⚡The honest answer up front
E-commerce platform cost is a layering problem. The subscription is the visible layer; underneath it sit apps, transaction fees, theme and customization work, and the developer time to keep it all coherent. Quoting only the subscription is how stores budget $50 a month and spend $2,000.
As labelled US market ranges we observe: a properly run Shopify or BigCommerce store costs $500 to $5,000 a month all-in, scaling with app stack and order volume. A custom build — Next.js storefront, a commerce engine or fully bespoke backend, payment and tax integration, admin tooling — runs $40,000 to $200,000 depending on complexity, then $8,000 to $40,000 a year in hosting and maintenance. Headless sits between: platform backend, custom storefront, and a build cost of $30,000 to $100,000 on top of the platform bill.
The decision is not "cheap versus expensive" — it is rent versus own, and the right answer moves with your revenue, your catalogue complexity and how much of your differentiation lives in the buying experience. The sections below price each layer and then put the 3-year totals side by side.
E-commerce systems development
| Path | Up-front cost | Ongoing monthly | 3-year total (typical mid-size store) |
|---|---|---|---|
| Shopify / BigCommerce, standard | $5,000–$25,000 (setup, theme, migration) | $500–$3,000 all-in | $25,000–$130,000 |
| Headless (platform backend + custom storefront) | $30,000–$100,000 | $1,000–$5,000 | $65,000–$280,000 |
| Fully custom build | $40,000–$200,000 | $700–$3,500 (hosting + maintenance) | $65,000–$325,000 |
| Marketplace-first (Amazon/Etsy), no DTC store | $0–$5,000 | 15–40% of revenue in fees & ads | Scales with revenue, not with cost |
📊The Shopify app stack: where $39 becomes $2,000
The platform pricing page lists the subscription. Your store's actual monthly cost is the subscription plus the app stack, and the app stack is where the number lives. The pattern is remarkably consistent: reviews, subscriptions, loyalty, search and merchandising, email capture, bundles, returns, advanced shipping — each is a gap between what the platform does and what your store needs, and each is filled by an app at $10 to $300 a month.
A typical growing store accumulates 8 to 15 apps. At a blended $30 to $150 each, that is $500 to $2,000 a month in apps alone, before the subscription tier upgrade your volume eventually forces, and before transaction fees if you are not using the platform's own payment processor — a surcharge of roughly 0.5 to 2 percent per order depending on plan, which at $100,000 a month in revenue is $500 to $2,000 more.
Two properties make this cost insidious. It creeps — each app is justified individually and the total is never reviewed. And it compounds with revenue in the transaction-fee layer, so the platform gets more expensive exactly when you succeed. None of this is an argument against Shopify; it is an argument for pricing the real stack, which the table below does honestly.
| Monthly cost layer | Lean store | Growing store | At scale |
|---|---|---|---|
| Platform subscription | $39–$105 | $399 (Shopify) / similar tier | $2,300+ (Plus) or negotiated |
| App stack (8–15 apps) | $150–$500 | $500–$2,000 | $1,500–$5,000+ |
| Transaction-fee surcharge (if not using platform payments) | ~2% of revenue | ~1% of revenue | 0.5–1% of revenue |
| Developer / agency retainer | $0–$500 | $500–$2,000 | $2,000–$10,000 |
| Realistic total (excl. % fees) | $200–$1,100 | $1,400–$4,400 | $5,800–$17,300+ |
Run the "app audit" once a year: list every app, its monthly cost, and what breaks if you remove it. Stores routinely find $300 to $800 a month in apps nobody would notice losing — and the ones that remain are the true monthly platform cost you should compare against a custom build.
🏗️When a custom build actually wins
Custom e-commerce wins in three situations, and only three. First, when the buying experience is the product — configurators, custom pricing logic, B2B quote workflows, multi-party marketplaces, subscriptions with unusual logic. If the thing that makes customers buy from you cannot be expressed in a platform's data model, you will pay for the workaround forever, and the workaround is usually uglier and slower than the build would have been.
