How to build a marketplace MVP without a co-founder

You build a marketplace MVP solo by killing the chicken-and-egg problem on day one: pick one side, fake the other side manually, and ship a single-sided product that looks like a marketplace to your first 50 users. You handle supply yourself for 60 to 90 days, use Stripe Connect for payments, and book a weekly engineer to write the code you can't. No co-founder required.

Marketplaces are the hardest startup shape to launch. They're also the most googled, because every solo founder eventually asks the same question: how do I get buyers without sellers and sellers without buyers, when I have neither? The answer that actually works isn't a feature, a growth hack, or a co-founder. It's a sequence of small unglamorous moves, most of which look nothing like "running a marketplace" for the first three months.

This post walks you through that sequence. Real stack, real budget, a 90-day plan, and the trick almost every successful marketplace founder used but rarely advertises.

The setup: where solo marketplace founders get stuck

You have an idea. Maybe it's a marketplace for pet sitters in your city, or freelance video editors for YouTubers, or used industrial equipment for small machine shops. You've talked to 20 buyers. They want it. You've talked to 10 suppliers. They'd list on it.

Then you sit down to build, and reality bites:

1. The product needs both sides to work, or neither side comes back.
2. You can't pay yourself to spend six months coding before launch.
3. Every "marketplace in a box" template (Sharetribe, CS-Cart, Marketplacer) costs $200 to $1,000 per month and still needs significant engineering to fit your category.
4. Stripe Connect, KYC, dispute handling, review systems, and search ranking each look like their own startup.
5. You can't afford a technical co-founder, and the technical advisor who promised to help has gone quiet.

Founders get stuck here for a year. The ones who ship in 90 days do something specific: they stop trying to build a real marketplace and instead build a believable illusion of one, with themselves wearing the costume on the supply side.

The single-sided MVP trick

The most useful concept in marketplace design is the "single-sided MVP," or supply-side concierge. The buyer side is real software, real payments. The supply side is you, in a spreadsheet, with a phone.

Airbnb did this. Brian Chesky and Joe Gebbia walked door to door in New York photographing apartments because hosts wouldn't upload good photos. That's the Paul Graham "do things that don't scale" essay made flesh. DoorDash founders drove the early deliveries themselves. Etsy's first sellers were personally onboarded. Thumbtack's first pros were cold-called.

The pattern is identical every time: build the buyer-facing surface as real product, source supply manually for the first 50 to 500 transactions, and replace the manual layer with software only after you've proven liquidity.

This works because buyers don't care how the sausage is made. If you can fulfill demand by texting a sitter you found on Craigslist and Venmoing them, the buyer is happy and you've validated demand without writing a matching algorithm.

The 4 marketplace MVP shapes (comparison)

Before you start, pick a shape. Each one trades off speed against eventual scalability.

Shape	What you build	Solo-friendly	Time to first txn	Becomes the real product?
Single-sided MVP (concierge)	Buyer site + you on the supply side	High	2 to 4 weeks	Yes, with rewrites
Off-the-shelf (Sharetribe, Yelo)	Configured template	Medium	4 to 8 weeks	Partially, capped at template limits
Custom from scratch	Full two-sided product day one	Low	4 to 6 months	Yes
Aggregator wedge	Scrape or list public supply (Zillow-style)	High	2 to 4 weeks	Yes, until rights holders sue

Most solo founders should run the single-sided MVP. It's the fastest path to a real transaction, the cheapest to test, and it teaches you the one thing software can't: what the actual supply problem looks like. The off-the-shelf path is tempting but lulls founders into shipping a generic marketplace that ranks for nothing and converts at 0.4%.

The aggregator wedge works in niches with public data (apartments, used cars, restaurant menus), but legal exposure is real. Check Terms of Service before you scrape.

What you actually need on day one

A marketplace MVP needs five working surfaces. Anything else is procrastination.

1. A landing page with two clear paths: "I want to hire" and "I want to be hired" (or buy/sell, book/list, etc.).
2. A request form that lets a buyer describe what they want and pay or commit.
3. A way to fulfill the request (this is you, with a phone, for the first 90 days).
4. A payment rail that takes money from the buyer, holds it, and pays the supplier after delivery. This is Stripe Connect's exact job.
5. A review or rating loop that produces social proof and pulls supply back for repeat work.

