Headless Commerce Is Turning Platforms Into Checkout Companies

For about fifteen years, the deal with a commerce platform was simple. You bought the whole thing: catalog, storefront, cart, checkout, orders, admin. It was one system, one login, one vendor to yell at. The platform’s value was that it did everything adequately, and the lock-in was that leaving meant leaving everything at once.

Headless broke the deal, and I don’t think the platforms have fully reckoned with how much.

The front end walked out the door

Headless commerce means the storefront is decoupled from the platform. The platform exposes APIs; someone else builds the thing customers actually see and touch.

That “someone else” is the merchant’s own team or their agency. And that matters, because the storefront is the single most visible, most differentiating, most billable part of a commerce build. It’s where brand lives. It’s where conversion lives. It’s where an agency earns its keep.

Once the platform isn’t rendering the storefront, it has given up the layer that customers form opinions about. Nobody has a feeling about a checkout API. Everybody has a feeling about the site.

APIs got easy, and then AI made them trivial

Ten years ago, “go headless” meant a large budget and a team that knew what it was doing. The APIs were incomplete, the docs were thin, and you were on your own for everything the platform used to handle.

That’s not the world anymore. Storefront APIs matured. GraphQL made shaping the data you need routine. Frameworks like Next.js and Astro turned a performant storefront into a weekend project for a competent developer.

And now AI-assisted development has dropped the floor again. A developer working with a coding agent can stand up a working storefront against a commerce API in days, not months. The knowledge that used to be the barrier — how do I structure this, what does the cart mutation look like, how do I handle variants — is available on demand.

When the hard part gets easy, the platform loses its second argument: “you couldn’t build this yourself.” You can. Cheaply.

Nobody wants to manage products in a platform admin

Here’s the part that I think gets underappreciated, especially by people who mostly see DTC brands with a few hundred SKUs.

For any catalog-heavy business — distributors, industrial suppliers, automotive, anyone with tens of thousands of SKUs and real attribute data — the platform admin is not where product data lives. It can’t be. It’s a lossy copy of the truth, which lives in an ERP or a PIM or, honestly, a very large spreadsheet somebody guards with their life.

Fitment. Contract pricing. Account-specific catalogs. Multi-warehouse availability. Units of measure. The platform’s product model was built for t-shirts in three sizes, and every serious B2B implementation spends its first six months bolting extensions onto it to make it stop being wrong.

So the catalog isn’t really the platform’s either. It’s syncing in from somewhere else, and the platform is a middleman that mostly gets in the way.

What’s left is checkout

Take away the storefront. Take away the catalog as system of record. Search has been a separate vendor for years. Order management for anything complex is its own system.

What remains that the platform does well and that you shouldn’t build yourself? Checkout. Payments, tax, fraud, PCI scope, wallets, buy-now-pay-later integrations. That’s hard, it’s regulated, and it benefits from scale.

Which makes Shopify, at the end of this process, a checkout company.

And I want to be clear: that’s fine. Checkout is a fantastic business. Stripe built a giant company doing roughly that and nothing else. Shop Pay is a real asset. Being the checkout layer is not a consolation prize.

But it’s a very different business than being the monolith. The monolith charged you for everything and made it painful to leave. The checkout layer competes on price and conversion rate, and can be swapped.

The monolith gets chipped, not toppled

Nobody rips out a platform in one move. What actually happens is that each layer gets pulled out when it becomes the bottleneck.

The storefront leaves first, because that’s where the pain and the budget are. Then search, because native search is never good enough for a big catalog. Then the catalog itself, once someone gets tired of the sync job. Then pricing and availability, because the ERP already has the answer and the platform is just in the way.

Each one of those is a piece you can now choose independently. Each one is a piece that used to be part of the reason you couldn’t leave.

The platforms see this. It’s why they’ve invested so heavily in their own headless frameworks and hosting, why they push their storefront tooling so hard, why the “best” way to go headless is always presented as staying inside their ecosystem. It’s an attempt to keep the storefront in-house even as the architecture says it doesn’t need to be.

Where I land

I run a consultancy that’s built on commerce platforms for a long time, and I’m building products that live in exactly the space this opens up. So I’m not neutral. But I’d say this even if I were.

If you’re a merchant, especially a catalog-heavy one: the platform is not the center of your commerce architecture anymore. Your index is. Your ability to answer “what do we have, what does it cost for this customer, and can we ship it” quickly and correctly is the asset. The platform is one consumer of that.

If you’re an agency: the front end walking out of the platform is the best thing that ever happened to you. Own it.

And if you’re a platform: being the checkout company is a good outcome. Just don’t price like the monolith once you’re not one.