Product-Market Fit in a Two-Sided Marketplace

TL;DR: Product-market fit in a two-sided marketplace is liquidity — buyers reliably finding supply, sellers reliably finding demand, with unit economics that survive without subsidies. Based on 200+ founder interviews on the PMF Show, winners solve both sides simultaneously in one narrow wedge: Mercado Libre compounded ~80% annual growth this way; Unhaggle grew three years without liquidity and pivoted.

After interviewing 200+ founders on the PMF Show, I've noticed that founders chasing product-market fit in a two-sided marketplace face a problem SaaS founders never do: they're building two products for two customers whose value depends entirely on each other. Demand means nothing without supply density; supply churns without demand velocity. The founders below each answered a different piece of the same question: how do you get both sides moving at once, and how do you know when the flywheel is actually spinning versus when you're just paying it to spin?

What does product-market fit look like in a two-sided marketplace?

It looks like a wheel that rolls on its own. According to Hernan Kazah, co-founder of Mercado Libre, the entire thesis of Latin America's largest marketplace was the simultaneous loop between the two sides — and the recognition that someone has to push the wheel first.

"As you could build an active, liquid marketplace with lots of buyers, then those buyers would attract sellers, and more sellers would put on sale more products, and more products would attract more buyers. So the wheel would start to roll, but you needed the initial traction." — Hernan Kazah, Mercado Libre

The counterintuitive part, as Kazah shared on the PMF Show, is that two-sided PMF rarely produces a hockey stick. Mercado Libre launched in 1999 and reached $78 million in revenue by its 2007 IPO — eight years in, with eBay (then a near-20% shareholder) openly comparing growth rates.

"When we were in year three, we're growing at 300%, and you guys are growing at 80%. And when we were at year four, we're growing at 250% and you guys are growing at 80%." — Hernan Kazah, Mercado Libre

The punchline: eBay's growth decayed toward 20-25% while Mercado Libre kept compounding. Because both sides of the network kept feeding each other, the company was still growing 70-80% year over year 25 years after founding. As Kazah put it, nobody keeps growing at almost triple digits 25 years later — unless the flywheel itself is the product.

Key stat: Mercado Libre took 8 years to reach $78M in revenue, then compounded ~80% annual growth for over two decades — into a $100B+ public company.

Should you solve supply or demand first — or both at once?

Both, but they demand different kinds of effort — and knowing which side needs manual labor is half the battle. Chris Slowe was Reddit's first employee and later its CTO, and he also helped build Hipmunk, a travel marketplace. The two companies sat at opposite ends of the supply problem, as he shared on the PMF Show.

At Reddit, supply was self-generating: users created the content that attracted more users. The site hit about 1 million monthly uniques in mid-to-late 2006 and, according to Slowe, was "on a pretty good three to six month doubling curve" with essentially zero paid acquisition on its way to 10 million monthly uniques by 2010.

Hipmunk's supply side — airlines and hotels — would never show up on its own. That forced a completely different org chart from day one.

"We had to hire a BD lead extremely early. That was because that was entirely to make deals with airlines and hotels and larger companies... The earliest part of the team, the first 12 employees give or take, included at least two people working on BD and at least one person doing marketing." — Chris Slowe, Reddit

The lesson for two-sided founders: audit which side of your marketplace generates itself and which side must be manufactured deal by deal. Reddit's demand-side monetization proof came fast when it arrived — Reddit Gold pulled in roughly $100,000 in its first week in 2010, more than the prior couple months of ad sales combined.

Key stat: At Hipmunk, 2 of the first 12 employees worked purely on supply-side BD; Reddit reached 1M monthly users on a 3-6 month doubling curve with no equivalent hire.

Can you fake marketplace liquidity with subsidies?

Yes — and that's exactly why signups are a false PMF signal in any transaction business. Jon Robin, co-founder of Dabble, the Australian social betting platform, is blunt about how easy it is to buy the appearance of liquidity in a business where money flows through the platform.

"This industry probably more than most can be really propped up by pouring money into marketing and pouring money into bonuses and generosity. 'Cause you're dealing with people's money. If you're giving them money, it's easy to prop up their revenue." — Jon Robin, Dabble

That's the two-sided marketplace trap in miniature: subsidize both sides hard enough and any marketplace looks liquid. Dabble launched in late July 2021 after 15 months of buildup, and by the end of that year roughly 50,000 people had registered — but only 20,000-30,000 actually deposited and bet. According to Robin, the real test came later.

