Custom E-commerce vs Shopify in 2026: When to Build, When to Buy, When to Go Hybrid

Shopify ships more SKUs than every custom build in the world combined, and for a reason: for 80% of e-commerce businesses, it's the right answer. The problem is that the other 20% — the ones with custom configurators, B2B account hierarchies, ERP-driven pricing, or product data that doesn't fit a row in a SQL table — keep getting talked into Shopify too. Then they spend the next two years fighting it. Here's how to tell which side of the line you're on.

When Shopify is right

You should be on Shopify if your operation looks roughly like this:

• Catalog under ~10,000 SKUs, with variants that fit the standard size/color/material model.
• Standard checkout — billing address, shipping address, credit card, maybe Klarna. No multi-step approvals, no quote-to-order flow, no purchase orders.
• Pricing is a number on a product, possibly with volume discounts or a coupon code. Not a calculated function of customer tier × region × contract term.
• Inventory lives in Shopify or a single integrated system. Not split across 14 warehouse management systems with different rules per location.
• Marketing-led revenue — performance ads, SEO, influencer, email. Not a long enterprise sales cycle with named accounts.

If that's you, Shopify Plus (€2k/month base, more with apps) costs less than three months of a custom build's maintenance budget. Use it. Use the apps. Use the themes. Save the engineering budget for the parts of your business that actually need custom work.

When custom wins

The crossover happens when at least two of these are true:

• Your product configurator is the product. Engagement rings with material + stone + setting + engraving + sizing combinations. Custom furniture with 200 fabric options and dimension sliders. Industrial machinery quoted from a parts matrix. Shopify can sort of handle this through metafields and custom apps, but you'll spend more on workarounds than on a clean custom build.
• B2B with account hierarchies. Parent companies, child accounts, branch-level purchasing limits, approval chains, customer-specific catalogs, customer-specific pricing. Shopify B2B exists, but it ceilings hard once your account model gets more nuanced than "customer with assigned price list."
• ERP-driven pricing and inventory. Real-time price calculation from your ERP, live stock from your warehouse system, customer credit checks before checkout proceeds. Shopify wants to be the source of truth. If your ERP is the source of truth, you're constantly fighting Shopify's data model.
• A bespoke buying flow. Multi-step configurators with 3D previews. Quote-request workflows that turn into orders after a sales call. Subscription products with complex change rules. Anything that doesn't fit "land on PDP, click add to cart, check out."
• Compliance or audit requirements Shopify can't meet — specific data residency, custom audit logs, regulatory reporting, restricted payment flows.

These are the patterns where Shopify costs you more to bend than custom costs to build.

What hybrid (Hydrogen / Medusa) looks like

The interesting middle has gotten more interesting. You don't have to pick all-Shopify or all-custom anymore.

Shopify Hydrogen keeps Shopify as the commerce backend — products, inventory, checkout, payments, fulfillment — and gives you a fully custom Next.js / React frontend. You write your own storefront in your own framework, query Shopify via Storefront API, and ship a marketing site that feels nothing like a Shopify theme. The checkout still happens on Shopify's hosted infrastructure, so PCI scope stays low and you get Shopify Pay, Shop Pay, and the rest of the wallet integrations for free.

This is the right answer for: brands that need a marketing-grade storefront (3D, video, interactive product stories, sub-second navigation) but whose actual buying flow is standard Shopify checkout. You buy 80% of the platform and rebuild only the customer-facing layer.

Medusa flips it. Medusa is open-source, self-hosted commerce — a Node.js / PostgreSQL backend you own end to end, plus a customizable admin and storefront. Use it when you need the data model flexibility of a custom build but don't want to write the cart, order, payment, and tax foundations from scratch.

This is the right answer for: operations that need full data ownership, deeply custom workflows, or commerce features Shopify can't model — but whose core flow is still recognizably commerce. You skip writing the boring 70% (cart, checkout, orders, customers, payments) and spend your engineering on the parts that differentiate.

The decision in one paragraph

If your business runs on a standard commerce model and your differentiation is brand, marketing, or product selection — use Shopify. If your differentiation is a buying flow Shopify can't model, your data lives in systems Shopify can't fully integrate with, or your scale puts you on Plus + 12 enterprise apps with a six-figure annual bill — look at custom or hybrid. Hydrogen for "Shopify backend, custom frontend." Medusa for "own the whole thing, don't reinvent commerce primitives." Full custom for the rare cases where neither fits.

What custom actually costs

For honest numbers — and assuming you're hiring a real engineering team, not a template shop:

• Hydrogen storefront on top of existing Shopify — €30k–€80k for the build, depending on configurator complexity, animation, and integrations. 6–14 weeks.
• Medusa-based custom commerce with admin, storefront, payments, and a couple of integrations — €80k–€200k. 4–8 months.
• Fully custom B2B commerce platform with bespoke account hierarchies, ERP-driven pricing, quote workflows, custom payment flows — €200k+. 6–18 months.

These are build numbers. Add a maintenance budget of 15–25% of the build cost annually for the first three years. Anyone quoting custom commerce without naming the maintenance number is selling you the iceberg's tip.

A test: how much custom code does your current Shopify have?

Open your Shopify admin. Count the apps. Look at your theme repo (if you have one). Count the custom Liquid sections, the metafield definitions, the webhooks firing into your own infrastructure, the scripts running on checkout.

If that count is under 10, Shopify is still working for you.

If it's 30+, your custom code is the actual product and Shopify is the runtime you're paying €X to host it. At some point past that, you're better off owning the runtime.