Second, when transaction-fee and app economics tip over. Past roughly $1M to $2M in annual revenue, the percentage-of-revenue fees plus a heavy app stack plus a developer retainer frequently total $60,000 to $150,000 a year — which is the maintenance cost of a custom system with money left over. The crossover moves with your mix, so do the arithmetic on your own numbers rather than trusting the threshold.
Third, when you need to own the data and the roadmap outright — deep ERP integration, custom fulfilment logic, or a storefront that is also your SaaS product. The honest counterweight: a custom build means you own uptime, security, payment-compliance posture and every future feature. Teams that build custom to save money without budgeting the ownership cost end up with an unmaintained system and a replatforming project — the most expensive outcome of all.
| Signal | Stay on platform | Build custom |
|---|---|---|
| Annual revenue | Under ~$1M–$2M | Above ~$2M with heavy fee/app spend |
| Catalogue & pricing | Standard products, standard discounts | Configurators, contract pricing, complex bundles |
| Checkout requirements | Standard checkout converts fine | Checkout logic is a differentiator (B2B, quotes, approvals) |
| Integrations | Apps cover ERP/3PL/PIM needs | Deep two-way ERP or custom fulfilment logic |
| Team | No engineering capacity to own a build | Engineering capacity (in-house or retained) exists |
| Differentiation lives in | Product, brand, marketing | The buying experience itself |
🔌Headless: the middle path, priced honestly
Headless commerce — keeping Shopify, BigCommerce or a commerce engine as the backend while building a custom storefront, usually in Next.js — is pitched as the best of both worlds. Sometimes it is. It keeps the platform's checkout, payments, tax and order management (the parts that are genuinely painful to own) while giving you full control of the presentation layer, which is where speed, SEO and brand differentiation live.
The cost shape: $30,000 to $100,000 for the custom storefront build, on top of the platform subscription and whatever app stack you keep. You also take on a permanent engineering dependency — a headless storefront is software you own, with hosting, deployments and a component library that someone must maintain. Budget $500 to $2,000 a month for that ownership, in-house or retained.
The failure mode to avoid is headless-by-fashion. If your reason is "the theme editor feels limiting" but your storefront is a standard catalogue with a standard cart, a premium theme and a few hours of a good developer solve it for a twentieth of the price. Headless earns its cost when the storefront itself is a product surface — rich interactive merchandising, content-commerce blends, or performance requirements that theme architecture cannot meet. Otherwise it is an expensive answer to a cheap problem.
🛒Marketplace vs DTC: the cost structure nobody models
The cheapest e-commerce launch is no store at all: sell on Amazon, Etsy or a category marketplace. Zero build cost, instant traffic, and a fee structure that takes roughly 15 to 40 percent of revenue once referral fees, fulfilment and the advertising you will inevitably need are counted. The economics are honest as long as you model them: you are paying a revenue share instead of a fixed cost, which is exactly right when volume is unproven.
The structural cost of marketplaces is not the fee — it is that you rent the customer. You get no email list, no retargeting pixel, no brand search equity, and your margin is one policy change away from compression. Marketplaces are a channel; the mistake is letting them be the business.
The pattern that works for most product companies: marketplace-first to validate demand and generate cash, then a DTC store (platform, not custom) once repeat purchase and brand search exist, then — and only then — consider custom when the DTC channel's fee and app spend justifies ownership. Each stage funds the next, and each step skips a stage at its peril. A custom build for an unvalidated product is the most expensive way to learn that nobody wanted it.
⚖️The 3-year TCO comparison, side by side
Here is the comparison done properly, as a composite illustration — labelled as such, not a quote. A store doing $80,000 a month in revenue (~$1M a year), 300 SKUs, subscriptions plus one-time sales, migrating off a legacy platform.
