The honest default for selling things online is Shopify. It handles payments, tax, fraud, inventory, shipping rates, and a checkout that converts better than one you'd build — and it does all of that on day one for less than a week of engineering time.
So the useful question isn't "Shopify or custom." It's: what is different enough about your business that a platform's assumptions cost you money?
Where Shopify wins outright
If your model is recognisably retail — products with variants, a cart, a checkout, some discounting — the platform is very hard to beat.
You get PCI compliance you didn't have to earn, a checkout that's been optimised across millions of transactions, an app ecosystem for reviews and email and loyalty, and someone else on call when a payment provider changes their API. Every one of those is a thing you would otherwise build, maintain, and be paged about.
Most stores that go custom "for flexibility" end up rebuilding a worse version of the checkout they left behind. It's the single most expensive component in e-commerce and the least visible to the person paying for it.
Where the platform starts to cost you
The seams show when your product isn't really a product.
- Configured or quoted items. Anything where price depends on a calculation — dimensions, materials, a survey, a lead time — sits awkwardly on a catalogue model. You can force it with line-item properties and draft orders, but past a certain complexity you're maintaining an app that fights the platform.
- Fulfilment that isn't shipping. Slots, sites, capacity, appointment windows. A restaurant group taking collection orders across six kitchens isn't picking a shipping rate — it's booking a capacity slot. That's a different data model.
- B2B account pricing. Customer-specific price lists, credit terms, approval chains, purchase orders. Shopify has grown real B2B features and they're worth trying first — but deep account hierarchies are still where teams reach for something else.
- Subscription logic with real rules. Simple recurring billing is a solved problem with an app. Pausing, prorating, swapping mid-cycle, and usage-based tiers are not.
- The catalogue is the product. If the differentiator is search, comparison, configuration, or a recommendation engine, you're building an application that happens to take payment. The storefront is the small part.
The option most people skip
You don't have to choose the whole stack at once. Headless Shopify keeps the checkout, payments, tax, and order management — the parts that are genuinely hard and genuinely regulated — and puts a custom front end in front of them.
You get to build the experience that's actually differentiated, and you keep the machinery you'd have been foolish to rebuild. It costs more than a theme and much less than a full custom build, and it's the right answer more often than either extreme.
The trade is complexity: two systems, a build pipeline, and a team that has to understand both. It's not a good fit for a store with no technical support behind it.
A decision you can make in five minutes
Work down this list and stop at the first yes:
- Is your model recognisably retail, and is the site's job to sell it well? → Shopify with a good theme. Spend the saved budget on photography and email.
- Is the model retail, but the storefront experience is a real competitive differentiator? → Headless Shopify.
- Is pricing, fulfilment, or the account model fundamentally not a catalogue? → Custom for that part, and keep Shopify or Stripe for payments if you can.
- Are you selling access to software rather than goods? → This isn't an e-commerce question at all. It's billing, and it belongs in your product.
The cost nobody quotes
Whatever you pick, the number that decides whether this works is not the build. It's the cost of changing your mind in eighteen months.
Platforms make small changes cheap and structural changes expensive. Custom builds do the opposite. So the question underneath the question is which kind of change you expect to be making — and if the honest answer is "I don't know yet," that's a strong argument for the platform, because it's the one you can leave.
