What a marketplace app costs in India in 2026
A working two-sided marketplace — buyer app, seller panel, admin console and live payments — typically costs Rs. 8,00,000 to Rs. 20,00,000 in India in 2026. A web-first pilot with manual operations can start near Rs. 3,50,000, and a mature platform with automated settlement, dedicated search and in-app chat runs Rs. 45,00,000 and upwards.
The range is that wide because a marketplace is not a feature list, it is a business model. Two apps that look identical to a buyer can differ several times over in cost depending on how money moves, who verifies sellers, and how much of the operation a human still does by hand.
A marketplace is three products, not one
This is the single thing most first-time founders price wrong. You are not commissioning an app. You are commissioning a buyer experience, a seller experience and an operations console, all sitting on one shared backend.
Every meaningful feature exists three times. An order is a screen the buyer tracks, a screen the seller acts on, and a record the admin can inspect, refund and override. Each version carries its own states, its own notifications and its own failure cases.
What that does to the estimate
A single-audience app with twenty screens is twenty screens. A marketplace with twenty buyer screens usually implies ten to twelve seller screens and fifteen or more admin screens, plus the APIs underneath all three. Cost rises with the number of interfaces, not the number of ideas, and every flow has to be tested twice over, from both sides.
Where to cut without damaging the product
Make the seller side a responsive web panel rather than a native app. Sellers work sitting down, with a keyboard, often uploading in bulk, and a browser serves them better in the first year anyway. Keep the admin console deliberately plain: tables, filters, buttons, no design system.
A native seller app is commonly the first Rs. 4,00,000 to Rs. 8,00,000 you can defer. If you want the like-for-like comparison against a normal single-audience build, our guide to mobile app development cost in India sets the baseline, and our mobile app development page explains what each build stage involves.
Version one should assume humans, not automation
A marketplace has a cold start problem: buyers do not arrive without sellers, and sellers do not stay without buyers. Supply almost always has to be solved first, and in the early months supply is solved by a person making phone calls, not by software.
Your first fifty or hundred sellers will be signed up by your team. They will not stumble onto a self-serve onboarding flow, read your policy page and upload their own documents unprompted. So do not pay to build that flow yet.
Do these by hand in version one
- Seller onboarding: an admin form your team fills in, not a public wizard with document scanning and automated approval.
- Payouts: a downloadable settlement report, with the actual bank transfer done by your accounts person.
- Catalogue creation: a spreadsheet import in admin, plus someone cleaning the data before it goes live.
- Disputes: a status field and a conversation, not an automated resolution engine.
- Seller analytics: a weekly emailed report, not a dashboard with charts.
What you must build correctly from day one is the ledger underneath all of that. A manual payout is a workflow choice you can change in an afternoon. A wrong ledger is a rebuild, and you will only discover it during your first serious refund.
One admin capability is worth insisting on: the ability to act on a seller’s behalf. Edit their listing, accept their order, correct their price. In the first year your operations team will do this daily, and without it they will ask a developer instead, which is the most expensive form of customer support there is.
Payments, commission and payouts: where the budget actually burns
Most of the invisible cost in a marketplace sits here. Taking a payment is easy and well-solved. Deciding whose money it is, and for how long, is not.
The money flow you are really building
The buyer pays the full amount into your payment gateway account. That money is not your revenue. Most of it is a liability you owe a seller once the order is fulfilled and the return window closes. Your commission is the only part you have earned.
So the system needs an order lifecycle with money attached to every state: paid, accepted, fulfilled, in return window, settled, refunded, reversed. Every transition writes a ledger entry that has to still make sense to an accountant eight months later. This is ordinary custom software work, but it is the part a template cannot give you.
Escrow-style holding, and why it is not really escrow
Founders often ask for escrow. In practice you are holding funds in your own gateway balance and releasing them on a schedule, which is a different thing legally and operationally. Some gateways offer split-payment or routed-payout products that hold and release each seller’s share for you, but those require every seller to be onboarded and verified with the gateway itself, which adds friction to the supply side you are already struggling to grow. Decide this before architecture, not after.
The refund nobody scopes
A buyer returns an item three days after you have already paid the seller. That money has left your balance. You now need negative seller balances, recovery from the next payout cycle, and a written rule for what happens when a seller never sells again. Skipping this is what turns month four into an emergency and a change request.