Notice what's missing. No search. No filters. No supplier dashboard. No real-time chat. No mobile app. No matching algorithm. Those are second-year problems. If you build them first, you'll run out of money before liquidity exists.

The stack (cost per month under $100)

You can run a real two-sided marketplace on a stack that costs $0 to $80 per month for the first year. Pretending you need more is how solo founders burn $40,000 before their first transaction.

Layer	Tool	Cost
Frontend + hosting	Next.js on Vercel	$0 (Hobby) to $20/mo (Pro)
Database + auth	Supabase	$0 to $25/mo
Payments	Stripe Connect (Standard)	2.9% + 30¢ per txn, $2/mo per active connected account
Email	Resend or Loops	$0 to $20/mo
Search (when you need it)	Algolia free tier or Postgres tsvector	$0
Analytics	PostHog Cloud	$0 (1M events)
Forms / spreadsheet ops	Airtable or a Postgres table	$0 to $20/mo

Total monthly burn before traffic: about $20 to $80. Total payments cost only kicks in when you have revenue. Use Stripe Connect Standard (not Custom or Express) for the MVP. Standard pushes KYC and tax handling onto Stripe-hosted onboarding, which removes a week of compliance work.

If you're a non-technical founder, this stack is also the cheapest because it's what every Cadence weekly engineer already ships in their sleep. Hand them the Stripe Connect docs and they'll have a working escrow flow by Wednesday.

Payment infrastructure: the only thing you can't fake

You can fake supply, search, and matching. You can't fake payments. The moment money moves between two strangers through your product, you owe both sides escrow, refunds, dispute handling, and (in the US) 1099-K reporting.

Stripe Connect handles 90% of this. The Standard flavor takes about a day to integrate. You'll write three things:

1. A "create connected account" flow that redirects suppliers to Stripe's onboarding.
2. A checkout that creates a PaymentIntent with transfer_data[destination] and application_fee_amount set to your take rate.
3. A webhook handler for payment_intent.succeeded, account.updated, and charge.dispute.created.

That's the whole MVP payment system. Don't build escrow yourself. The take-rate cost (about 0.25% to 0.5% on top of card fees for Connect Standard) is the cheapest engineering you'll ever buy.

Trust and reviews: do the boring social-proof work

A marketplace dies if buyers don't trust suppliers. You bootstrap trust with the things you can control:

• Verify supplier identity yourself. ID + a 15-minute video call. You're doing the first 50 anyway.
• Offer a no-quibble money-back guarantee. You eat the refund cost. It's cheaper than churn.
• Publish a public "first 10 transactions" log. Names (with consent), what they bought, what went wrong.
• Launch reviews after txn 20, not txn 1. Empty rating sections look worse than no review system.

Avoid star ratings as your only trust mechanism. Free-form testimonials with the buyer's name and city convert 3 to 5 times better than 4.7-star aggregates.

A 90-day plan you can run solo

Here is the plan that gets you from idea to live marketplace with revenue, no co-founder, and a budget of roughly $5,000 to $15,000 in tooling and engineering.

Days 1 to 14: pick the wedge

You should already have 20 buyer interviews and 10 supplier interviews in the bank. If not, run them now. The wedge is the single category, single geography, and single transaction type you'll launch with. Not "marketplace for freelancers": "freelance YouTube thumbnail designers for channels with 10k to 100k subs, priced per thumbnail." Narrow wins.

While the interviews run, register the domain, set up the Vercel + Supabase + Stripe accounts, and write the landing page yourself. This is not engineering work. It's product clarity.

Days 15 to 28: build the buyer-side product

Scope is small: landing page, two-path signup, buyer request form, Stripe Connect onboarding for the (fake, you-controlled) supplier account, payment intent on submission, transactional emails.

A mid-level engineer on Cadence ($1,000/week) ships this in two to three weeks. The 48-hour free trial covers scoping and the first commit. Daily ratings drive auto-replacement so you don't have to fire anyone yourself. For comparison, the same scope through Toptal typically runs $80 to $120 per hour with a 20-hour minimum, often $4,000+ for the same MVP.

This phase pairs well with a clear engineering job spec you can hand over without three follow-up calls.

