Medusa vs Shopify: which one fits, and when switching is a mistake

Choose Shopify if you want to be selling within days and have nobody to keep a platform running. Choose Medusa if you can point at a limitation that blocks your sales model, and you have a team or a partner willing to own the code, the hosting and the upgrades.

These are not two versions of the same thing. One side hands you a finished product and keeps control of the checkout and the infrastructure. The other hands you control and hands you the maintenance duty with it. Every other difference follows from that.

One note on the numbers before you start comparing them. A Shopify subscription and the cost of building a Medusa platform are two different lines, not two sides of one bill. Every platform needs implementing, Shopify included. That is why the table below separates them.

Eleven dimensions, side by side

Current as of 12 August 2026. Plan prices, quotas and deadlines come from both vendors’ pricing pages and documentation (checked 11 August 2026); the implementation ranges come from our own review of industry publications, compiled 12 August 2026. Shopify plan prices are localised per market, so the only plan-level figures quoted here are the ones the sources state for every market. Implementation ranges differ several-fold between publishers, so read them as orders of magnitude rather than as a quote. Our own ranges are set in Polish złoty (from 60,000 PLN, typically 80,000 to 200,000 PLN) and converted at a rounded 4.30 PLN to 1 EUR, the NBP mid rate of 11 August 2026. Sources and publishers are listed at the end.

A licence is not the cost of a platform

The most common error in comparisons like this one, including in the first version of this page: a Shopify subscription on one side, the cost of building a Medusa platform on the other. That puts a rental price next to a construction price, and what comes out is not true.

The honest split has two lines. Licence: Shopify charges a plan fee from day one, Medusa charges nothing. Implementation: both need it, because no platform turns itself into a store holding your catalogue, your pricing rules and your integrations.

That proportion cuts both ways, which is why it is worth knowing. On the Shopify side it means the subscription is the visible part of the bill, not the whole bill. On ours it means a zero in the licence row excuses nobody from costing the build.

What these numbers do not prove

We do not claim that building on Medusa is cheaper than building on Shopify Plus at a comparable scope. The public data does not settle it: Medusa publishes no implementation ranges, and the ranges quoted for Plus overlap with ours.

Part of the gap you see in comparisons like this is geography rather than technology. Shopify Plus Certified Partners quote 175-300 USD per hour; agencies in Poland typically quote 50-99 USD. The same work costs less here whatever it is built on, and we are one of those agencies.

Time to launch depends on what you are comparing

Same trap. A store on a stock theme goes live faster than we could finish planning a project, and that will not change. At a complex scope the gap narrows: Medusa takes 8-12 weeks, and migrating onto Shopify Plus from WooCommerce or BigCommerce takes 12-16 weeks.

Those two numbers only belong next to each other when the scope on both sides is the same. If yours includes an ERP integration and an order history migration, “live in days” does not describe your project on either route.

So the solid argument sits somewhere other than the cost of going live once. It sits in the row labelled cost to change after launch, which looks like the least dramatic line in the table and then recurs every month for three years.

Who this choice is actually for

Before you read on, check whether this decision is yours to make. The question “Medusa or Shopify” is a real question for a narrow set of stores. For everyone else the answer is Shopify and there is nothing to weigh. Five markers tell us a conversation about an owned platform has a basis:

• Your catalogue does not fit a standard product model: made-to-measure goods, bundles, configurators, prices derived from parameters, technical documentation attached to the item. Variants will not cover it.
• Selling logic is an advantage rather than an add-on: contract prices per customer, order approvals, credit limits, subscription rules of your own. Usually the hardest of the five to work around.
• ERP and warehouse integrations set the pace of operations: they decide how fast the company works, so you do not want somebody else’s request quota setting their rhythm.
• You sell through several channels at once: the store, marketplaces, wholesale, sometimes an integrator on top. Every channel wants different data and different rules, and all of them come back to one stock figure.
• Revenue is high enough that a percentage of sales starts to matter: a charge calculated on turnover and a stack of app subscriptions grow with sales; the margin does not necessarily follow.

How many have to be true? One is not enough, because a single problem can usually be absorbed by an app or a higher plan. At two it is worth doing the arithmetic. At three the workarounds start costing more than building the thing properly.

If none of them sounds familiar, the rest of this piece is background reading rather than a decision. Without those markers you pay for a build and for maintenance, and you get what a subscription already gave you. That spend has no way of returning.

