Medusa vs Shopi­fy Plus: the B2B math no ven­dor puts on the ta­ble

The Medusa vs Shopi­fy Plus com­par­i­son is al­most al­ways run as a fea­ture duel, and al­most al­ways run wrong. Lay two check­box lists side by side and you've mis­tak­en a strate­gic de­ci­sion for a pro­cure­ment ex­er­cise.

Shopi­fy Plus is fast, com­fort­able, and ex­cel­lent­ly main­tained. Medusa is open source, self-hostable, and hands you full con­trol over data mod­el and process­es. Both sen­tences are true, and both say lit­tle about the real ques­tion: do you want to rent your process­es or own them, and will you car­ry the op­er­a­tional re­spon­si­bil­i­ty that own­er­ship means? The case is de­cid­ed at that fork, not by whether plat­form A has a fea­ture plat­form B lacks.

A fea­ture list com­pares two things that aren't ar­chi­tec­tural­ly com­pa­ra­ble. Shopi­fy Plus is a host­ed SaaS suite: you get a fin­ished prod­uct that a US ven­dor runs, up­dates, and se­cures for you. Medusa is a kit of de­cou­pled com­merce mod­ules that you as­sem­ble, de­ploy, and main­tain your­self. One is a car on a lease; the oth­er is a very good build kit plus a work­shop.

Once you ac­cept that, the ques­tions that mat­ter shift away from the fea­ture set to­ward four strate­gic axes: how cost be­haves over time, con­trol over process­es and check­out, data sov­er­eign­ty, and op­er­a­tional re­spon­si­bil­i­ty. A head­less com­merce ap­proach like Medusa wins none of these au­to­mat­i­cal­ly. It only moves them into your hands, with every­thing that en­tails. If you don't want that shift, take Shopi­fy Plus and stop com­plain­ing about lock-in you know­ing­ly bought.

The base fee sits open­ly on Shopi­fy's pric­ing page: as of mid-2026, Plus starts at €2,250 per month on a one-year con­tract and €2,100 per month on a three-year con­tract, each quot­ed as "start­ing at" (ver­i­fy). That's the pre­dictable, hon­est part. The dis­hon­est part is every­thing that grows with rev­enue.

First, the trans­ac­tion fees. Run a third-par­ty PSP as your pri­ma­ry gate­way and you pay its fees *plus* 0.20 per­cent per trans­ac­tion to Shopi­fy, which the ven­dor at­trib­ut­es to "se­cu­ri­ty and com­pli­ance costs." Only Shopi­fy Pay­ments waives these third-par­ty fees glob­al­ly; then only the coun­try-de­pen­dent card rates ap­ply. It's a clean in­cen­tive to stay in­side the ven­dor's pay­ment ecosys­tem. Sec­ond, the "vari­able plat­form fee for more com­plex busi­ness struc­tures," which Shopi­fy prices rev­enue-linked, with no pub­lished fig­ure, through sales only. Third, the app costs: a ma­ture Plus store rarely runs few­er than a dozen paid apps, each its own month­ly line.

The point is­n't that Shopi­fy is ex­pen­sive. The point is how the cost be­haves. Trans­ac­tion fees scale with your suc­cess. The bet­ter the store runs, the more ex­pen­sive the plat­form gets, and that re­la­tion­ship shows up on no sales slide, be­cause it makes the prod­uct cost­lier for ex­act­ly the cus­tomers who are hap­pi­est with it.

As open-source soft­ware un­der an MIT li­cence, Medusa it­self charges no trans­ac­tion or rev­enue fee. You car­ry only in­fra­struc­ture and pay­ment-provider costs. Use Stripe and you pay Stripe's card fees, but no sur­charge to a com­merce plat­form sit­ting be­tween you and your mon­ey.

