Most of the money wasted on e-commerce web development is spent before anyone writes code. A brand signs off a 40-page design file, discovers in week six that the product data can't support the filters in the mockup, and pays twice: once for the design, once for the rework. We see this pattern more than any other. So this guide runs in the order the work actually happens, not the order it appears on a proposal.
If you read nothing else: pick the platform in week one based on your catalogue and your checkout needs, spend the next three weeks on product data structure, and treat design as the last decision, not the first.
The brief is the expensive part
A good brief for a growing brand is four or five pages and answers things nobody enjoys answering. How many SKUs today, how many in eighteen months. How many variant axes, and does any product have more than 100 variants. Do you sell to GCC and India from one store or two. Is there a subscription. Is there a made-to-order lead time that changes by product. Do you need HSN codes on the invoice. Does anything you sell need an FSSAI licence number on the PDP.
Those answers determine the build. Colour palette does not. We have rebuilt stores where the entire architecture was decided by one line in a brief that said "we'll add B2B later", because B2B later means a second price list, and a second price list means either Shopify's B2B features on a higher plan or a custom app, and that changes the quote by lakhs.
Platform first: where e-commerce web development actually diverges
For a D2C brand doing between ₹50 lakh and ₹30 crore a year, Shopify is the default and the burden of proof is on anything else. That is not vendor loyalty. It is because the three things that break custom builds at scale — payment reliability, checkout conversion, and peak-traffic stability during a Diwali sale — are the three things you would be rebuilding from scratch and maintaining forever.
Where another platform genuinely wins:
- WooCommerce if your business is primarily content and the store is secondary, and you already have WordPress people. Cheaper to start. You own the hosting problem.
- A custom stack if the commerce logic is the product. Rental, marketplaces with seller payouts, complex quoting, regulated goods with per-state rules. If you can describe your checkout in two sentences, you do not need custom.
- Headless Shopify if you have an in-house front-end team that will still be there in two years. Not otherwise. Headless shifts cost from licence fees to salaries, and the brands that regret it are the ones who lost their React developer and then couldn't change a banner for six weeks.
Shopify Plus is a separate question from Shopify. Most brands ask about it a year too early. The honest test is whether you will use checkout extensibility, Shopify Functions for custom discount logic, or multiple expansion stores. If your answer is "we want to look serious", stay on the standard plan and spend the difference on inventory. We say this often enough that it has become a running joke internally, and it has never once cost us a client who was actually ready for Plus.
What e-commerce web development costs in India, with the arithmetic
Quotes vary wildly because people compare different things. Here is a three-year comparison with the assumptions written down, so you can swap in your own numbers.
Custom build: ₹20,00,000 initial development. One maintenance developer at ₹1,20,000 a month for 36 months is ₹43,20,000. Hosting, CDN, monitoring and a payment-gateway integration you maintain yourself at ₹15,000 a month is ₹5,40,000. Total: ₹68,60,000.
Shopify build: ₹6,00,000 initial. Platform plan and apps at a combined ₹20,000 a month is ₹7,20,000. A part-time retainer at ₹40,000 a month for continued CRO and feature work is ₹14,40,000. Total: ₹27,60,000.
The gap is about ₹41 lakh, and the custom build still doesn't have a better checkout. Your numbers will differ. The point is to run the three-year version, because comparing a ₹20L custom quote with a ₹6L Shopify quote and stopping there is how brands end up with a stack they can't afford to maintain. We've broken down the line items in more detail on our Shopify development cost page.
One caveat on the cheap end. A ₹60,000 theme-install job is a real option and sometimes the correct one. If you have 30 SKUs, one warehouse, and no custom logic, buy a good theme, configure it properly, and spend your budget on photography and ads. Come back for a custom build when the catalogue or the operations demand it.
Your catalogue data is the product
The single highest-leverage week in any build is the one spent on the product data model, and it is the week clients most want to skip.
Decide what is a product and what is a variant. Decide which attributes are metafields and which are tags. Write the taxonomy down in a spreadsheet before design starts. If you sell apparel, you need size and colour as variants, fabric and fit and occasion as metafields, and probably a care-instruction block that is structured rather than pasted into the description. If you sell jewellery, weight, purity, metal type and making charge are structured fields, because they drive both filtering and pricing.
Get this wrong and two things happen. Filters can't be built without a bulk re-tag of the whole catalogue. And your product feed to Google Shopping and Meta is garbage, which quietly caps your paid performance for as long as you leave it.
A rough rule: if a customer might want to narrow a collection by it, or a marketplace feed needs it, it is a structured field. Not free text.