Rates, and what to assume
Gateway charges are a percentage of each transaction and differ by provider, by payment method and by your volume. UPI, cards, net banking and wallets are not priced alike, and negotiated rates move as you grow. Never let a proposal build your unit economics on one fixed number. Ask your provider for a written schedule that includes payout, refund and failed-transaction charges.
There is also a tax layer specific to Indian marketplaces. An operator that collects payment on behalf of sellers carries tax deduction and collection-at-source obligations, with monthly filings and per-seller statements. The applicable rates have been revised more than once, so confirm the current position with your chartered accountant before anyone hard-codes a percentage into the payout job.
Identity and trust: what makes strangers transact
A marketplace only works if a buyer believes the stranger on the other side will actually deliver. Trust features are not polish, they are the product.
Seller verification
Expect phone OTP login on both sides, PAN and GST capture where applicable, and bank account verification before the first payout. Document upload with a manual review queue in admin is enough for version one; automated verification services charge per check and can wait until volume justifies them.
Ratings, reviews and disputes
Review systems look trivial and are not. Only a verified completed order should be able to leave a review, sellers must never be able to delete an unfavourable one, and an admin needs a path to remove genuinely abusive content with a reason recorded against it.
Budget a dispute flow too: a buyer raises an issue, the order freezes, the payout holds, and an admin decides where both sides can see it. Holding the payout is the part cheap builds forget.
Leakage
Buyer and seller will try to move the deal off your platform to avoid commission. This is the quiet killer of marketplace revenue, particularly in services. Masked calling or in-app chat with numbers hidden is the usual defence, and it costs real money: a telephony provider, per-minute charges and a chat service. Decide early whether you can afford to police it, or whether your commission is modest enough that nobody bothers.
Search, discovery and notifications
Search costs what your catalogue demands
With a few thousand listings, well-indexed database queries and good category filters are enough and cost almost nothing to run. Typo tolerance, synonyms, relevance tuning and instant suggestions need a dedicated search engine, which is build effort plus a recurring monthly bill. Buy it when your data says to, not at launch.
Log every search that returns nothing. That log is the cheapest roadmap you will ever own, because it tells you exactly which supply to go and sign up next. Design the zero-result screen properly too, with a request-this option, since it is a common moment a new buyer leaves for good.
Ranking is a business decision, not a technical one
Someone has to decide what appears first: nearest, cheapest, best rated, newest, or whoever paid. Sponsored placement is often the second revenue line after commission, but it needs a labelled slot, a budget system and a click ledger. Even if you build it in year two, leave the hook in the data model now.
Notifications, and why WhatsApp carries the load
Push notifications are free to send and unreliable in India, because aggressive battery optimisation on many Android phones kills background delivery. Treat push as a bonus channel, never as the one carrying order and payment updates.
SMS still works for OTP, and in India it needs DLT header and template registration before anything sends. Budget the setup time, not just the per-message price.
WhatsApp is where Indian order updates actually get read. Sent through the official WhatsApp Business Platform, confirmations, dispatch updates and payout notices reach both sides in the app they already live in. Message charges are set by Meta and billed by Meta to your own WhatsApp Business Account. ZupiChat, our own WhatsApp CRM, adds the templates, automation and shared team inbox on top without any markup on Meta’s charges: Starter is free forever, Growth is Rs. 999/month for 3 WhatsApp numbers and 3 team members, and Enterprise is Rs. 2,499/month with unlimited agents. There is no separate free trial, because the free plan is the trial, and API access is included on every plan.
Realistic price bands for India in 2026
These are market ranges across Indian studios and agencies, not a quote. Far below the band usually means a template with the marketplace logic missing. Far above usually means a bigger team than your stage needs.
