number is small, in-house delivery will just give you idle riders. If you have a genuine repeat base — a neighbourhood following, a strong takeaway counter, a WhatsApp list, regulars who ask for your number — the economics change completely.

If you need proof of how hard the demand side is: a well-funded, Jubilant FoodWorks-backed platform called Thrive built a restaurant-first ordering app with ~9,500 restaurants across 76 cities on a 3% commission model, and its consumer app served its last order in December 2024. Ola pushed food delivery across about 100 cities and quietly pulled the plug in December 2025. Fixing fulfilment is solvable. Manufacturing demand is brutal.

Conclusion: don't think of this as leaving the aggregators. Think of it as building a second, cheaper channel alongside them, and slowly shifting your regulars to it.

Problem 2: Your own riders are a fixed cost. Orders are not.

Here's the arithmetic that decides everything, and you can do it on the back of a bill book tonight.

An in-house rider costs you the same whether he does 8 deliveries or 28. Salary, fuel, phone, insurance, and the shifts you have to cover on his day off — that's a fixed monthly number.

An on-demand rider from a third-party fleet costs you nothing when there are no orders. As a reference point, hyperlocal delivery services advertise rates starting around ₹10 per km with per-booking prices from about ₹45.

So work out your own break-even:

**Cost per delivery with own rider** = (rider's fully-loaded monthly cost) ÷ (deliveries he actually completes in a month) Then compare it to your per-order cost from a third-party fleet at your average delivery distance.

Two things almost always fall out of this exercise:

1. In-house wins only at high, predictable density. If one rider genuinely does 20+ deliveries a day within a 3 km radius, own fleet is cheaper. That's a busy single-kitchen QSR, not a mid-sized casual dining place doing 15 orders spread over lunch and dinner.
2. Your peak is the killer. You don't need average capacity, you need Friday 8:30pm capacity. Staffing your own fleet for peak means paying for that capacity all week. This is precisely why almost every restaurant that scales delivery ends up hybrid: a small owned fleet for the base load, third-party riders for the spike.

There's one cost people forget: pickup time. Your rider isn't paid to deliver, he's paid to wait — in your kitchen while the order finishes, then again at the customer's gate. Fleet platforms know this; some explicitly pay riders a first-mile component for the distance travelled to collect an order. If your kitchen runs 12 minutes late at peak, that inefficiency is now on your payroll instead of theirs.

Problem 3: Someone has to decide which rider gets which order

This is the invisible work, and it's what actually exhausts restaurant owners.

At 8pm on a Saturday you have four orders ready, two riders out, one rider who says he's 10 minutes away, and a customer calling to ask where his food is. Somebody has to make those decisions in real time, order after order, every night. If that somebody is you or your manager, you have quietly added a full-time operations job to a business that already has too many.

This is the exact problem the delivery integrator category exists to solve — and it's the piece most restaurant owners have never heard of.

The middle path: keep the money, skip the fleet

There is a layer of companies that sit between your kitchen and many delivery fleets. You send them the order; they decide which rider from which fleet picks it up, at what price, and they retry when the first one fails. You never sign a contract with an individual fleet, never manage riders, never do the 8pm triage.

Part of what makes this cheap now is ONDC — the government-backed open network that unbundles food delivery into separate pieces. Discovery, ordering and delivery become three separate purchases instead of one 30% bundle. Guides written for restaurateurs in 2026 put total channel cost on ONDC at roughly 3–5% versus 25–30% on the big apps — on a ₹500 order that's about ₹85–135 more staying in your pocket. Settlement is also typically faster, commonly 1–3 business days.

Here's the practical shortlist, in plain terms:

If you want everything from one vendor — your own ordering app plus ONDC listing plus riders: uEngage (Chandigarh) is the most restaurant-native option. Its Edge product gives you a branded ordering site/app, Flash handles rider and third-party fleet allocation from one dashboard, and it onboards you to ONDC. It integrates with the POS you probably already run — Petpooja, Restroworks/Posist, Billberry, UrbanPiper. Free entry tier. Smaller company, so ask hard questions about support in your city.

If you mainly want the delivery muscle and demand on the network: magicpin is the biggest player on ONDC food by a wide margin and now calls itself India's third-largest food delivery app. It runs Velocity (delivery-as-a-service aggregating fleets like Shadowfax, Rapido, Porter, Zypp) and MagicFleet, which crossed a million deliveries a month by mid-2025 with 100,000+ delivery partners across Delhi NCR, Bengaluru, Mumbai, Pune, Hyderabad, Chennai and Kolkata. Be clear-eyed: it is also a food delivery app competing for the same customer.

