An iceberg shown in three cross-sections, each with a small tip above water and a different-sized mass below, with small lights in the submerged parts.

Every team weighing a multi-vendor marketplace asks "should we build it or buy it?" and prices the wrong thing. The build cost and the licence fee are the visible tip; the mass underwater is integration and, above all, operations — the bill that arrives every month for as long as the marketplace exists. This article costs all three routes honestly in effort terms: building the engine, integrating licensed software, and having it operated for you. If you want the conceptual framing first, the pillar on white-label marketplaces and the Marketplace-as-a-Service explainer set it up; this is the money-and-effort version.

The three routes

There are exactly three ways to end up with a marketplace, and they trade the same three resources — money, time and engineering capacity — in different proportions.

  • Build from scratch. Write the ingestion, enrichment, search, cart, split checkout, vendor consoles and monitoring yourself. Maximum control, maximum cost, longest time to live.
  • Buy (license) software. Install a marketplace platform under your brand. You skip writing the core engine but own integration, hosting, tuning and operations.
  • Managed (Marketplace-as-a-Service). An operator runs everything on your domain. You bring the audience and brand; they bring the machinery and share revenue.

The trap is comparing them on the wrong axis — sticker price — when the axis that decides the outcome is total cost of ownership over several years, most of which is operations.

The iceberg: what each route actually costs

Break the true cost into three layers and the routes separate clearly.

Cost layerBuildBuy (license)Managed
Engine / platformMulti-quarter buildLicence feeIncluded
IntegrationN/A (you built it)Heavy — connect vendors, payments, taxIncluded
Hosting & infraYours foreverYours foreverIncluded
Operations (perpetual)Yours foreverYours foreverIncluded
Time to liveQuartersMonthsWeeks
ControlTotalHighModerate
You need engineers?A teamAt least someNone

The row that dominates the decision is operations (perpetual), and it is the one spreadsheets omit because it has no line item. Build and buy both leave it entirely on you; managed absorbs it. That single row is usually larger than the engine and integration combined, because it never ends.

What building really involves

A credible multi-vendor platform is not one project but a dozen. Catalog ingestion from several ecommerce platforms and feeds. Product data enrichment to make messy vendor data searchable. A real search engine with typo tolerance, synonyms, facets and semantics. One cart with per-vendor shipping math. Split checkout that fans one payment into VAT-correct sub-orders. Vendor and operator consoles. Compliance — EU price rules, VAT, consent. Monitoring, deploys, alerting. Each is a specialty; together they are a multi-quarter effort for an experienced team, and the work does not stop at launch because catalogs drift, vendors switch platforms, and payment and tax rules change under you. Building is justifiable when the marketplace is the company. It is rarely justifiable when it is a monetisation channel for an existing brand.

There is a second, quieter cost to building: opportunity. Every engineer-quarter spent on ingestion and checkout plumbing is a quarter not spent on the content, product or community that earns your audience in the first place — the actual source of your advantage. For a media or content brand, redirecting scarce engineering into commerce infrastructure that dozens of teams have already built is rarely the highest-value use of the team, even when they are perfectly capable of building it. Capability is not the same as it being worth their time.

The hidden cost of buying software

Licensing looks like a shortcut because it skips the engine, but it relocates the cost rather than removing it. You still connect every vendor, map their categories, wire payments and tax, tune search for your catalog, host and secure the stack, and — forever — keep it healthy. The licence fee is the small number; the integration and the standing operations team are the large ones. Licensed software is the right call when you have engineers to spare and want deep low-level customisation. It is the wrong call if you chose "buy" specifically to avoid needing an engineering team, because you still do.

From production

The operations layer is not theoretical. Running our marketplace means catalog syncs across two ecommerce platforms plus feeds, enrichment re-runs when data changes, search tuning, per-vendor shipping mirrored from each seller's own shop, split checkout with VAT-correct invoices, plus monitoring, a daily SEO audit and a weekly health sweep. None of that ends after launch — it is the perpetual cost the build-or-buy spreadsheet always leaves out.

A decision framework by team and ambition

Match the route to the resource you are actually short of.

Your situationBest route
Marketplace is your core product; funded engineering teamBuild
Have engineers; need deep customization; commerce is strategicBuy / license
Have an audience; no commerce engineering team; want to be live fastManaged
Content or community brand monetising existing trafficManaged
Uncertain demand; want to validate before committing capitalManaged (lowest sunk cost)

The through-line: if commerce is your product, own the engine; if commerce is a way to monetise an audience you built doing something else, do not spend your scarce engineering and attention rebuilding plumbing that already exists. The revenue mechanics that make the managed share worthwhile are in the business model guide, and whether the model fits your niche at all comes back to the three-ingredient test in niche marketplace ideas.

One more factor belongs in the decision: uncertainty. If you are not yet sure the marketplace will earn its keep, the route with the lowest sunk cost is the one that lets you find out cheaply. Building and licensing both demand large commitments before you learn whether demand converts; a managed arrangement lets you validate with real vendors and real orders first, and scale the bet only if the numbers work. Costing the downside of being wrong — not just the upside of being right — is what separates a sound infrastructure decision from an expensive act of faith.

Key takeaways

  • Total cost of ownership, not sticker price, decides it — the engine and licence are the visible tip of a much larger iceberg.
  • Operations is the dominant, perpetual cost and the one spreadsheets omit; build and buy leave it on you, managed absorbs it.
  • Building is a dozen specialties, justified only when the marketplace is the company itself.
  • Licensing relocates cost rather than removing it — you still integrate, host and operate, and still need engineers.
  • Match the route to your scarcest resource: own the engine if commerce is your product; go managed if it is a channel for an audience you already have.

Frequently asked questions

Should I build or buy a marketplace platform?
Build if the marketplace is your core product and you have a funded engineering team; buy or license if you have engineers and need deep customization. But if commerce is a way to monetise an audience you built doing something else, a managed marketplace is usually the right answer, because the dominant cost is not the engine or the licence but the perpetual operations, which building and buying both leave entirely on you.
How much does it cost to build a marketplace?
In effort terms, a credible multi-vendor platform is a multi-quarter build for an experienced team, because it is really a dozen specialties — ingestion, enrichment, search, cart, split checkout, vendor consoles, compliance and monitoring. And the cost does not stop at launch: catalogs drift, vendors switch platforms, and payment and tax rules change, so the operations bill arrives every month thereafter. That ongoing cost usually exceeds the initial build.
Is licensed marketplace software cheaper than managed?
The licence fee is cheaper than a managed revenue share on paper, but licensing relocates cost rather than removing it. You still integrate vendors, wire payments and tax, tune search, host the stack and keep it healthy forever, which means you still need an engineering and operations team. If you chose licensing specifically to avoid needing that team, managed is usually cheaper once you count the standing operations cost.
What is the real total cost of ownership of a marketplace?
Three layers: the engine or platform, integration, and perpetual operations — plus hosting. The mistake is pricing only the first, when operations dominates and never ends. Building leaves all of it on you; licensing removes the engine but keeps integration and operations; a managed service absorbs the lot in exchange for a revenue share. Costing the operations row honestly is what changes most build-vs-buy decisions.

Skip the iceberg. Keep the marketplace.

We carry the engine, the integration and the perpetual operations, so your team never has to budget for them — you keep the audience and the revenue share.

Request early access See the live marketplace →