| Version | What is in it | Typical cost in India, 2026 |
|---|---|---|
| Web-first pilot | Responsive marketplace site, admin console, listings and enquiries or a single payment flow, sellers onboarded by your team, payouts by hand | Rs. 3,50,000 to Rs. 8,00,000 |
| MVP marketplace | Buyer app on Android and iOS, responsive seller panel, admin console, live gateway payments, commission ledger, ratings, full order lifecycle, WhatsApp and SMS updates | Rs. 8,00,000 to Rs. 20,00,000 |
| Growth build | Native apps on both sides, routed or automated payouts, seller KYC workflow, dedicated search, in-app chat or masked calling, coupons, sponsored listings, reporting | Rs. 20,00,000 to Rs. 45,00,000 |
| Scale platform | Multi-category or multi-city, logistics and a delivery partner app, wallets and subscriptions, fraud and risk rules, analytics warehouse, a permanent product team | Rs. 45,00,000 and upwards |
Timelines follow the same shape: roughly 8 to 12 weeks for a pilot, 4 to 6 months for an MVP, and 8 to 14 months for a growth build with a team that keeps shipping afterwards. If your model is closer to a single-seller store than a true marketplace, compare against ecommerce website cost in India first, because you may be buying complexity you do not need. Platforms that eventually sell seats or subscriptions to sellers drift into SaaS development territory, which changes both the architecture and the pricing conversation.
What it costs to run every month
The build is the deposit. Running it is the rent, and a marketplace with no operations budget stalls quietly a few months after launch.
| Running cost | How it is charged | Early-stage monthly range |
|---|---|---|
| Cloud hosting and database | Per server, storage and bandwidth | Rs. 4,000 to Rs. 30,000 |
| Payment gateway | Percentage of each transaction plus payout and refund charges; varies by provider, method and volume | Scales with turnover |
| SMS and OTP | Per message, after DLT registration | Rs. 1,000 to Rs. 15,000 |
| WhatsApp messaging | Set by Meta and billed by Meta to your own WhatsApp Business Account | Depends on conversation volume |
| Maps and geocoding | Per request, free allowance then metered | Rs. 0 to Rs. 20,000 |
| Hosted search and media delivery | Per index size or bandwidth | Rs. 0 to Rs. 15,000 |
| Operations and support people | Per person, per month | Rs. 18,000 to Rs. 40,000 each |
| Maintenance and updates | Retainer, or planned at 15 to 20 per cent of build cost a year | Rs. 15,000 to Rs. 60,000 |
Two more line items people forget. The app stores charge an annual developer fee on Apple and a one-time registration fee on Google, both billed in dollars. Store commission applies to digital goods bought inside the app, while physical goods and real-world services paid through an external gateway are generally outside it, but these policies get revised, so read the current developer policy before you design your checkout.
Maintenance is the line founders delete and then pay anyway, in a panic, when an OS update breaks the camera or a gateway retires an API. Our guide to mobile app maintenance cost in India covers what that year-two number contains.
When a web marketplace beats an app
An app is not automatically the right first product, and for a lot of Indian marketplaces the responsive website wins on every measure that matters in year one.
Go web first when
- Purchase frequency is low: property, weddings, machinery, annual services. Nobody installs an app for something they do twice a year.
- You need Google traffic. Listings on a website can rank; screens inside an app cannot. For a discovery-led marketplace that single difference decides where your demand comes from, which is why website development and search visibility usually belong in the same plan.
- Your sellers work at a desk with bulk uploads, invoices and a printer.
- Your budget covers exactly one product properly. One well-built responsive marketplace beats three half-built apps every time.
Go app first when
- Buyers come back weekly, so push re-engagement genuinely drives revenue.
- You need the camera, precise location, background tracking or offline use, as a delivery partner app does.
- Repeat ordering and payment must be two taps, not a fresh login each time.
The middle path is a progressive web app: one codebase, installable from the browser, reachable from a search result, no store review cycle. It cannot do everything a native app can, but for a marketplace still hunting its first hundred sellers it is very often the correct compromise.
What to settle before you ask anyone for a quote
Write down four answers. Who your first hundred sellers are and how you will reach them. What you charge, and at which moment you take it. What a buyer must believe before they pay a stranger. And which parts of the operation a human will do for the first six months.
A developer can price that list accurately. Nobody can price a vague request to build something like a well-known app, and a firm that quotes one without asking is guessing at your expense. Compare estimates on scope and money logic, not on the bottom line.
Codelith Lab is based in Pune and builds marketplace platforms for businesses across India. Send us those four answers through our contact page and we will come back with an itemised scope for version one, what it costs to build, and what it will cost you to run every month afterwards.
