Choosing your e-commerce platform — from Shopify to headless to fully custom

Every founder building a product that sells physical or digital goods will face this question. The default answer — Shopify — is correct more often than not. But the default answer is also the one that agencies, influencers, and comparison sites are incentivized to give you. Nobody makes affiliate revenue recommending “build it from scratch in Rails.”

Every recommendation here comes from real builds, not marketing pages. I rebuilt a 10-year legacy PHP e-commerce platform into Rails over 2.5 years at High Stickers — a project where no off-the-shelf platform could have served the business model. I built a custom e-commerce system for RV dealerships at kwali — an industry where no platform even comes close. And I have integrated Shopify, set up WooCommerce stores, and evaluated every headless option on this list for real clients.

The angle is simple: start with the cheapest, fastest option that does not constrain your product model. Graduate when you hit real limitations — not when a consultant tells you to.

The spectrum runs from “zero infrastructure, maximum constraints” to “full control, full responsibility.” Every step toward more control is a step toward more cost, more complexity, and more maintenance. The question is not “which platform is best?” — it is “which constraints can your business live with?”

Tier	Examples	Best for	Launch time	Cost range
Hosted SaaS	Shopify BCBigCommerce	Founders launching, DTC brands under ~€10M ARR	Days to weeks	€29–2,500+/mo
Self-hosted legacy	WooCommerce PSPrestaShop	WordPress-first merchants, content-led commerce, European markets	Days to weeks	“Free” + €1,500–50K+/yr
Modern headless	MeMedusa SaSaleor VnVendure	Dev teams with TypeScript/Python, genuine multi-region needs	Weeks to months	Free core + €25K–150K+ build
Ruby headless	SoSolidus SpSpree	Rails teams, complex product models, rich data modelling	Weeks to months	Free core + Rails dev cost
Fully custom	Ruby on Rails Django Next.js	Unique product model IS the competitive moat	Months	€100K–500K+ build

Hosted SaaS

Best forFounders launching, DTC brands under ~€10M ARR

LaunchDays to weeks

Cost€29–2,500+/mo

Shopify BCBigCommerce

Self-hosted legacy

Best forWordPress-first merchants, content-led commerce, European markets

LaunchDays to weeks

Cost“Free” + €1,500–50K+/yr

WooCommerce PSPrestaShop

Modern headless

Best forDev teams with TypeScript/Python, genuine multi-region needs

LaunchWeeks to months

CostFree core + €25K–150K+ build

MeMedusa SaSaleor VnVendure

Ruby headless

Best forRails teams, complex product models, rich data modelling

LaunchWeeks to months

CostFree core + Rails dev cost

SoSolidus SpSpree

Fully custom

Best forUnique product model IS the competitive moat

LaunchMonths

Cost€100K–500K+ build

Ruby on Rails Django Next.js

Most businesses should start at Tier 1 (Shopify) and graduate only when they hit specific, measurable limitations. The most expensive mistake is not choosing the wrong platform — it is over-engineering your first choice.

Shopify dominates for good reason. It powers 28.8% of the top 1 million e-commerce sites (BuiltWith), processed $88 billion in GMV in Q2 2025 alone, and its international GMV grew 42% year-over-year. The checkout converts 5% better than Magento on average. Shop Pay reduces checkout time 60%+. The app ecosystem has 8,000+ integrations.

Speed to market

A Shopify store can launch in hours to days for a standard DTC catalog. Basic at $29/month (annual billing) — operational instantly. WooCommerce takes days to weeks for comparable setup. A custom build takes weeks to months. If speed to market is your competitive advantage, Shopify is the answer.

Pricing at different scales

Upgrade thresholds: Basic → Grow at ~$8,000/month revenue. Grow → Advanced at ~$40,000/month. Advanced → Plus at ~$200,000/month when API flexibility and B2B features become necessities.

The sweet spot

• DTC brand with straightforward catalog: Excellent fit
• Under ~$10M ARR, standard product model: No reason to leave
• Need to ship fast — days not months: Nothing comes close
• Migrating from Magento: The most common enterprise migration pattern. Serge Blanco (French menswear, 50+ stores) cut platform costs 75–80% moving from Magento to Plus
• Escaping WooCommerce maintenance: Typical gains are 8–15% checkout conversion improvement plus managed hosting

Shopify’s constraints are architectural, not configurable. No amount of apps, custom code, or Shopify Plus budget can fix these. When you hit them, you have outgrown the platform — not the plan.