"True product market fit can only really come in this space when the unit economics stack up and... the lifetime value of the customer look like they're gonna be higher than the customer acquisition cost." — Jon Robin, Dabble

It took Dabble 12 months of retention data to prove customers were sticking around without being paid to. Once that held, the compounding was dramatic: about $10 million in net revenue in the first full financial year (on roughly $100 million of bets taken), then $30 million, then $75 million, then $150 million in year four.

Key stat: Dabble waited 12 months for LTV > CAC proof, then grew net revenue from $10M to $150M in four years.

Can you build a marketplace on someone else's supply?

Sometimes the smartest answer to the two-sided problem is to only solve one side. Hussein Fazal, co-founder and CEO of Super.com, started with SnapTravel — a hotel-booking business that borrowed existing supply (hotel inventory via wholesale channels) and manufactured only demand, through messaging-first booking.

Because supply already existed at scale, every dollar went into demand acquisition and conversion, and the model scaled fast. As Fazal shared on the PMF Show, the pre-COVID business was substantial:

"We're doing at that time at least over a hundred million in GMV in top line, call it a 10 to 15% margin." — Hussein Fazal, Super.com

The funding path tracked that focus: a $1.2 million seed round, an $8 million Series A plus extension, roughly $40-50 million raised in total pre-COVID, with investors including Telstra Ventures and Steph Curry. The trade-off of borrowed supply is a thinner moat — which is why Super.com later layered financial products and a Super+ membership on top of the travel demand engine, making both businesses independently profitable.

If one side of your market already exists in aggregatable form, you're not really building a two-sided marketplace — you're building a demand machine with a supply integration.

Key stat: SnapTravel passed $100M in GMV at a 10-15% take rate without owning a single hotel room — by borrowing supply and building only the demand side.

What happens when you own the supply side yourself?

You get control — and capital intensity that punishes weak unit economics. Dan Park, CEO of Clutch, Canada's largest online used-car marketplace, holds actual inventory: Clutch buys, reconditions, and delivers the cars. When capital was free, that model scaled explosively.

"We went from 10 million in revenue in '19 to 200 million in revenue... we 20x'd the revenue in three years and then all of a sudden it was very quick." — Dan Park, Clutch

The "all of a sudden" was late 2022: capital dried up, a $95 million round signed four days after Halloween collapsed on January 5th, and within 12 days Clutch cut 150 people. Headcount fell from a peak of 350 to a low of 87. The rebuild, according to Park, came from treating each car as a unit-economics problem: filling a windshield chip for $20 instead of replacing it for $800-900, attaching warranty and financing revenue, cutting inventory aging, and shutting three unprofitable west-coast markets that were dilutive to EBITDA.

Hiring discipline followed the same logic:

"Licensed mechanics — if you sell 10 cars, you need, call it, one. If you're going to sell 20, you need two. So those are linearly variable, but we didn't really do a lot of dev or product hiring." — Dan Park, Clutch

Clutch doubled the business while getting profitable, reaching $400 million in revenue — double its previous peak.

Key stat: Clutch went from 350 employees to 87, then rebuilt to $400M in revenue — 2x its prior peak — by fixing per-unit economics instead of chasing GMV.

What if one side of your marketplace never converges?

Then sell your product to the side that already loves it. Andrew Tai co-founded Unhaggle (later MotoInsight), a car-buying marketplace. For nearly three years it grew month over month — but never fast enough, because the demand side had a fatal frequency problem: consumers buy a car every 4-8 years, making demand acquisition brutally expensive. The supply side, though — dealers and automakers — used the product constantly.

"We saw dealers using it. We saw consumers using it. We just decided, hey, what if we actually enabled the dealers and the automakers that we had started working with to deliver that kind of experience to the audience that they already own?" — Andrew Tai, MotoInsight

That side-switch — from consumer marketplace to vertical software sold to the supply side — is one of the most repeatable escape hatches in marketplace history, as Tai shared on the PMF Show. And the PMF signal on the new side was unmistakable and countable:

"If you can get one customer, maybe it's luck. You get two, it's like, oh. And then you get three. It's like, that's it. That's the pattern now." — Andrew Tai, MotoInsight

The third automaker contract told Tai the pattern was real in a way three years of steady-but-slow marketplace growth never had. If your marketplace's demand side transacts too infrequently to retain, the supply side's willingness to pay is often the business you should have built first.

