How to build an e-commerce site: decisions, costs and sequence

Building an e-commerce site can technically be a weekend job. You open a Shopify account, pick a theme, add ten products and go live. That part isn't hard. What's hard is that the system still holds up six months later when orders pick up, and that you don't have to rebuild from scratch a year in when you want to grow.

This piece covers the decisions you make when starting from zero, with their real costs. Whether you build it yourself or hire someone, the things you need to know are the same.

Platform first: when Shopify is right, and when it isn't

Let's be honest: Shopify isn't the right answer for every business. Where it is right is clear — if you sell physical products, don't want to run payment and shipping infrastructure yourself, have no technical team or a small one, and want to spend your growth on selling rather than on plumbing, Shopify is probably the fastest route.

There are places where it isn't. If your sales model is genuinely unusual (complex configurators, subscriptions tangled up with one-off sales, heavy B2B tender processes), if you carry hundreds of thousands of products, or if software is already your core business, the platform's rules may constrain you. In that case you either use Shopify purely as infrastructure and write the front end separately (headless), or choose a different road.

Platform choice is made once and is expensive to undo. Choose on where you expect to be in two years, not on the monthly fee.

The decisions that are expensive to reverse

Some choices made in five minutes at the start cost weeks to correct later. Getting these right up front is the highest-return work in the whole project.

• URL structure: changing category and product addresses later means carrying your SEO value across with a redirect map. Set it up cleanly and permanently from the start.
• Product data model: variant or separate product? How do size and colour work? This structure directly drives stock, pricing and reporting.
• SKU discipline: every variant needs a unique, consistent code. If you will ever connect an ERP or a warehouse system, this is non-negotiable.
• Tax and invoicing setup: VAT rates, exemptions, country-specific rules. A misconfigured tax setting comes back as a financial problem months later.
• Payments and shipping: which methods, which countries, which thresholds? Easy to add later, but planning it up front shapes how you can run campaigns.

Where the money actually goes

The cost breakdown is where newcomers are most often wrong. The theme price or the monthly platform fee is usually a small share of the total. The real cost sits in three places: preparing the product data (photography, descriptions, specifications, translation), integrations (accounting, shipping, marketplaces) and everything after launch.

Preparing a thousand-product catalogue properly takes longer than building the site itself in most projects. If you don't plan for it, the project finishes technically but cannot launch — because there is nothing to put in it.

A realistic timeline

A standard Shopify store, with scope kept tight, goes live in four to eight weeks. That includes design, development, product migration and testing. Add an ERP integration and plan another three to six weeks; add a mobile app and plan another six to ten.

The only healthy way to shorten these is to cut scope, not to add people. Deciding what will not be in the first release matters as much as deciding what will.

Should you build it yourself or hire an agency?

The honest answer: if you have few products, a standard business model and the time to learn, build it yourself. Shopify's off-the-shelf themes are genuinely good now, and what you learn building your first store pays off for years. Paying an agency a large sum to end up with something that looks like a stock theme helps nobody.

An agency makes sense when: you have an existing operation and data has to be migrated; you need to connect an ERP or warehouse system; you have a non-standard sales model such as B2B; performance has to be a measurable target; or simply when you don't have the time and your hour costs more than theirs.

A good agency will also tell you what you don't need to build. A proposal that says yes to everything in the first call is a sign the scope wasn't understood.

Launch is a beginning

Going live is not the end of the project, it's the start of measuring. In the first month, watch which page people leave from, where they abandon the cart, and which searches return nothing. The small corrections you make from that data usually produce more sales than the big pre-launch design debates ever did.