Where the difference stops being theoretical

The table says how the platforms differ. What matters is where that difference is felt, and three places settle most decisions: the bill, the checkout and the integration ceiling.

The bill: what scales with what

On Shopify the cost follows revenue. You pay a plan fee and processing on every order, and if you settle payments outside Shopify Payments there is an extra charge on sales: 2% on Basic, 1% on Grow, 0.6% on Advanced and 0.2% on Plus.

On Medusa the cost follows complexity. There is no licence fee, but there is infrastructure and there are people: Medusa Cloud starts at 29 USD per month on the developer tier and 99 USD on the tier positioned for going live, while self-hosting needs Node.js 20+, PostgreSQL and an owner. Which bill ends up lower is settled by a three-year TCO.

Checkout: the first hard wall

Checkout is where you meet the edge of the platform soonest. Shopify documentation is unambiguous about it: checkout styling customisations are available only to Shopify Plus merchants. Everything else goes through defined extension points rather than page code.

The second edge is somebody else’s calendar. Scripts and the checkout.liquid template on the Thank you and Order status pages were sunset on 28 August 2025 for Plus stores, and script tags are sunset for the rest on 26 August 2026. The deadline for rewriting those integrations is set by the vendor.

API limits: how much room integrations get

API quotas tell you how much room an ERP, a warehouse or a marketplace channel actually gets. The GraphQL Admin API runs on a points budget tied to your plan: 100 points per second on standard plans, 200 on Advanced, 1,000 on Plus and 2,000 on the enterprise offering.

On Medusa the limit is whatever your infrastructure allows. That is an advantage and a cost at once: nobody throttles your ERP sync, and nobody looks after your database performance or your queues either. The same trade, seen from the integration side.

The honest case for Shopify

We write this as a company that builds on Medusa, which is exactly why it is worth saying plainly: for a large share of stores Shopify is the right answer, and switching would be spend without a return.

• You want to sell this month: an account, a theme from the catalogue and your products are enough. An owned platform is measured in weeks, not days.
• Nobody is available to run a platform: hosting, updates and payment compliance stay with the vendor. A platform nobody looks after is a worse problem than fees.
• You need an integration that already exists: with over 16,000 apps, the common requirement is a configuration task rather than a project.
• You sell in several markets on one operation: the Plus plan adds expansion stores, unlimited staff accounts, round-the-clock priority support and a published 99.99% uptime SLA, none of which you have to build.
• Your catalogue and processes fit the standard: a straightforward assortment, one sales channel, no contract pricing and no per-customer terms.

None of that is politeness towards a competitor. If none of the limits above blocks your sales model, an owned platform buys control you will not use and adds a duty you did not need.

The case for Medusa

The markers under “Who this choice is actually for” tell you whether the problem is yours. This section asks something else: whether you can afford the answer. Three conditions, and good intentions substitute for none of them.

• The limitation can be named and priced: not “we want headless”, but “we cannot quote a contract price per customer without doing it by hand”. Work out what that manual handling costs per month. Without that figure the decision is a preference, not a calculation.
• The arithmetic has turned: processing fees, turnover-based charges and app subscriptions grow faster than sales, and cost per order rises instead of falling.
• Someone will own it: an internal team or a maintenance partner under contract, not an assumption that it will be looked after. That is a budget line, not goodwill.

If any of those sound familiar, start by naming the problem rather than picking a technology. The twelve signs of outgrowing a SaaS platform help with the naming, and the choice between platform families is laid out in criteria instead of a ranking.

What neither pricing page covers: AI automation and agents

This dimension stayed out of the table because it cannot be set out column against column. Nobody sells automation built around your process as part of a subscription. The difference is what such a workflow can reach, and who draws its boundaries.

On somebody else’s platform, automation runs at the extension points the vendor defines and inside API quotas tied to your plan. On a backend you own, it reaches the data model, the order flow and the integrations directly, because you set the permissions.