Days 29 to 60: seed supply manually

This is the part everyone skips. You personally find the first 25 suppliers. Cold DM, Reddit, Discord, IndieHackers, in-person events, Craigslist, Upwork, niche Slacks. Pay above-market for the first 20 transactions if you have to.

Your goal is not profit. It's 25 to 50 completed transactions that prove buyers will come back. Track every transaction in a spreadsheet. Note what broke, what the supplier complained about, what the buyer complained about.

The rhythm is similar to working with a remote developer in a different timezone: short async updates, fixed handoff windows, written briefs.

Days 61 to 90: replace yourself with software

Now you build the second wave, informed by what actually broke: a supplier dashboard, basic search or filter, automatic payout schedule, review prompt.

You may need a senior engineer here ($1,500/week on Cadence) if you're rebuilding matching or adding real-time chat. Avoid the lead tier ($2,000/week) until you have product-market fit; that's fractional CTO money.

By day 90 you should have 50 to 200 transactions, real revenue, real reviews, and a buyer-side product that no longer requires you to personally text every supplier. That's a marketplace MVP.

Common solo-founder mistakes

The five mistakes I see most often, in order of money burned:

1. Building both sides at once. You'll ship neither well. Pick the harder side to fake (usually supply) and build the other.
2. Hiring a senior or lead to write CRUD. A mid ships the MVP. Save the senior for the day you have real load.
3. Building escrow from scratch. Use Stripe Connect. Always.
4. Launching reviews on day one. Empty rating sections kill conversion. Launch reviews at transaction 20.
5. Skipping the manual supply phase. If you can't fulfill 25 transactions by hand, you don't understand the supply side well enough to automate it.

The sixth honorable mention: hiring a freelance designer for a custom Figma system before you have transactions. Use shadcn/ui or Tailwind UI ($249 lifetime). A real designer is the right call at month six, not month one.

What to do next

If you have an idea and 90 days, start the wedge interview process this week. If you have the wedge and need to ship, the cheapest path is to write the brief yourself and book a mid-level Cadence engineer to handle the build while you handle supply. The 48-hour trial means you spend nothing if the fit is wrong. Most founders who run this play hit their first paid transaction inside week three.

If you want a structured version of this same arc applied to non-marketplace startups, the 90-day idea-to-revenue playbook covers the sprint cadence in more detail.

Ready to ship the buyer-side build? Book a vetted weekly engineer on Cadence in 2 minutes. Every engineer is AI-native by default, vetted on Cursor / Claude / Copilot fluency before they unlock bookings, with a 48-hour free trial so you only pay if the fit is right. Start your booking.

FAQ

Do I really need a technical co-founder to build a marketplace?

No. A co-founder gives you long-term commitment and equity alignment, which matter at year three. They don't ship faster than a booked weekly engineer. For an MVP, a $1,000-per-week mid-level engineer plus a clear spec will outpace most cofounder searches, which typically take 4 to 9 months and end without a match.

How much does a marketplace MVP cost to build?

Roughly $3,000 to $12,000 in engineering plus $20 to $80 per month in tooling for the first 90 days. The range depends on category complexity (a video-editing marketplace needs file handling and review threading; a pet-sitter marketplace doesn't) and whether you can write any code yourself.

What's the best marketplace platform for non-technical founders?

Sharetribe Flex and Yelo are the most-cited turnkey options, starting around $99 to $200 per month plus transaction fees. They get you a generic two-sided site in 1 to 2 weeks. The trade-off is template lock-in: when you want a category-specific flow (custom request forms, niche pricing models, vertical search), you'll outgrow the template and have to rewrite anyway. For most serious founders, the cleaner path is Next.js + Supabase + Stripe Connect from day one.

How do I get the first 100 suppliers without paid ads?

Cold outreach (DM, email, in-person), niche Reddit and Discord communities, Upwork and Fiverr sellers in your category, founder Slacks, and personal network. Pay the first 20 above-market to remove the risk. Expect 5% to 15% reply rates. The single highest-converting channel for most marketplaces in year one is in-person events for the relevant trade.

When should I move from a single-sided MVP to a real two-sided product?

After roughly 50 completed transactions with a buyer return rate above 25% and supplier complaints concentrated on the same 2 or 3 ops problems you've been doing manually. That's the signal that the friction is now in software, not in the market itself. Before that point, scaling code investment scales the wrong thing.