Checkout is where Indian builds leak money
Cash on delivery is still a meaningful share of orders for most Indian D2C brands, and it is the part of the build that deserves actual engineering attention rather than a checkbox.
Things worth doing at build time, not later: OTP verification on COD orders above a threshold, a small prepaid discount shown at the point the customer selects payment method, pincode-level COD availability tied to your courier's serviceability list, and an address-quality check that catches a missing landmark or a six-character pincode before the order is placed.
Do the RTO arithmetic for your own store. Take 1,000 COD orders a month at an average order value of ₹1,500. At 24% RTO you lose 240 orders of revenue and still pay forward and reverse shipping on them, say ₹160 a shipment both ways, so ₹76,800 in freight plus ₹3,60,000 of revenue that never lands. Move 15% of those COD orders to prepaid with a ₹75 discount and you've spent ₹11,250 in discounts to remove roughly 36 returns. That is worth doing. Run it with your own RTO rate before you copy the numbers.
On gateways: Razorpay, Cashfree and PayU all work on Shopify. The deciding factors are settlement cycle, UPI success rate on your customer base, and whether you need international cards. Test two in parallel for a month if the volume justifies it, and watch the success rate per method rather than the blended number.
Set a speed budget before the design is signed off
Write the performance target into the contract. We use: Largest Contentful Paint under 2.0s on a mid-range Android over 4G, Interaction to Next Paint under 200ms, total JavaScript on the homepage under 400KB compressed. Those are achievable on Shopify with a well-built theme. They are not achievable if the homepage carries a hero video, three carousels, a live chat widget, a reviews widget, a wishlist app, an upsell app and two pixels.
The damage is usually cumulative rather than dramatic. Each app adds 40–90KB of script and nobody notices until the store is at 3.8s LCP and the agency is being blamed for the theme. Audit the app list quarterly and delete anything that isn't earning its weight. Our own SwiftStore app handles the scanning and monitoring part of that loop, and the deeper theme-level work is covered on our speed optimisation page.
What moves the needle most, in our experience, in order: compress and correctly size hero images, lazy-load everything below the fold, remove unused app scripts, defer third-party tags through a tag manager, and only then start on theme code. Most agencies do that list backwards because theme code is billable.
Search and filtering break quietly as the catalogue grows
Under 200 SKUs, default collection navigation is fine. Between 200 and 1,000 it starts to cost you. Past that it is one of the biggest single conversion levers in the store, and nobody notices because the failure is silent: a shopper searches "cotton kurta xl", gets eleven irrelevant results, and leaves. No error, no complaint, no line in your analytics unless you've specifically instrumented zero-result and low-result searches.
Do instrument them. Export the top 200 internal search queries monthly. The ones returning nothing are either products you should stock, synonyms you should add, or a naming mismatch between what you call a thing and what customers call it. That report costs an hour and routinely finds revenue.
Migration: plan the redirects, not the redesign
Replatforming posts fail for boring reasons. The 301 map is incomplete, so 4,000 ranking URLs 404 for three weeks. Customer passwords don't transfer, so the login flow has to force a reset and nobody warned the support team. Historical order data stays on the old system and the finance team discovers it in April, mid-GST-filing.
Sequence it: export and clean the catalogue, build the URL map including old collection and blog paths, migrate customers and orders, run the new store on a password-protected domain for a full week of internal ordering, then switch DNS on a Tuesday morning. Never on a Friday, never in October. We've written up the full sequence under Shopify migration, including the bits that only bite you a month later.
Budget two weeks of search-ranking wobble even when the redirects are perfect. If someone promises none, they haven't done many.
Who should build it
In-house makes sense when you have continuous change: weekly landing pages, frequent merchandising experiments, a roadmap that doesn't end. A single mid-level Shopify developer in Bengaluru costs ₹8–14 lakh a year fully loaded and is worth it from roughly ₹10 crore in revenue, earlier if you're running a lot of campaigns.
An agency makes sense for the build itself, for specialist work you need twice a year, and for the months either side of a festive peak when your own team is already stretched. The hybrid most of our clients land on: one in-house person who owns the day-to-day, with an external team for the heavy builds and the performance work. If you're weighing that up, our notes on hiring Shopify developers in India cover the rates and the interview questions that actually sort people.
Start here
Before you brief anyone, open a spreadsheet and fill in four columns: SKU count today and in eighteen months, variant axes per product, payment methods you need live at launch, and the five things your current store can't do. That document will cut a month off your build and will tell you, honestly, whether you need a ₹6 lakh project or a ₹60,000 one.
If you want a second pair of eyes on the existing store before you commit, our free audit covers speed, checkout flow and the catalogue structure, and we'll tell you if the answer is to fix what you have rather than rebuild it.