Complex product configuration

The variant limit was raised to 2,048 in October 2025 (from 100). But the real problem was never the count — it is the lack of conditional logic between options. Products requiring “if customer selects option A, show pricing tier B, hide options C/D” require a custom configurator layer. Industrial equipment, automotive parts, motor assemblies — these cannot be modelled in Shopify’s variant system. Workarounds exist (virtual SKUs, external configurators pushing custom line items) but they add significant complexity.

Custom pricing logic (CPQ)

Shopify Functions (the Scripts replacement, mandatory by June 30, 2026) can handle tiered discounts and customer-tag pricing. But they cannot do:

• Real-time CPQ based on product configuration
• Contract pricing across line item × customer segment × quantity break × region simultaneously
• Complex matrix pricing (3+ dimensions)
• Dynamic pricing tied to live external data feeds

The B2B pricing is actively broken: discounts apply to the base price, ignoring Market-adjusted regional prices. A B2B customer in Region X getting 10% off pays £90 instead of £99 because Shopify ignores the Market price (£110) and applies the discount to the base (£100). This has been documented since December 2025.

Multi-vendor marketplace

Shopify is architecturally incapable of being a marketplace. No native vendor panel, no commission management, no multi-vendor cart splitting, no automated payouts, no per-vendor analytics. “Shopify allows you to experiment with a marketplace idea, not to build a scalable one.” Minimum viable marketplace on Plus: $30,000+/year before custom development. For real marketplaces, look at Spree (native multi-vendor), Sharetribe, or custom.

B2B beyond basic wholesale

B2B features are Plus-only ($2,300+/month). Even on Plus: only 2 pricing tiers natively, no parent-company/subsidiary hierarchies, no built-in CRM or sales rep tooling, B2B checkout doesn’t support Checkout Extensibility (separate code path). Multi-tier pricing (distributor/dealer/VIP/contract) requires stacking third-party apps at $3,500–6,000/month total.

Deep ERP integration at scale

A single product GraphQL query costs ~100–115 points. On Shopify Plus (1,000 points/sec), querying 10–12 products simultaneously triggers throttling. Real case: WTB (American bike brand) on Plus + SAP Business One — API throttling during Black Friday required an iPaaS solution for retry/queue logic.

The “Shopify tax” at scale

At $10M/month GMV on Plus, you pay $40,000/month to Shopify in platform fees alone — $480,000/year before apps, development, or processing fees. Add the typical Plus app stack ($1,000–3,000/month) and dev retainer ($7,000–17,000/month), and total cost of ownership reaches $10,300–29,300/month. At that spend, a custom build starts to pencil out.

Companies that left Shopify

• Unimig (Australian welding equipment): Rigid architecture for complex pricing, no PIM for technical specs, B2B limitations. Moved to Next.js + Directus + Medusa.js + Odoo.
• XeroShoes (minimalist footwear, WooCommerce → Shopify): Shopify was nearly twice as slow — homepage on 3G mobile: 5.76s → 11.44s. Three of five variation apps failed. Cart bug on Thanksgiving had “five-figure sales impact.” Yet the founder would still make the same decision for managed infrastructure.
• Detroit Ammo Co.: Multi-vendor marketplace for online firearms — Shopify architecture could not support it.
• Anonymous fabric store (Reddit, 2025): Carrier-calculated shipping requires $399/month Advanced plan. Flat-rate tiers based on yards (not grams) — impossible without custom code. Reversed the migration back to WooCommerce.

WooCommerce — the WordPress powerhouse

WooCommerce leads by raw store count: 4.23 million live stores (33.4% of all tracked e-commerce). But it is declining — down 6% year-over-year, losing ~8,599 stores to Shopify per 90 days. The trend tells you where the market is going. But the absolute numbers tell you the ecosystem is massive and healthy.

When WooCommerce wins:

• Content-led commerce: WordPress is the superior SEO platform. Yoast + WooCommerce provides more technical SEO control than any hosted alternative
• Full data ownership: Your server, your database, no platform can change pricing or shut you down. Critical for European GDPR-conscious merchants
• European market: Strongest European coverage by absolute store count — Germany (111K), Italy (102K), Netherlands (100K), France (99K), Spain (96K), UK (163K)
• Cost at small scale: $2,000–5,000/year, significantly cheaper than Shopify + apps at similar GMV
• WPML multilingual + multi-currency: The most mature European multi-language solution

When WooCommerce hurts:

• Security: World’s most targeted CMS. A store with 20–30 plugins has 20–30 potential attack surfaces. A critical Store API vulnerability was patched in December 2025 — stores with auto-updates disabled were exposed.
• Plugin conflicts: The #1 operational headache. Any plugin update can break other plugins.
• Performance at scale: PHP-rendered, database-heavy. Degrades without Redis caching, CDN, and proper hosting. At $50M+ GMV requires major infrastructure investment.
• No central support: Fragmented across hosting provider, plugin developers, and forums.