If you want a neutral, no-conflict delivery layer: Pidge (Gurugram) sells only orchestration — it owns no consumer app. It claims 20,000+ brands across 50+ cities and works with F&B names including KFC and EatClub, and it raised ₹120 crore in late 2025, so it's reasonably funded. Good choice for multi-outlet operators who want one API and honest allocation.

If you want the biggest balance sheet behind you: Shiprocket Quick is live in roughly a dozen cities including Hyderabad, aggregating Ola, Shadowfax's Flash, Loadshare and Borzo, with 2/3/4-wheeler options and POS/API integration. Shiprocket listed on the exchanges in August 2026, so it's the most financially durable vendor in this set. Hyperlocal is one line in a much bigger business, so restaurant-specific workflow is thinner.

There is also a smaller, cheaper tier — ProRouting in Bengaluru being a good example, bootstrapped and built by a team that has been solving hyperlocal orchestration since before ONDC existed. Great for a pilot or a single outlet. Don't make it your only option for a five-outlet chain.

Before you sign up for anything new: check your billing software

This is the shortcut most owners miss. You may already be paying for half of this.

Most Indian restaurant POS companies have already wired up ONDC, third-party riders, or both — usually as a switch you can turn on rather than a new vendor you have to onboard. Here's what the POS players themselves claim, with links so you can check the current version before you call your account manager:

Petpooja has a dedicated ONDC page pitching the network as a way to reach more ordering and delivery aggregators at lower commission, with ONDC access available through the POS you already use (petpooja.com/ondc). On the delivery side, it says its billing software supports 150+ third-party integrators including ride-management apps like Dunzo, Shadowfax and Pidge, and charges no commission of its own for connecting a third party — you only pay the aggregator (online order management, integrations directory). Its ONDC work goes back to an early pilot with GrowthFalcons in which a Paytm order landed on the restaurant's Petpooja POS and was delivered by LoadShare.

UrbanPiper took a different route: rather than building its own network presence, it announced that it enables merchants to go live on ONDC through partners uEngage and Bitsila (announcement). Practically, that means if you're on UrbanPiper you reach ONDC via a partner's registration, not UrbanPiper's.

Chefdesk provides ready to use online web channel and AI based whatsapp integration that connects buyer and delivery partner for last mile delivery. You incur only delivery fees with no commission.

Rista (by DotPe) lists Swiggy, Zomato and ONDC as order sources feeding a single queue, with auto-acceptance and automatic stock-outs across channels (product page).

uEngage publishes its integration list openly, and it's the most useful single page in this whole category: POS integrations with Petpooja, POSist/Restroworks and Rista, and logistics partners including Shadowfax, Shiprocket, Porter, Ola Logistics and LoadShare (uengage.io/partners). If you run POSist or Billberry, this is typically how you get to ONDC and to third-party riders.

Bitsila is POS and ONDC seller app in one, selling across website, QR, aggregators, ONDC, WhatsApp and in-store, with self-delivery, 3PL and ONDC fulfilment as options (bitsila.com). It was acquired by Pai Platforms in 2024.

Gofrugal matters if you also run a retail or grocery counter alongside the kitchen. It claims to be the first ERP to integrate with ONDC, and its GoDeliver app covers delivery management plus ONDC network integration, pushing your catalogue to consumer apps like Paytm and PhonePe (POS software, integrations, company timeline).

Two honest caveats about that list.

First, everything above is a company claim on its own website. Vendor marketing pages are the least reliable part of this research — they describe what has been built at some point, not necessarily what is live in your city this month. Treat them as a reason to ask a specific question, not as an answer.

Second, and more concretely: some POS solution integrations page still names Dunzo**, whose app and website went offline after its co-founder left. That single detail tells you how to read every partner list in this industry. So when you call your POS provider, ask exactly one question:

*"Which delivery fleets are live and contracted in my city today, on my plan — and can you show me a recent order that went through one?"*

If they can't answer that in a single reply, the integration exists in a brochure, not in your kitchen.

And one question specific to the ONDC piece: whose network registration are you listing under? If it's your POS vendor's or their partner's, find out whether your listing, ratings and order history are portable when you switch POS. Being locked into a billing system by your delivery channel is a worse trap than the commission you're trying to escape.

One thing to check before you sign anything

Ask every vendor for a live, per-city list of which delivery fleets are actually contracted and running today — not which ones are "available on the network."

Two more questions worth asking:

• Who owns the failed delivery? When food is dropped, delayed or damaged, who refunds the customer — you, the fleet, or the platform? Get this in writing.
• Is my ONDC presence portable? If you list through a vendor and later want to leave, can you take your listing with you, or does it live under their registration?

