The Build Is the Cheap Part of an Online Store

Most people planning an online store budget for one thing: the build. They get three quotes for the website, pick one, and treat the number as the cost of going online. It isn't. It is the cost of the door. Everything that happens after the door opens is where the money actually goes, and almost all of it is monthly, not one-time.

This piece walks through the recurring costs in the order they hit you: what a payment costs, what a delivery costs, what a failed delivery costs, what the tax paperwork costs in time, and why the product catalogue is usually the largest single piece of work in the whole project. None of this is a reason not to sell online. It is a reason to know your numbers before you commit.

What a payment actually costs you

Every rupee a customer pays arrives slightly smaller. The payment gateway — the service that takes the card or UPI payment and settles it to your bank — charges a cut on each transaction. Indian gateways commonly advertise around 2% for domestic cards and net banking, and then GST at 18% is charged on that fee. So a 2% fee is really closer to 2.36% of the transaction value once tax is added. International cards are typically charged at a higher rate again.

The important exception: UPI currently carries zero merchant discount rate by government mandate. If your customers pay by UPI, that transaction costs you nothing in gateway fees. This is not a small detail. If half your orders come through UPI, your blended payment cost is roughly half what the rate card suggests. It is worth making UPI the first and most obvious option at checkout, not the third tab down.

Two further things nobody mentions in the sales call. First, settlement is not instant — gateways settle on a cycle, often T+2 working days, so festival weekends and bank holidays stretch the gap between selling and having the cash. Second, a refund usually does not return the gateway fee to you. Refund an order and you are out the fee on a sale you did not make.

Shipping, returns and the RTO problem

Courier charges in India are billed on the higher of actual weight and volumetric weight — length × breadth × height in centimetres, divided by 5000 for most domestic couriers. A light but bulky item, say a lampshade or a cushion, can be billed at three times its real weight. If you sell anything bulky, work out the volumetric weight of your actual packed box before you set a shipping price. Sellers routinely discover this on the first invoice rather than in the spreadsheet.

Then there is cash on delivery. COD is still widely used in India, particularly outside metro pin codes, and refusing it will cost you orders. It also carries three costs stacked on top of each other: a COD handling fee per shipment, a delay in getting your money (the courier remits collected cash on its own cycle), and RTO.

RTO — return to origin — is the killer. The customer doesn't answer the phone, or changes their mind, and the parcel comes back. You pay forward shipping, you pay return shipping, you get no sale, and you receive a box that may need repacking before it can be sold again. On a ₹600 order with ₹70 each way, an RTO turns a modest profit into a clean loss. The single most useful thing you can do about it is not a courier tier, it is a phone call or WhatsApp confirmation on every COD order before it ships. That is a workflow, and it can be largely automated — see business automation — but somebody still has to own it.

Prepaid returns cost less but are not free. If you promise free returns, you are paying reverse pickup plus the gateway fee on the refunded amount, plus your time inspecting the item.

GST, invoices and the paperwork

Selling goods from your own website is, for tax purposes, just selling goods. The usual registration thresholds apply, and once registered you must issue a proper tax invoice for every order with the correct HSN code and the correct rate for that product. Your store software has to generate that automatically and store it, because reconciling hundreds of small orders by hand at the end of the month is not a job anyone does well twice.

Three things that catch people out:

• Place of supply. A customer in Delhi buying from a Delhi-registered seller attracts CGST and SGST. A customer in Lucknow attracts IGST. Your checkout needs the delivery state captured cleanly, and your invoice logic needs to split correctly. This is a settings problem, and it is worth testing with real orders to both cases before launch.
• Marketplace TCS. If you also sell through Amazon, Flipkart or similar, the operator collects tax at source on your behalf and it shows up in your returns. The rate for e-commerce operators was reduced with effect from 10 July 2024. Your accountant handles it; you just need to know it exists and that marketplace sales and own-website sales are reported differently.
• Rates move. The GST slabs were rationalised in September 2025. If your catalogue stores a rate per product rather than pulling from a maintained tax class, a rate change becomes a day of manual editing across every SKU.

The cost here is not a fee. It is hours — yours or your accountant's — every single month, forever.

The catalogue is the real project

Ask an agency what an online store costs and you get a build number. Ask who is writing 400 product descriptions and the room goes quiet.

A single product listing that actually sells needs: a clear title, three to six usable photographs on a consistent background, a description written for a buyer rather than a spec sheet, variants (size, colour) with separate stock counts, packed weight and dimensions for shipping, an HSN code, a tax class, and a URL that won't need changing later. That is genuinely twenty to forty minutes per product if you are being honest, more for the first fifty while you settle the format.

Three hundred products is therefore weeks of work, not an afternoon of "importing the Excel sheet". And it does not stop at launch: new stock, discontinued lines, price changes, seasonal photography. If nobody owns the catalogue as an ongoing job, the store quietly rots — out-of-stock items still listed, last year's prices, photographs that don't match what ships.

Budget for catalogue work as a recurring line, the way you budget for a shop assistant. It is closer to that than to a software cost.

The monthly lines nobody quotes for

Line	How it is charged	Who it goes to
Hosting and domain	Monthly or annual, scales with traffic	Host
Payment gateway	Per transaction, plus 18% GST on the fee	Gateway
Shipping	Per shipment, by billed weight and zone	Courier or aggregator
COD handling and RTO	Per shipment; RTO is a pure loss	Courier
Packaging	Per order, and it is not trivial	Supplier
Order handling time	Staff hours per day	You
Catalogue upkeep	Ongoing hours	You or an agency
Software subscriptions	Monthly, often per seat	Various
Security patching and backups	Monthly retainer or a crisis later	Your support arrangement
Acquisition	Ads, or the slower work of search visibility	Meta, Google, or patience

Add those up per order at your expected volume. If the number is uncomfortably close to your gross margin, the answer is usually a higher average order value — bundles, minimum free-shipping thresholds — not a cheaper courier.

What I would not buy

A store built from scratch. Unless you have a genuinely unusual model, a custom-coded cart is money spent rebuilding things that established platforms already do correctly — tax logic, refunds, PCI-compliant payment handling, stock deduction under concurrent orders. We build on established platforms for exactly this reason. If someone is quoting a ground-up cart for a normal retail catalogue, ask what it does that an off-the-shelf platform does not.

An app, at launch. A mobile app adds store listings, review cycles, two more codebases and a download barrier for customers who were perfectly happy on a website. Revisit it when you have repeat customers asking for it.

A multi-vendor marketplace build before you have a single vendor who has agreed to join. This gets sold a lot. It is three times the build and needs payouts, commission logic and dispute handling on day one.

Bulk catalogue import as a headline feature. Importing 2,000 rows of supplier data produces 2,000 thin, duplicated, unsellable pages. Fewer products, properly done, outsell a large bad catalogue.

An ERP-grade inventory system for one warehouse. A well-configured store plus a simple CRM for customer follow-up covers most businesses until they are genuinely running multiple locations.

What to do next

Before you approve any build quote, do one hour of arithmetic. Take your three best-selling products. For each, write down the selling price, the cost of goods, the packed volumetric weight, a realistic courier charge to a zone you actually ship to, the packaging cost, and 2.36% for a card payment. Subtract. Then assume one in six COD orders comes back and see what that does to the average.

If the number survives, you have a business that will work online and the build becomes the easy part. If it doesn't, you have saved yourself a year. Either outcome is worth the hour.

When you are ready to talk mechanics — platform choice, checkout flow, where to automate the boring parts — have a look at how we approach web development, or tell us what you sell and we will tell you honestly whether an online store is the right next spend.

Questions we get asked