Key stat: Unhaggle spent nearly 3 years on the consumer marketplace; on the supply side, 3 automaker contracts were enough to confirm PMF.

Key Takeaways: How to Find PMF in a Two-Sided Marketplace

1. Liquidity is the PMF signal, not signups. Dabble had 50,000 registrations in four months, but only the 20,000-30,000 who deposited — and stuck around for 12 months with LTV above CAC — proved product-market fit. 2. Push the flywheel by hand before it rolls on its own. Mercado Libre needed initial traction before buyers attracted sellers and sellers attracted buyers — then the loop compounded at ~80% annually for two decades. 3. Audit which side self-generates. Reddit's content supply created itself; Hipmunk's airline and hotel supply required 2 of the first 12 hires to be BD. Staff for the side that must be manufactured. 4. Subsidized liquidity is not liquidity. If you're paying both sides to transact, strip out the incentives before you call it PMF — "if you're giving them money, it's easy to prop up their revenue." 5. Borrow supply if it already exists. SnapTravel reached $100M+ GMV at a 10-15% take rate by aggregating existing hotel inventory and solving only demand. 6. Owning supply multiplies both upside and burn. Clutch 20x'd revenue in three years, nearly died when capital vanished, and only became durable after fixing per-car unit economics. 7. Frequency kills demand-side economics. A purchase that happens every 4-8 years (Unhaggle) makes demand CAC nearly impossible to recover; high-frequency categories reach liquidity faster. 8. When one side won't converge, sell to the other. Unhaggle's pivot to supplying dealers and automakers found in 3 contracts the conviction 3 years of marketplace growth never delivered.

FAQ: Common Questions About Two-Sided Marketplace PMF

Q: How do you know when you have product-market fit in a two-sided marketplace?

A: When both sides transact repeatedly without subsidies and unit economics hold — lifetime value above acquisition cost on the paying side. On the PMF Show, Dabble's Jon Robin said true PMF only arrived 12 months post-launch, when retention data showed customers sticking around without bonuses propping up revenue.

Q: Should a marketplace startup focus on supply or demand first?

A: Start with whichever side does not self-generate, and concentrate both sides in one narrow category. Mercado Libre seeded supply by hand until buyers attracted sellers on their own, while Hipmunk had to dedicate 2 of its first 12 employees to supply-side BD. The side that shows up organically needs product; the other side needs people.

Q: How long does it take a two-sided marketplace to reach product-market fit?

A: Longer than SaaS — expect years, not months. Mercado Libre took 8 years to reach $78M in revenue, Dabble needed 15 months to launch plus 12 months to prove unit economics, and Unhaggle learned after 3 years that its marketplace would never converge. Liquidity, not revenue alone, is the tiebreaker.

Q: What should you do if your marketplace isn't reaching liquidity?

A: First narrow the wedge — one category, one geography — and check whether transaction frequency is the bottleneck. If one side transacts too rarely to retain (like car buyers every 4-8 years), consider selling your product directly to the side that uses it daily, as MotoInsight did with dealers and automakers.

Sources: Listen to the Full Founder Stories

• Hernan Kazah, co-founder of Mercado Libre (S4) — The buyer-seller flywheel, patient sequencing, and compounding ~80% growth for two decades.
• Chris Slowe, first employee and CTO of Reddit (S3) — Self-generating supply at Reddit versus manufactured supply at Hipmunk, and Reddit Gold's $100K first week.
• Jon Robin, co-founder of Dabble (S3) — Why subsidized revenue isn't PMF, and the 12-month wait for LTV > CAC before scaling to $150M.
• Hussein Fazal, co-founder and CEO of Super.com (S4) — Building a $100M+ GMV travel business on borrowed hotel supply.
• Dan Park, CEO of Clutch (S4) — Owning inventory, the 350-to-87 crash, and rebuilding to $400M on per-unit economics.
• Andrew Tai, co-founder of MotoInsight/Unhaggle (S2) — Three years of marketplace grind and the pivot to selling the supply side its own experience.

Last updated: July 2026