For us this is a separate stage of the framework with its own price: AI Automation is scoped from about 1,400 EUR per workflow (6,000 PLN, converted at a rounded 4.30 PLN to 1 EUR, the NBP mid rate of 11 August 2026). The scopes we build most often:

• Product descriptions: generating and standardising copy from catalogue data, with a brand glossary and a sample check before anything publishes.
• Catalogue translation: bulk localisation into new markets, covered separately in AI catalogue translation.
• Marketplace feeds: preparing and policing the data that goes out to sales channels, each of which wants it differently.
• Order status handling: answering “where is my parcel” from the data in your system rather than from a model’s general knowledge.
• Attribute mapping: pulling supplier files into one schema, in the common case where every file looks different.

The second thread is running a store through AI agents. An agent meant to read stock levels, correct product data and prepare price changes needs access to the backend, to the data and to clearly defined permissions. On a platform you own, you design that access alongside everything else.

This is a range, not an offer. The exact figure comes out of a conversation about a specific workflow: how much data, how many rules, and what has to be checked before publication. The full billing model sits on our pricing page.

What the open source label does not cover

An open licence does not mean free forever. The Medusa repository core is MIT licensed, but role-based access control belongs to the Enterprise Edition and requires a separate commercial agreement with MedusaJS.

The documentation also points to Medusa Cloud first and self-hosting second. The Cloud pricing page states zero platform fees on GMV and the option to export your data and run Medusa yourself, which is a question worth settling before the build, not after it.

What this comparison leaves out

The honest boundaries of this piece, so it is not read as more than it is:

• Performance under load: we have no measurements of both platforms under the same conditions, so we make no claims about it.
• Negotiated terms: Shopify Plus starts at 2,250 euro per month on a one-year term and 2,100 euro on a three-year term, and larger businesses are offered a variable platform fee based on revenue and business model, agreed with sales.
• Local pricing: Shopify shows plan prices in the currency of the visitor market, so the only plan-level figures we quote here are the percentages the pricing page states for every plan.
• An answer to who builds cheaper: Medusa publishes no implementation ranges, and the ranges quoted for Shopify Plus overlap with ours. The public data does not settle that question, so neither do we.
• A quote for your build: the ranges in the table are orders of magnitude on both sides, not an offer. Catalogue size, integration count and migration scope move the result in both directions.
• Other platforms: WooCommerce, PrestaShop, Saleor and regional SaaS vendors follow a different cost logic and deserve their own comparison.

Medusa is the foundation, BEAM is how it ships

Medusa is a technical foundation rather than a finished product. What we build on it is BEAM: one place a client runs their whole eCommerce from, plus the four stages we deliver it in. The build-or-stay decision is made during the Blueprint.

If the decision leans towards an owned platform, the move itself, including redirects and customer data, is described on our Shopify migration page, and the sequencing that keeps the store selling throughout is covered in migration without stopping sales.

Current as of 12 August 2026

Prices, quotas and deadlines were checked at source on 11 August 2026, and the implementation ranges and timelines were compiled on 12 August 2026. We revisit this page quarterly, because an out-of-date comparison does more harm than good. If a figure has moved since, tell us and we will correct it together with the check date.

• Shopify pricing: shopify.com/pricing and shopify.com/plus/pricing.
• API limits and checkout: rate limit documentation, checkout styling and the sunset timeline for scripts.
• App count: apps.shopify.com. Plus plan features: Shopify Help Center.
• Medusa: Medusa Cloud pricing, installation requirements and the repository licence.
• Shopify implementation cost: the “$2,000 to $20,000 or more” range for custom development comes from the Shopify blog. Mid-market replatforming onto Plus is priced by agencies at 50,000-120,000 USD (Blackbelt Commerce), 50,000-150,000 USD (IWD) and 80,000-250,000 USD (Elogic). Polish agencies quote roughly 3,000-15,000 PLN for a simple themed store and 20,000-80,000 PLN once integrations and a data migration are involved.
• Licence as a share of project cost: the 20-40% figure comes from Elogic, an agency that implements Shopify, in its material on replatforming costs. Hourly rates: Shopify Plus Certified Partners 175-300 USD, agencies in Poland in a 50-99 USD band (Clutch data).
• Implementation timelines: Medusa 8-12 weeks, a figure agencies on several markets converge on; the vendor itself publishes neither a time nor a cost to launch. Shopify: 2-4 weeks for a simple store, 12-16 weeks to migrate onto Plus from WooCommerce or BigCommerce.
• Our ranges and billing model: the pricing page. The two complete packages compared with a three-year bill: BEAM + Medusa vs Shopify Plus and an agency.