Shopify path: $399 a month subscription, an honest app stack of $900 a month, a developer retainer of $1,000 a month, setup and migration at $15,000, and the transaction-fee surcharge if not using platform payments (up to ~$800 a month at this volume on the relevant plan). Three-year total: roughly $95,000 to $130,000. Headless path: the same backend costs plus a $60,000 storefront build and $1,000 a month to own it — three-year total roughly $160,000 to $210,000.
Custom path: a $120,000 build covering catalogue, subscriptions, checkout, ERP sync and admin, then $18,000 a year in hosting and maintenance. Three-year total: roughly $175,000. The reading: at $1M a year in revenue, the platform is still the value pick and it is not close. Re-run the same table at $5M a year with a 1 percent transaction-fee layer and a $3,000-a-month app-plus-dev stack, and custom pulls ahead by a margin that widens every year. The crossover is arithmetic; the mistake is never doing it.
| 3-year scenario | Platform | Headless | Custom build |
|---|---|---|---|
| ~$1M/yr revenue, standard catalogue | $95K–$130K — cheapest | $160K–$210K | ~$175K |
| ~$5M/yr, heavy app stack, % fees | $250K–$400K (fees dominate) | $220K–$350K | $150K–$300K — pulls ahead |
| Complex B2B / configurator product | Workarounds accumulate; cost + friction | Storefront solves UX, not backend logic | Usually the honest fit |
Compare platforms on the 3-year total at your projected year-3 revenue, not the month-1 subscription. Replatforming is a $20,000 to $100,000 project — the store you choose today should be the store your year-3 self still fits in.
📦If you do switch: what replatforming costs
Replatforming is the second e-commerce project most stores eventually run, and it prices by data and by URL. The data side — products, customers, order history, subscriptions, reviews — runs $5,000 to $30,000 depending on volume and cleanliness, with subscription data being the delicate part because active billing relationships have to migrate without charging anyone twice or losing them silently.
The URL side is the one that gets forgotten until the traffic report craters. A platform migration that changes URL structure without a complete redirect map hands Google a brand-new site, and organic revenue takes months to recover. A proper migration includes a full redirect map, metadata carryover and a launch-window monitoring plan — typically $3,000 to $10,000 of the budget and consistently the highest-ROI line in it.
Timing matters too: never replatform in the quarter of your peak season, and run the old and new stores in overlap for one full order cycle. The overlap feels like paying for two stores; it is actually the insurance premium on the revenue stream that pays for everything else.
E-commerce systems: builds, headless and replatformingHow to build a marketplace app
🧭How to make the call
The decision procedure is mechanical. First, price your real platform stack: the subscription tier your volume forces, the apps you actually need (audit them, do not guess), the transaction-fee layer on your payment setup, and honest developer time. Second, price custom: the build scoped to your real feature list, plus hosting and maintenance at 15 to 25 percent of build per year. Third, compare the 3-year totals at your projected year-3 revenue — and check which column your differentiation actually lives in.
Then apply the tiebreakers. No engineering capacity to own a build stays on platform regardless of the arithmetic, because an unowned custom system is a depreciating asset. A buying experience that is itself the differentiator pushes toward custom even before the numbers cross. And an unvalidated product goes marketplace-first or platform-first, full stop — the cheapest store is the one you do not build until someone has proven they will buy from it.
Get a scoped number for your e-commerce build
Platform when revenue and complexity are modest
Under ~$1M–$2M a year with a standard catalogue, the platform stack is cheaper, faster and maintained by someone else. Spend the difference on inventory and marketing.
Headless when the storefront is the surface
Content-commerce blends, interactive merchandising or hard performance requirements justify the custom front end while the platform keeps checkout, tax and orders.
Custom when the fees and workarounds compound
When app stack plus transaction fees plus retainer exceed the maintenance cost of ownership — or when your pricing and checkout logic simply do not fit the platform data model.
Marketplace-first when demand is unproven
Pay the revenue share while you validate. Build the store when repeat purchase exists; build custom when the store itself strains the platform.