What I'd actually do, in order

1. Build the demand channel first, before touching delivery. QR code on every table and every bill. A WhatsApp number that takes orders. Your own ordering link in your Instagram bio. Ask takeaway customers for their number. Do this for 60 days and measure how many direct orders you get. This costs almost nothing and answers the only question that matters.
2. Call your POS provider before you call anyone else. If you're on Petpooja, Rista, UrbanPiper, POSist or Gofrugal, ONDC and third-party riders are likely a switch on your existing plan. Cheapest possible starting point, and no new vendor relationship.
3. Turn on ONDC through a seller app (uEngage or magicpin are the obvious starting points; if you're on Petpooja, ONDC is available from inside the POS you already use). Treat it as a cheap extra channel, not a replacement. Realistically it will be 5–20% of your orders at first.
4. Use a third-party fleet through an integrator for those orders. Zero fixed cost, no rider management, no 8pm triage. Track your actual cost per delivery for a month.
5. Only then consider hiring riders — and only if the numbers say one rider will consistently do 20+ deliveries a day in a tight radius. Start with one, not four. Keep the integrator switched on for peaks and for his day off.
6. Stay on the aggregators while you do all this. They're your discovery engine. The goal isn't a dramatic exit; it's shifting your repeat customers to a channel where you keep ₹480 instead of ₹370, one regular at a time.

The honest summary

Doing your own deliveries is rarely about riders and bikes. It's about whether customers will come to you directly, and whether you want to add a nightly dispatch job to your life.

The good news is that in 2026 you no longer have to choose between paying 30% and running a fleet. There is a genuine middle option: your own ordering channel, a low-commission network listing, and on-demand riders you don't employ — with someone else's software deciding who picks up what. In many cases it's already sitting unused inside the POS on your counter.

Start with demand. Check your billing software. Rent the riders. Buy the bikes last, if ever.

Frequently asked questions

Is it cheaper for a restaurant to run its own delivery fleet in India? Only at high, predictable order density. An in-house rider is a fixed monthly cost whether he completes 8 deliveries or 28, while a third-party rider costs nothing on a slow Tuesday. Divide your rider's fully loaded monthly cost by the deliveries he actually completes, then compare that to a per-order third-party rate at your average distance. Most single-outlet restaurants doing fewer than roughly 20 deliveries a day in a tight radius are cheaper off with on-demand riders.

How much commission does ONDC charge restaurants compared to Swiggy and Zomato? Guidance published for restaurateurs in 2026 puts total ONDC channel cost at roughly 3–5%, against 25–30% on the major aggregators. On a ₹500 order that's about ₹85–135 more retained. Settlement is also typically faster, commonly 1–3 business days.

Can I leave Swiggy and Zomato entirely? You can, but you'd be giving up your discovery engine, not just your delivery. Those customers belong to the app. The realistic approach is to run a cheaper direct and ONDC channel alongside the aggregators and migrate your repeat customers over time.

Who delivers my food if I list on ONDC? Delivery is bought separately from third-party logistics providers on the network — Shadowfax, LoadShare and others — and is usually assigned automatically when the order comes in. You don't need your own riders to sell on ONDC.

Does my restaurant POS already support ONDC and delivery partners? Very likely. Petpooja, Rista by DotPe, Bitsila and Gofrugal all advertise ONDC connectivity, and UrbanPiper enables it through partners uEngage and Bitsila. Most also list third-party delivery integrations. Call your account manager and ask which fleets are live and contracted in your city today — partner lists on vendor websites go stale.

What is a delivery integrator or aggregator? A company that sits between your kitchen and multiple delivery fleets. You send it the order; it picks the rider, prices the trip, and retries when a fleet fails — so you never contract individual fleets or do dispatch triage at 8pm. Examples include magicpin's Velocity, Pidge, Shiprocket Quick and uEngage Flash.

Reference links

Vendor pages cited above, for you to verify current terms directly:

• Petpooja ONDC — https://www.petpooja.com/ondc
• Petpooja online order management (delivery integrations) — https://www.petpooja.com/poss/online-order-management-software
• Petpooja integrations directory — https://www.petpooja.com/poss/restaurant-integrations
• UrbanPiper on going live on ONDC via uEngage and Bitsila — https://www.urbanpiper.com/blog/urbanpiper-enables-merchants-to-go-live-on-ondc-through-partners-uengage-and-bitsila
• Rista by DotPe POS — https://ristaapps.com/restaurant-point-of-sale-software
• uEngage POS, payments and delivery partner list — https://www.uengage.io/partners
• Bitsila (POS + ONDC seller app) — https://www.bitsila.com/
• Gofrugal POS — https://www.gofrugal.com/pos-software/
• Gofrugal integrations — https://www.gofrugal.com/integration-pos.html
• Gofrugal company timeline (ONDC claim) — https://www.gofrugal.com/about.html
• Shiprocket Quick for restaurants and cloud kitchens — https://www.shiprocket.in/quick/food-delivery-services/