PSPrestaShop — the declining European legacy

PrestaShop is at 162,832 stores and declining 11% year-over-year. Only 3% of new European stores in 2025 chose PrestaShop, versus 14.3% of existing French stores still running it. The platform was acquired twice in four years (MBE Worldwide in 2021, Cyber_Pixel/Cyber_Folks in February 2026 for ~€55M).

If you have an existing PrestaShop store with significant customization, migration cost may exceed the cost of maintaining it. But for a new store in 2026: the declining trajectory, module conflict problems, and ownership instability make it hard to recommend. The Cyber_Pixel acquisition does create a potential upgrade path to Sylius (headless, same ownership group), but that is a bet on future execution.

The headless commerce market is real: $1.74 billion in 2025, projected $7.16 billion by 2032 (22.4% CAGR). But “headless” has become the “microservices” of e-commerce — a technically sound architecture that gets recommended far more often than it should be.

The back-migration pattern

A significant portion of headless Shopify projects in 2025–2026 involve migrating back to native Shopify. Not because headless is wrong, but because the organisation could not maintain it long-term, content updates slowed down, and the cost/value gap never closed. The original build was theoretical, not driven by real needs.

Total cost of ownership

Enterprise headless averages $2.6 million (platform, dev, integration, migration). Mid-market implementations run $300K–$800K. The hidden “integration tax” arrives 6–24 months post-launch: API orchestration complexity grows non-linearly with vendor count, marketing team autonomy is lost, and DevOps overhead compounds. Legacy-heavy organisations spend nearly 60% of IT budget on upgrades and maintenance — making the “headless is more expensive” argument moot when you are already drowning in tech debt.

The headless platforms

MeMedusa — the fastest-growing open-source option

Copenhagen-based, MIT licensed, 32,300+ GitHub stars growing at 33.4%/month. The key differentiator: zero GMV fees — pricing is infrastructure-only ($29–$299/month on Medusa Cloud). At scale, this saves tens of thousands monthly versus Saleor Cloud (0.8%+ GMV). Fully modular architecture with 17 built-in commerce modules. Used by Eight Sleep ($1B+ valuation, built in 6 months with a single engineer) and Matt Sleeps (24x revenue growth in 3 years).

SaSaleor — enterprise multi-channel

Polish origin, Python/Django, GraphQL-first with the most mature headless GraphQL schema. Zalando is a strategic investor. Multi-channel is the flagship feature: each channel has independent pricing, stock, currency, payment methods, tax, and shipping. Used by Lush Cosmetics (£690M turnover, 869 shops, 50 countries). But: GMV-based Cloud pricing (0.8%+ on Starter) gets expensive at scale, Python requirement means a separate hiring pool from the React frontend, and GraphQL-only means no REST.

VnVendure — TypeScript-native B2B

German company (Vendure GmbH), self-funded, NestJS architecture. The strictest TypeScript typing end-to-end. Enterprise customers include IBM and Swile. Made a bold pivot in February 2026: closing their €3M/year Elevantiq agency business to go all-in on product. No managed cloud yet (announced but not GA) — use Medusa Cloud or Saleor Cloud if you need managed hosting today.

For Rails teams specifically

SoSolidus — forked from Spree in 2015, maintained by Nebulab (Italian agency). Installed as gems in a standard Rails app. All ActiveRecord, ActiveJob, Rails conventions — zero context switching. Best for: clean, highly-customised DTC or B2B Rails stores. Added reverse charge VAT fields in v4.6 for EU compliance. But: not API-first by design, default frontend is dated ERB templates.

SpSpree — 15,000 GitHub stars, 865 contributors, the largest Rails e-commerce community. Spree 5 (April 2025) was a transformation: headless-native API-first architecture, modern JavaScript admin UI, no-code storefront builder, native multi-vendor marketplace, Price Lists engine for B2B. But: the AGPL-3.0 license change (v4.10+) means all modifications must be open-sourced if distributed — commercial licence from Vendo required for proprietary use.

Your team	Platform	Why
Rails	SoSolidus or SpSpree	All Rails conventions, ActiveRecord, gems. Solidus for MVC; Spree for API-first + marketplace.
TypeScript / Node.js	MeMedusa or VnVendure	Medusa for best DX + fastest community. Vendure for B2B complexity + NestJS teams.
Python / Django	SaSaleor	Best multi-language + multi-currency. Mature GraphQL. Enterprise-grade (Lush, Breitling).
No dev team	Shopify	Managed infrastructure, no hosting decisions, ship in days. Graduate when you hit real limits.