The in­fra­struc­ture is real and has to be quan­ti­fied hon­est­ly. A pro­duc­tion Medusa stack needs Post­greSQL and Re­dis, two op­er­at­ing modes (a serv­er for API and ad­min, a work­er for back­ground jobs), and at least 2 GB of RAM. Self-host­ed on Het­zn­er in Ger­many, the en­try lev­el runs about €20 to €50 per month, and a high­ly avail­able set­up with a re­dun­dant, sep­a­rate data­base €150 to €400 per month (es­ti­mate). A clean first im­ple­men­ta­tion of a self-host­ed stack we bud­get as a one-off at €5,000 to €20,000, with on­go­ing op­er­a­tion at two to five hours a month. Over three years a mid-sized shop re­al­is­ti­cal­ly lands around €40,000 to €80,000 all in (part­ner fig­ure, ver­i­fy). We ran the full cal­cu­la­tion, sov­er­eign­ty ar­gu­ment in­clud­ed, in the Het­zn­er TCO com­par­i­son.

The de­ci­sive thing is the di­rec­tion of the curve. Self-host­ed costs fall over time as hard­ware gets cheap­er, op­er­a­tion be­comes rou­tine, and the one-off im­ple­men­ta­tion is writ­ten off. SaaS costs rise with rev­enue, be­cause trans­ac­tion fees and the vari­able plat­form fee are tied to growth. At low rev­enue Shopi­fy wins the math al­most every time. At high gross mer­chan­dise val­ue, with a third-par­ty PSP, or across sev­er­al cat­a­logue seg­ments, it flips. Where ex­act­ly the break-even sits de­pends on your spe­cif­ic vol­ume, but that the two curves cross is not an opin­ion, it's a con­se­quence of the pric­ing mod­els.

This is where the com­par­i­son gets in­ter­est­ing, be­cause both sides like to over­state. The old sto­ry that "B2B com­merce only runs on Plus at Shopi­fy" has been out­dat­ed since April 2026. Core B2B fea­tures, com­pa­ny pro­files, net pay­ment terms (net 7 to 90), vol­ume pric­ing, stored cred­it cards, and up to three ac­tive cat­a­logs, have been avail­able on all paid plans since. Plus-ex­clu­sive are un­lim­it­ed cat­a­logs as­signed di­rect­ly to in­di­vid­ual com­pa­ny lo­ca­tions, cus­tomer de­posits, par­tial pay­ments, and the ex­tend­ed check­out cus­tomiza­tion. That's sol­id, main­tained B2B func­tion­al­i­ty you don't have to wor­ry about.

Medusa's B2B side is the most mis­un­der­stood part of the com­par­i­son. Yes, there's com­pa­ny and em­ploy­ee man­age­ment, per-em­ploy­ee spend­ing lim­its with re­set fre­quen­cies, ap­proval work­flows at com­pa­ny and mer­chant lev­el, and quote man­age­ment with ne­go­ti­a­tion and bulk add-to-cart. But all of that lives in the B2B starter, which is ex­plic­it­ly boil­er­plate "de­signed to be cus­tomized and ex­tend­ed," ref­er­ence code to fork and self-main­tain. These fea­tures are im­ple­ment­ed there as cus­tom mod­ules (com­pa­ny, quote, ap­proval), not shipped in the Medusa core. The core it­self con­tains no com­pa­ny, quote, or ap­proval mod­ule.

That dis­tinc­tion is­n't a de­tail, it's the heart of the mat­ter. A Shopi­fy B2B fea­ture is a main­tained SaaS fea­ture with a roadmap and sup­port. A Medusa B2B fea­ture from the starter is code *you* own, patch, and keep cur­rent against Medusa up­dates from the first git clone. It's a de­gree of free­dom and an oblig­a­tion in one.

Now the hon­esty the pro-open-source camp tends to skip. Medusa's con­trol is worth ex­act­ly as much op­er­a­tional re­spon­si­bil­i­ty and build ef­fort as you can ac­tu­al­ly shoul­der, and not a euro more.

Two lim­its need nam­ing plain­ly. First: you don't re­build Shopi­fy's ecosys­tem. The app store, the pay­ment in­te­gra­tions, the ful­fill­ment con­nec­tions, the themes, that's the sum of tens of thou­sands of de­vel­op­er-years. Ex­pect the same breadth from Medusa and you're bad­ly un­der­es­ti­mat­ing the in­te­gra­tion work. Every con­nec­tion that's a sin­gle click in Shopi­fy's app store is a cus­tom mod­ule with its own life­cy­cle in Medusa.

