.webp)
A UK founder we spoke to last month had a £110k marketplace MVP quote. Nine months and £180k later she had a working buyer app, an incomplete seller portal, no dispute resolution, and Stripe Connect Express doing more than she realised. Her next investment round wanted to see 30-day payout guarantees she could not enforce.
This is the marketplace build story most founders live through. The buyer app is what they see in the pitch deck. Payments, seller onboarding, and dispute resolution are what actually take the time.
This article is a candid 2026 guide to building a marketplace platform. Real cost bands per scope. The eight modules founders always under-budget. Sharetribe vs custom trade-offs. What Stripe Connect handles and what it does not. And the failure modes that turn a £150k MVP into a £400k rebuild.
The Marketplace Software Stack Nobody Fully Costs
Eight modules any working marketplace platform must cover.
Buyer discovery and search. Category browsing, search with filters, ranking algorithm. The visible module. Usually 25 to 35 percent of engineering effort.
Seller onboarding and KYC. Registration, identity verification, bank account setup, tax details, terms acceptance. Non-trivial in 2026 with EU and UK KYC requirements. 8 to 12 percent of effort.
Listing management. Seller creates and edits listings, sets pricing, uploads media, manages inventory. Sounds simple. Multiplies fast with categories. 10 to 15 percent of effort.
Payments and split payouts. Buyer pays, platform takes commission, seller receives payout. This is where Stripe Connect earns its price. 8 to 12 percent of effort.
Escrow and holding. Money held until service delivered or product received. Required for high-value or services marketplaces. 5 to 10 percent of effort.
Dispute resolution. Buyer complains, seller defends, platform decides. The hardest module to design well. 10 to 15 percent of effort.
Reviews and trust. Buyer reviews seller, seller reviews buyer, moderation of fake reviews. 5 to 8 percent of effort.
Admin and moderation. Platform team suspends bad actors, resolves disputes, handles reports. 8 to 12 percent of effort.
Founders often budget for modules 1 to 4 and discover 5 to 8 in month four. Budget for all eight from day one.
Real 2026 Cost Bands: MVP, Growth, Scale
Three scopes with real 2026 numbers.
MVP: £60k to £140k over 3 to 5 months. Single category, basic search with 3 to 5 filters, Stripe Connect Express, seller portal with essential features, basic dispute button that escalates to email, admin panel to suspend accounts.
Growth-stage: £180k to £400k over 6 to 10 months. Multi-category, advanced search with facets, Stripe Connect Standard or Custom, full seller portal with analytics, structured dispute workflow, review authenticity, admin moderation queue, KYC automation, tax reporting.
Scale: £500k to £1.5m+ over 12 to 18 months. Multi-country, complex payment splits, custom escrow, fraud detection with pattern matching, native mobile apps for both sides, machine learning for search and recommendations, in-house moderation team dashboards, integration with logistics or delivery partners.
Add £30k to £120k per year in ongoing platform run cost and maintenance from year two. Add £3k to £15k monthly in Stripe Connect fees at scale (2.9 percent + fixed fee per transaction plus platform fees). See Stripe Connect for marketplaces for the full pricing structure. See our related read on cost of mobile app development for the native mobile portion.
Sharetribe vs Custom Build: When Each Wins
Off-the-shelf marketplace platforms are viable in 2026, particularly for MVP validation.
Sharetribe (£95 to £450 monthly). Two products. Sharetribe Flex (developer-friendly, customisable) and Sharetribe Go (no-code, faster launch). Best for services marketplaces, community marketplaces, and MVPs where you need to validate the two-sided demand before committing to build. Caps out cleanly around 10,000 monthly active users; above that you migrate.
CS-Cart (£495 one-time to £8,850 lifetime). Best for product marketplaces with multi-vendor scenarios similar to Amazon. Feature-rich, PHP-based, requires PHP engineering to customise. Enterprise licence commonly needed at scale.
Yo!Kart, Arcadier, Marketplacer. Similar platforms with different positioning. Yo!Kart competitive for Middle East and Asian markets. Arcadier stronger for B2B. Marketplacer enterprise-tier.
Custom build. Defensible when your marketplace has unusual workflow (industry-specific escrow, complex commission structures, unusual integration with existing systems), when you plan to scale past 50,000 monthly active users within 18 months, or when the software itself is core to your competitive moat.
For most first-time marketplace founders, launch on Sharetribe or CS-Cart to validate demand, then migrate to custom around 5,000 to 10,000 MAU. Skipping this validation step often costs £150k to £300k in wasted early engineering.
What Stripe Connect Actually Solves
Six things Stripe Connect handles that would otherwise cost you 3 to 6 months of engineering.
KYC and seller verification. Identity checks against government databases across 50+ countries. Compliant with EU AML and US KYC rules.
