Custom E-Commerce vs Shopify: When Building From Scratch Is the Right Call
Short answer: use Shopify (or WooCommerce, or any platform) — unless the thing you sell only makes sense through an experience the platform can't model. Then a custom build stops being an indulgence and becomes the cheaper option. Most brands should not build from scratch. The ones who should, usually know why within five minutes of trying to fit their model into a cart.
What platforms are genuinely great at
Catalogue, checkout, tax, shipping, payments, fraud, PCI, refunds, inventory. That is years of unglamorous, high-risk work you get for a monthly fee. If your business is “products in a catalogue, added to a basket, shipped to an address,” building that yourself is burning money to arrive at a worse version of Shopify.
We tell most clients exactly that. It is the honest answer far more often than the alternative.
The four signals you should build custom
- The product is configured, not picked. The customer composes it — with rules, minimums and constraints the cart has to understand and enforce.
- Price is decided by a human. Commissions, quotes, bespoke work: someone reviews the request and proposes a value before money moves.
- Access is the product. Invite-only tiers, private membership, content and channels that only exist for certain customers.
- The interface is part of the brand promise. If the buying experience has to feel like the object itself, a theme with your logo on it will always undercut you.
One signal? Bend the platform. Two? Have the conversation. Three or four? You are building something.
Platform vs custom, honestly
| Platform | Custom build | |
|---|---|---|
| Time to first sale | Days–weeks | Months |
| Upfront cost | Low | High |
| Ongoing | Fees + apps, forever | Hosting + your own maintenance |
| Payments/tax/fraud | Solved for you | Yours to integrate and own |
| Unusual buying models | Fought with apps and workarounds | Modelled properly in the schema |
| Ceiling | The platform's | Yours |
Note the row that matters: ongoing. Platforms feel cheap until the workarounds arrive — five apps, a headless front end, a developer maintaining glue between them. That is when “cheap” quietly becomes the expensive path.
A case where custom was correct
Delicious Diamonds is a chocolate atelier in Leiria making handmade bonbons in small numbered runs. Four of four signals:
- Configured, not picked. A made-to-measure coffret: pick a format, choose the bonbons, hand-assembled the day it ships — with per-variety minimums enforced live in the basket.
- Human pricing. A bespoke commission where the chef reviews each request personally and proposes a value back through the platform.
- Access as product. An invite-only collectors' circle: a private members' chat, a direct line to the chef, priority on numbered editions, and a bonbon that exists only for members.
- Interface as promise. A seven-step palate profile that maps intensity, texture and the wine at the table into pairings and the right box — plus an AI assistant helping the chef develop and pressure-test new creations.
Frontend and backend written from zero on a schema past 30 tables, bilingual throughout. None of that is a plugin. Full write-up in the case study.
What custom actually costs
Expect months, not weeks, and a real budget. You are buying the parts a platform gave away: accounts, baskets, orders, payments integration, admin, email, permissions. Budget for the second year too — you now own maintenance that Shopify was quietly doing for you.
The trade is control and ceiling. You stop designing around someone else's data model.
What we'd never build custom
Payments and card handling — integrate a provider, never touch card data. Nor tax engines, nor shipping rate calculation, nor auth if a good provider fits. Custom should mean your business logic, not re-implementing solved infrastructure.
FAQ
Is custom e-commerce more expensive than Shopify?
Upfront, always. Over several years it depends entirely on how many apps, workarounds and developer hours the platform version needs to fake your model.
Can't headless Shopify do all this?
Often, yes — headless is a genuinely good middle path when the catalogue and checkout are conventional but the front end must be bespoke. It struggles when the back of the business (commissions, memberships, human pricing) is what doesn't fit.
How long does a custom platform take?
For something of this scope, months. A narrower build with one unusual mechanic can ship far faster.
How do I know which side I'm on?
Count the signals above. If you're writing “we'll just work around it” more than twice while scoping, you already have your answer.