"Headless e-commerce" gets used for everything from "Shopify with a custom theme" to "we rebuilt all of Amazon". This post defines what we mean at Schedars in 2026, when headless actually wins, and the specific stack we use when we go that route.

TL;DR

Headless = decoupling the storefront UI from the commerce backend, communicating via API. It wins above $5M GMV/year when standard Shopify perf or flexibility hits a ceiling. Below that, monolithic Shopify (theme + apps) is faster, cheaper, and equally good. Our 2026 default headless stack: Medusa.js for the backend, Next.js App Router for the storefront, Stripe for payments, Sanity for content, Algolia for search.

What "headless" actually means

In a headless setup, the commerce engine (cart, catalog, checkout, inventory, orders) is one service. The storefront (the UI customers see) is a separate app that calls the engine via API. They deploy and scale independently. The "head" is the storefront; the "body" is the commerce engine.

What headless is NOT: a custom theme on Shopify is not headless. Liquid templates rendered server-side by Shopify are still tightly coupled. True headless means you’re running your own frontend (Next.js, Astro, Nuxt) that calls Shopify Storefront API or Medusa REST/GraphQL.

When headless makes sense

We push back on headless requests when they’re vibes-driven ("we want flexibility"). We support headless when there’s a specific business reason. Use the four-question test:

  • Is your annual GMV above $5M and Shopify’s 0.5–2% fees are real money? Headless pays back.
  • Does your storefront need integrations or page types Shopify themes can’t cleanly support (configurators, AR try-on, complex bundle builders, B2B portals)?
  • Are you launching omnichannel where the same backend serves web, mobile, kiosk, point-of-sale? Headless is mandatory.
  • Does brand experience (perf, custom interactions, design system) matter more than time-to-market? Headless wins on the perf ceiling.

Two yeses = headless probably wins. One yes = you can argue both ways. Zero = stick with Shopify and a custom theme.

The 2026 stack we use

Backend: Medusa, Saleor, or Shopify Storefront API

Three credible options in 2026. Medusa.js (open-source, self-hosted, Node-based) is our default — control + ecosystem + matched 70% of Shopify Plus features. Saleor (open-source, Python+GraphQL) is strong for B2B and complex catalog work. Shopify Storefront API is the right call when you want headless storefront but need to keep Shopify’s checkout, payment, and admin.

Frontend: Next.js, Astro, or Hydrogen

Next.js (App Router with RSC) is the default for app-like storefronts with auth-gated content, account pages, complex carts. Astro is right for content-heavy storefronts where most pages are static — better Lighthouse scores out of the box, smaller bundles. Shopify Hydrogen is a Remix-based framework optimized for Shopify Storefront API — good if you’re committed to Shopify backend.

Search, CMS, payments

Search: Algolia for managed (instant facets, autocomplete, $$$), Meilisearch for self-hosted (cheaper, slightly less polish). CMS for content + product enrichment: Sanity (structured content, real-time collab) or Contentful. Payments: always Stripe for custom; Shopify Payments if you’re hybrid on Shopify.

Cost reality

A custom headless build in 2026 is $80k–$200k for the initial launch with the standard feature set: catalog, cart, checkout, account, basic admin, search, content. Add $20k–$50k per advanced feature (subscription, B2B portal, AR try-on, multi-currency).

Maintenance: 1 engineer ongoing minimum for security patches, dependency updates, framework version bumps, integration evolution. Realistic ongoing cost: $80k–$150k/year for a small team. Compare to Shopify Plus at $24k/year for the platform plus app subscriptions — you need to be saving more than $100k/year on Shopify fees and platform-feature gaps to justify custom.

Common mistakes we see

  • Headless to "future-proof" without a current pain point — sunk cost trap
  • Rolling custom checkout without PCI scope reduction (Stripe Elements + setup intents) — multiplied compliance burden
  • Underbudgeting search (treating it as "we’ll add Algolia later") — search is 30%+ of conversion in commerce
  • Skipping the admin panel because "we’ll use the database directly" — operations breaks first month
  • Building the storefront before the catalog data model is locked — every storefront refactor follows a backend change

What we ship at Schedars

Two recent headless e-commerce builds: a B2B-meets-DTC hybrid (custom on Next.js + Medusa + Stripe Connect, $140k initial, 16 weeks) and a premium DTC brand storefront (headless Hydrogen + Shopify backend, $90k initial, 11 weeks). Both clients had specific business reasons — the B2B hybrid needed multi-step quote → invoice → fulfillment that Shopify couldn’t support; the DTC brand needed Lighthouse 95+ on a content-rich storefront and the existing Shopify Plus theme couldn’t hit it.

Bottom line

Headless e-commerce is the right answer when monolithic Shopify can’t support your business model or your scale makes Shopify fees real. It’s the wrong answer when "we want flexibility" is the only justification. The cost is real, the maintenance is real, the wins are real — pick based on numbers, not vibes.

Weighing headless for your store? Send us the GMV, the catalog size, and the integration list — we’ll tell you what we’d build and why.