Split payments. Take payment from buyer, split between platform commission and seller payout automatically.
Multi-currency and multi-country payouts. Sellers in 50+ countries paid in local currency automatically.
Chargeback and dispute handling on the payment side. Card network chargeback protocols managed by Stripe.
Tax reporting. 1099-K reporting in the US, VAT reporting in EU, sales tax obligations across jurisdictions.
Fraud prevention on the payment layer. Radar and machine learning to catch payment fraud before it hits your books.
What Stripe Connect does not solve.
Marketplace dispute mediation. Buyer says the product was not delivered; seller says it was. Stripe handles the card chargeback if buyer wins that route. Your platform handles the mediation before it gets there.
Listing quality moderation. Fake listings, prohibited items, misrepresented products. Your problem.
Review authenticity. Fake reviews, review-boosting rings, review-for-refund schemes. Your problem.
Cross-seller fraud rings. Coordinated bad actors across multiple accounts. Your fraud detection layer, not Stripe's.
Assume Stripe Connect handles about 40 to 50 percent of the payment and compliance stack. The rest is your engineering.
What We Learned Building TrackVid's Multi-Seller Architecture
TrackVid is our video proof and claim management platform used by ecommerce sellers primarily in India and expanding internationally. Not a two-sided marketplace, but the multi-seller architecture challenges are directly transferable. Two lessons.
Isolation must be architectural from day one. In TrackVid v1, we had a shared database with a seller_id column and permissioning layer. By month six, we recognised the audit and compliance risk. Rebuilt to schema-per-seller with a shared metadata layer. Six months of engineering we could have avoided by getting it right on day one. Marketplaces face the same choice: pretend "multi-tenant means WHERE clause" or architecturally isolate. Choose isolation.
Reconciliation determines whether you can invoice. In TrackVid, each seller has usage-based fees on top of subscription. If our reconciliation runs are wrong, we cannot invoice. Same problem shape as marketplace payment splits: if commission calculation drifts, sellers dispute payouts. Build reconciliation as first-class engineering with reproducible daily runs, not month-end batch scripts.
You can see TrackVid at our portfolio. If you want to talk about a marketplace build for your specific idea, get a marketplace platform estimate from WhiteStone.
Common Failure Modes
Three failure modes we see repeatedly.
Building the buyer app first, discovering seller onboarding at month four. Founders spend 60 percent of budget on the visible module and run out of runway before shipping the rest.
Committing to Stripe Connect Express when Standard or Custom is needed. Express is fast but limits customisation. Some marketplaces discover in month six that dispute customisation or payment split logic requires Standard or Custom. Migration is not free.
Skipping the moderation and admin tooling. Marketplace goes live. Bad actors appear. Ops team has no tools to suspend accounts, resolve disputes, or moderate listings. Trust collapses.
Frequently Asked Questions
How much does a two-sided marketplace cost to build in 2026?
MVP £60k to £140k over 3 to 5 months. Growth-stage £180k to £400k over 6 to 10 months. Scale £500k to £1.5m+ over 12 to 18 months. Add £30k to £120k per year ongoing plus Stripe Connect fees (2.9% + fixed fee per transaction plus platform-configurable fees).
Should I use Sharetribe or build custom?
Sharetribe (£95 to £450 monthly) for MVP validation up to about 10,000 monthly active users. Migrate to custom when you exceed that, need unusual workflow, or the software becomes core competitive advantage. Skipping Sharetribe validation often costs £150k to £300k in wasted early engineering.
What is the biggest cost founders miss?
Seller onboarding and KYC, dispute resolution, moderation tooling, and admin infrastructure typically total 30 to 40 percent of the build. Founders often budget for the buyer app (25 to 35 percent) and discover the rest in month four.
Does Stripe Connect solve marketplace payments completely?
It solves about 40 to 50 percent: KYC, split payments, multi-country payouts, chargebacks, tax reporting, payment fraud. It does not solve marketplace dispute mediation, listing quality moderation, review authenticity, or cross-seller fraud rings. Budget for these separately.
How long does a marketplace MVP take?
3 to 5 months for a genuine MVP (single category, basic search, Stripe Connect Express, essential seller portal, basic dispute button). Faster than 3 months usually means shipping without seller portal or dispute handling, which fails quickly at first traction.
The One Thing to Remember
The marketplace build cost founders quote in the pitch deck is the buyer app. The real cost is the buyer app plus seller onboarding plus payments plus dispute resolution plus moderation plus admin. That is 2.5 to 3x the buyer-app figure. Budget for all eight modules from day one, or expect month four to be painful.
If you want a candid conversation about your specific marketplace idea, browse our custom software development services or come to the estimate call.