Rails

PlatformSoSolidus or SpSpree

WhyAll Rails conventions, ActiveRecord, gems. Solidus for MVC; Spree for API-first + marketplace.

TypeScript / Node.js

PlatformMeMedusa or VnVendure

WhyMedusa for best DX + fastest community. Vendure for B2B complexity + NestJS teams.

Python / Django

PlatformSaSaleor

WhyBest multi-language + multi-currency. Mature GraphQL. Enterprise-grade (Lush, Breitling).

No dev team

PlatformShopify

WhyManaged infrastructure, no hosting decisions, ship in days. Graduate when you hit real limits.

The fully custom path is warranted when:

1. The product model itself is the business — not just how you sell it
2. Pricing logic is genuinely novel — not “complex” but actually impossible to model in any framework
3. The industry has no off-the-shelf fit — the product category has unique data, regulatory, or workflow requirements no platform contemplated
4. You have already tried platforms — and the workarounds cost more than the custom build
5. The founder wants to own the engineering outcomes — technical differentiation IS the competitive moat

High Stickers — when you outgrow any platform

High Stickers had been running on a legacy PHP e-commerce platform for 10 years. They went through a refactor attempt first, then came to me when the complexity proved too much for off-the-shelf solutions. We rebuilt the entire platform from scratch in Rails: 53 domain models, a custom pricing engine, and an operational command centre. The founder became the top code contributor on his own codebase (796 of 2,241 commits). That cultural outcome — a non-technical founder owning his own product — is something no SaaS platform can give you.

Cost and timeline reality

• Simple custom platform: €50,000–150,000 (small team, 3–6 months)
• Complex custom platform (High Stickers scale): €150,000–500,000+ (multi-year engagement)
• Annual maintenance: €80K–150K+ for a senior Rails developer + €500–5,000/month infrastructure
• The financial argument: At €50M+ GMV/year, Shopify Plus costs €480K/year in platform fees alone. Custom build maintenance at €100K–200K/year breaks even within 2–3 years.
• Below €5M GMV: Custom build almost never financially justified

European e-commerce reached €819 billion in 2024. 77% of EU internet users aged 16–74 made an online purchase. The market is enormous — and the regulatory landscape shapes platform decisions in ways that US-focused comparisons miss entirely.

B2B e-invoicing mandates (2026–2028)

No mainstream e-commerce platform natively handles EU B2B e-invoicing (Peppol/XML format). Belgium is live since January 2026. Poland since February 2026. France goes live September 2026 for large/mid-sized companies. Germany follows January 2027. This is a universal gap requiring third-party solutions (Quaderno, Billomat) regardless of platform. Shopify’s PDF invoices (added December 2024) are NOT compliant.

Payment method coverage

Method	Shopify	WooCommerce	MeMedusa	SaSaleor
iDEAL (NL)	✓ (restrictions)	✓ Mollie	✓ Stripe	✓
Bancontact (BE)	✓	✓ Mollie	✓ Stripe	✓
Wero (new)	✗ Not yet	Via Mollie	Via Adyen	Via Adyen
Klarna	✓ Native	✓ Plugin	Custom only	✓ Official
SEPA Direct Debit	✓ Via Stripe	✓	✓	✓
Mollie (no lock-in)	Via plugin	✓ Native	Via plugin	Via app

Key developments: Giropay was discontinued December 31, 2024 — remove from all checkouts. SOFORT was consolidated into Klarna (March 2025). Wero, the pan-European instant payment scheme backed by 16 banks and 1,100+ organisations, launched for e-commerce in Germany (November 2025) and Belgium/France (January 2026). Shopify does not yet support Wero. Headless platforms can integrate it today via Adyen.

Shopify’s critical EU limitation: Multi-currency checkout requires Shopify Payments. EU merchants preferring Mollie or Adyen cannot offer local-currency checkout. Medusa (regions) and Saleor (channels) have the best native multi-currency for EU multi-market selling without payment gateway lock-in.

GDPR and data sovereignty

Self-hosted platforms (WooCommerce, Solidus, Spree, Medusa, Saleor, Vendure) give maximum data control — your server, your database, no third party holds customer data. Shopify and BigCommerce are US-based SaaS with DPAs available but data on US infrastructure. For strict GDPR requirements, self-hosted is the clearest path.

