Every founder gets sold the same story. Pick a hosted platform, pay the monthly fee, add your products, and never think about the plumbing again. For the first year, that story is mostly true. Then you try to do one thing the platform was not built for, and you find out exactly how much of your business you handed to someone else.
We ran our storefront the normal way for a while. Then we moved it to Medusa, an open-source commerce engine, and rebuilt the front end ourselves. This is the honest version of why we did it, what it actually cost, and who should absolutely not follow us.
What "headless" actually means, in plain English
Strip the jargon out of it. A traditional platform welds four things together: the storefront your customer sees, the cart, the checkout, and the admin you manage it from. It is one sealed box. That is why it is easy, and it is also why you hit a wall the moment you want the storefront to behave in a way the box did not anticipate.
Headless commerce splits the box in two. The commerce engine (products, inventory, carts, orders, payments, customers) runs on its own. The front end (the actual pages, the design, the checkout experience) is yours to build however you want, calling the engine through an API. You get the boring, hard, must-not-break commerce logic as a foundation, and total freedom on everything the customer touches.
The bill nobody reads until it hurts
Hosted commerce is cheap until you are successful, and then it becomes a tax on being successful. The plans themselves are not the part that hurts. The percentage on every sale is (Shopify, n.d.).
| Plan | Monthly | Online card rate | Extra fee if you skip their processor |
|---|---|---|---|
| Basic | $39 | 2.9% + 30¢ | +2.0% |
| Grow | $105 | 2.7% + 30¢ | +1.0% |
| Advanced | $399 | 2.5% + 30¢ | +0.6% |
| Plus | from $2,300 | Negotiated | +0.2% |
Do the math on real volume. At a few hundred thousand dollars a year in sales, the processing and platform fees are not a rounding error. They are a real line on your P&L, they scale with your growth, and you do not control them. You are renting your own checkout.
Why Medusa
Medusa is that engine, and it is open source (Medusa, n.d.-a). It bills itself plainly as the open-source Shopify alternative, and it has the traction to back the claim. It raised an $8 million seed round led by LocalGlobe and Dawn Capital (Rindom, 2022), and it is one of the most-starred commerce projects on GitHub, with a large developer community behind it (Medusa, n.d.-b). That matters, because open source is only a real option when the project is alive.
We run Medusa v2 with a Next.js storefront. Here is what that buys us that a hosted plan never could:
- We own the code and the data. Our products, orders, and customers live in our database, not in an account we are renting.
- There is no per-feature paywall. When we wanted our own product manager, our own admin console, and our own first-party analytics, we built them. We did not file a support ticket asking permission.
- We are not paying a percentage of every sale to a landlord. We pay for hosting, which is cheap, and for the processor we actually chose.
What it actually cost us
This is where most "we switched and it is amazing" articles go quiet, so I will not.
Headless is not free. You are trading a monthly fee and a sales percentage for engineering time. Somebody has to build and maintain the front end, keep the engine updated, handle security, and own uptime at 2 a.m. On a hosted platform, that is someone else's job and it is priced into your bill. On Medusa, it is yours.
We could take that trade because we already build software. A large share of our storefront was written with AI agents doing the heavy lifting under our direction, which is the only reason a small team can own this much surface area. If you do not have engineering capacity, in-house or contracted, that cost does not disappear. It just shows up later as a stalled project.
Who should not do this
I will talk you out of it if you are the wrong fit, because watching someone rebuild their store out of principle and lose three months of selling is painful.
| Stay on a hosted platform if… | Move to Medusa if… |
|---|---|
| You sell a handful of products and nobody is technical | Your volume is high enough that the platform's cut is real money |
| The fees do not hurt yet | You need customization the platform flatly will not allow |
| "Excellent and boring" beats "custom and broken" | You already have engineering capacity, in-house or contracted |
Independence is worth a lot. It is not worth rebuilding something that already worked just to say you own it.
The part that actually mattered
Under all the technical arguments, the real reason we moved is ownership. When your store is your business, renting the foundation is a risk you only feel the day you want to change something and cannot. We wanted to change a lot, and we wanted to keep every decision in-house.
Medusa gave us a commerce engine we control and a front end we can take anywhere. For a company that builds its own software, that was not a close call. For a company that does not, it might be the wrong call entirely. Both of those can be true, and knowing which one you are is the whole decision.
References
Medusa. (n.d.-a). Medusa: The open-source Shopify alternative. https://medusajs.com/
Medusa. (n.d.-b). medusajs/medusa [Source code repository]. GitHub. https://github.com/medusajs/medusa
Rindom, S. (2022, July 15). Announcement: Medusa's $8M USD seed round to build the leading e-commerce platform for developers. Medusa. https://medusajs.com/blog/announcement-8m-usd-seed-round-to-build-the-leading-ecom-platform-for-devs/
Shopify. (n.d.). Pricing. https://www.shopify.com/pricing