Sec­ond, and this is the hard­er lim­it: Medusa's B2B starter is *not* en­ter­prise pro­cure­ment ei­ther. It de­liv­ers sim­ple com­pa­ny ap­provals and spend­ing lim­its, not mul­ti-stage ap­proval chains with cost-cen­tre log­ic, no real ERP bud­gets, no pun­chout, no cXML, no OCI, no req­ui­si­tion-to-PO, no net-terms in­voic­ing rec­on­ciled against the ERP. Those func­tions are demon­stra­bly ab­sent from the starter. Real en­ter­prise pro­cure­ment runs over pun­chout cat­a­logues, where the buy­er jumps live from their SAP, Ari­ba, or Coupa sys­tem into your store, and req­ui­si­tion, ap­proval, and pur­chase or­der hap­pen in the *buy­er's* sys­tem. You im­ple­ment that serv­er-side your­self, and we've tak­en apart sep­a­rate­ly why that gap mat­ters for de­mand­ing B2B cas­es: Medusa has no turnkey pro­cure­ment.

Fail to plan for that in­te­gra­tion work and you buy, with Medusa, a con­trol you can't ex­er­cise, and you'd have slept faster, cheap­er, and calmer on Shopi­fy Plus. Con­trol you have no ca­pac­i­ty to act on is­n't an ad­van­tage, it's an open in­voice.

Ven­dor lock-in usu­al­ly gets pinned to data ex­port, the wrong place. At Shopi­fy the lock-in sits in the check­out and the op­er­at­ing mod­el. The check­out is a con­trolled black box: check­out.liq­uid is no longer sup­port­ed for the cen­tral steps, and free Liq­uid edit­ing and ar­bi­trary JavaScript have giv­en way to Check­out Ex­ten­si­bil­i­ty (UI ex­ten­sions, web pix­els, func­tions). Even on Plus you move in­side an ex­ten­sion sand­box, with no ac­cess to the check­out code it­self. For 95 per­cent of stores that's no prob­lem. For the 5 per­cent with a gen­uine spe­cial process it's a wall.

On top of that comes the di­men­sion Ger­man de­ci­sion-mak­ers have to de­fend be­fore a CFO and the board: data sov­er­eign­ty. Shopi­fy is a US ven­dor, with every­thing that third-coun­try trans­fer and the CLOUD Act raise as GDPR ques­tions. With self-host­ed Medusa, or­ders and per­son­al data sit in your own Post­greSQL in­stance, in an EU data cen­tre of your choice; pay­ment data stays with the pay­ment provider un­der its own data-pro­cess­ing agree­ment. This is not le­gal ad­vice but an ar­chi­tec­tur­al fact: where the data lives is your de­ci­sion with Medusa and not with Shopi­fy. The strate­gic overview of Medusa in the com­pos­able land­scape sits on our Medusa agency page.

Don't an­chor the com­par­i­son on the fea­ture set. An­chor it on three ques­tions you have to an­swer be­fore the board any­way. Do you have be­spoke B2B or pro­cure­ment process­es that break Shopi­fy's check­out and cat­a­logue lim­its? Is data sov­er­eign­ty a hard cri­te­ri­on in your mar­ket, reg­u­la­to­ry or po­lit­i­cal? And do you have, in-house or through a part­ner, the ca­pac­i­ty to car­ry op­er­a­tional re­spon­si­bil­i­ty and in­te­gra­tion work over the long run?

Three times no: take Shopi­fy Plus. The com­fort, the ecosys­tem, and the time-to-mar­ket are real, and the lock-in is a know­ing­ly bought, fair price for them. Com­plain­ing af­ter­ward about SaaS de­pen­den­cy would be dis­hon­est.

Three times yes: then Medusa is not the risky choice but the ex­ten­si­ble one. You pay in op­er­a­tional re­spon­si­bil­i­ty and build ef­fort, and you get falling in­stead of ris­ing costs, a portable data mod­el, and process­es that be­long to you. What you should nev­er do in ei­ther case is treat the de­ci­sion as a pure cost ques­tion. A tech­nol­o­gy de­ci­sion is al­ways strate­gic and po­lit­i­cal too. Com­pare only the base fee and you buy your ven­dor's in­ter­ests along with the prod­uct, and you no­tice only once the trans­ac­tion fees start grow­ing with your suc­cess.