Forget the feature matrix for a moment. The right platform depends on three things: your product model, your team, and your growth trajectory.

Cost at different scales

Annual GMV	Shopify	WooCommerce	MeMedusa	Custom
<€500K	€350–2,400/yr	€1,500–3,000/yr	€350–1,500/yr infra	Not cost-effective
€500K–5M	€950–10K/yr	€5K–20K/yr	€3,600/yr + dev	€50K–150K build
€5M–50M	€28K–50K+/yr	€20K–100K+/yr	€10K–50K/yr	€100K–200K/yr
>€50M	Up to €480K/yr platform fee	€50K–200K+/yr	€0 GMV fees + infra	€100K–200K/yr maintenance

The default recommendation: start with Shopify

For founders launching and non-technical operators: Shopify Basic → Grow → Advanced → Plus as GMV grows. The exit criteria matter more than platform sophistication. The goal is learning your product model before over-engineering your tech stack.

Specific triggers to start on Shopify:

• Straightforward DTC catalog (physical products, standard variants)
• Under €500K projected first-year revenue
• No existing technical team
• Speed to market is competitive advantage
• Migrating from Magento 1 (end-of-life pressure)

Specific triggers to graduate from Shopify:

• Frontend performance measurably impacting conversion at scale (with A/B test data, not assumptions)
• Running 5+ storefronts from one backend (multi-brand, multi-market)
• Shopify’s Checkout Extensibility genuinely cannot support your checkout flow
• Technical team has the expertise AND ongoing maintenance bandwidth
• The Shopify tax exceeds the cost of maintaining your own platform

The full comparison across the six most relevant platforms for 2026. PSPrestaShop is excluded from this matrix because of its declining trajectory; SoSolidus is excluded because its feature set overlaps with SpSpree (choose Solidus for traditional Rails MVC, Spree for API-first).

Feature	Shopify	BCBigCommerce	WooCommerce	MeMedusa	SaSaleor	SpSpree
Hosted/managed	✓	✓	✗	Cloud option	Cloud option	✗
Multi-currency	✓ (SP lock-in)	✓ Native	Plugin	✓ Native	✓ Native	✓
Multi-language	✓ Markets	✓ MSF	Plugin (WPML)	CMS needed	✓ Native	Limited
Multi-storefront	Plus only	✓ All plans	Plugin	✓ Channels	✓ Channels	✓
Native B2B	Plus only	✓ B2B Edition	Plugin	Basic	Limited	Enterprise
API-first	✓ GraphQL	✓ GQL+REST	Partial	✓ REST	✓ GraphQL	✓
Open source	✗	✗	✓	✓ MIT	✓ BSD-3	✓ AGPL
Transaction fees	0.2–2.0%	0%	0%	0%	0% (OSS)	0%
Marketplace native	✗	✗	Plugin	Plugin	Limited	✓
Custom checkout	Plus Extensibility	Catalyst	✓ Custom	✓ Full	✓ Full	✓ Full
Stack	Liquid / React	Handlebars / React	PHP / WordPress	TypeScript / Node	Python / Django	Ruby / Rails
App/plugin ecosystem	8,000+	1,000+	70,000+ WP	Dozens	App Store	Gems

After rebuilding a 10-year legacy platform, building custom e-commerce from scratch for niche industries, integrating Shopify for DTC brands, and evaluating every headless option on this list for real clients — here is what I actually tell people:

Start on Shopify unless you have a specific, articulated reason not to. “Shopify limits us” is not specific. “Our pricing logic requires contract × customer segment × quantity break × region, and it changes daily” is specific. “We want more control” is not specific. “We need a marketplace with vendor panels, commission management, and automated payouts” is specific.

When you graduate, choose based on your team, not on feature lists. A Rails team building with Medusa is fighting their own toolchain. A Python team trying to work with Solidus is in the wrong ecosystem. The best platform is the one your team can maintain at 2 AM when something breaks.

Go fully custom only when the three questions all point the same way:

1. Can you express your pricing/product model in any existing framework’s data model? If no: custom.
2. Is the technical differentiation core to your competitive moat, or is it operational complexity you would rather not own? If moat: custom.
3. Do you have (or can you hire) committed engineering talent for a 3+ year build? If no: you are not ready for custom, regardless of the product model.

And the European angle matters more than most comparisons acknowledge. If you are selling B2B in France, Germany, or Belgium, e-invoicing compliance is coming for you in 2026–2027. If you need iDEAL without restrictions, Wero support, or Mollie without Shopify Payments lock-in, the headless and self-hosted platforms are genuinely better-positioned than Shopify today.