How to Solve the Chicken and Egg Problem in Marketplaces

Every marketplace founder hits the same wall in month two. You've built the thing. Sellers won't list because there are no buyers. Buyers won't come back because there's nothing to buy. The product works fine. Nobody's using it. That's the chicken and egg problem, and it's the most common reason marketplace startups die, ahead of bad ideas, bad code, and running out of money.

Here's the part most founders get wrong: the chicken and egg problem isn't solved with product. It's solved with a deliberate, unscalable, often embarrassing seeding strategy that runs for months before the marketplace looks like a marketplace at all. Airbnb's founders went door to door in Brooklyn photographing apartments. DoorDash's founders printed restaurant menus, took the orders themselves, and drove the food over in their own cars. Those weren't growth hacks. They were the entire business for the first year.

Below: the seeding strategies that work, which one is 10x more capital efficient than the rest, and how to tell whether your cold start is working before you have enough volume to trust any metric.

What is the chicken and egg problem in a marketplace?

The chicken and egg problem is the fact that a marketplace has no value to either side until both sides are present, so neither side has a reason to show up first. A ride-hailing app with no drivers is useless to riders. With no riders, it's useless to drivers. The value is entirely in the aggregation, which means at launch you have a product with a real cost and zero benefit.

This is different from a normal cold start. A SaaS tool with one customer still works for that customer. A marketplace with one seller and zero buyers works for nobody. That asymmetry is why marketplace startups burn more capital and take longer to reach product market fit than almost any other model, and why "just launch and iterate" falls apart here.

Sangeet Paul Choudary calls the fix "standalone mode." Chris Dixon calls it "single player mode." Same framing either way: make one side of the marketplace valuable on its own, before the other side exists.

Which side should you build first, supply or demand?

Supply, in most cases, but the real answer is whichever side is more constrained. Supply usually is, because sellers are fewer, more concentrated and easier to find. There are 200 restaurants in your city and 400,000 people who eat.

Supply also compounds differently. A seller who lists once stays listed. A buyer who visits an empty marketplace never comes back. You get one shot at demand, so spend it after supply is dense enough to convert.

The exception is when demand is scarce. Recruiting marketplaces are the classic case: candidates are plentiful, hiring budgets are not. Same with high-ticket B2B where a handful of buyers control most of the spend. If ten buyers represent 80% of transaction volume, go get the ten buyers and supply follows.

Test it before you commit. Try to get 20 suppliers to list with no buyers, and 20 buyers onto a waitlist with no supply. Whichever conversation is harder tells you which side to build first.

What is single player mode and why is it the most capital-efficient way to start?

Single player mode means giving one side a product that's useful even if the marketplace never materialises. It's the most common seeding strategy among the largest marketplaces, and by a wide margin the most capital efficient.

Eli Chait's team studied the 100 largest marketplaces after selling their company to OpenTable. They found 34% used some form of single player mode as the initial value proposition. More interesting is the economics. Marketplaces seeded this way averaged a revenue-to-funding ratio of 10.03. Marketplaces that seeded by promising to fill empty seats for suppliers averaged 1.01. Ten times the revenue per dollar raised.

OpenTable is the canonical example. Before it was a reservation marketplace, it sold restaurants an electronic reservation book: table management and CRM software, on a subscription, useful on day one with zero diners on the platform. Once it had hundreds of restaurants in a city, it had something worth showing diners. It was still making most of its revenue from software subscriptions when it filed to go public, ten years after founding.

Amazon ran the same play from the other direction. Book retailer first, buying inventory and reselling it, which created value for buyers without any third-party sellers. Once it had the buyers, it opened the platform to sellers.

Three things drive the capital efficiency, per Chait's analysis:

• Less competition for supply. If your pitch is "we'll fill your empty seats," so is every competitor's, and switching is free. A tool your supplier runs their business on locks them in.
• Lower churn. Operationally critical software is painful to leave. A listing is not.
• Cash flow funds growth. OpenTable collected subscription revenue years before the marketplace had liquidity. Uber and Lyft needed thousands of drivers and hundreds of thousands of riders before the model produced real gross margin, and had to raise against that gap.

If there's a job your supply side currently does in spreadsheets, build that. The marketplace comes later.

How did Airbnb, Uber and DoorDash actually seed their first users?

They filled empty seats, and they did it manually, one supplier at a time. Filling empty seats was the second most common strategy in Chait's study at 33%: you find businesses with underused inventory and offer them incremental revenue at no upfront cost and no downside risk.

Uber started with existing black car and limo drivers who spent most of their day parked between pre-booked appointments. Uber handed them an iPhone running the driver app and told them to accept rides during downtime. No new supply was created. Existing idle supply got monetised. Later, in specific San Francisco neighbourhoods, Uber paid drivers by the hour to circle empty streets so that the few riders who opened the app saw a car nearby.

Groupon did the same with merchants: a discounted gift card, nothing paid upfront, Groupon taking a cut only of what it sold. One deal per day, marketed hard on Google and Facebook, meant it needed very little supply to launch.

Airbnb went narrower. It built dense supply in specific cities during events where hotels sold out, then walked Brooklyn brownstones taking listing photographs itself because bad photos were killing conversion.

DoorDash picked different geography on purpose. Instead of fighting for dense urban markets, it launched in suburbs and mid-sized cities where competition was thin and it could dominate quickly. In the earliest days the founders posted menus from restaurants near Stanford, took the orders, picked up the food and delivered it themselves. The restaurants didn't know they were suppliers yet.

None of these companies built a marketplace and waited. They did the marketplace's job by hand until enough people showed up to do it for them.

How small should your first market be?

Small enough to be the obvious best option in it, which for most founders is one city, one category or one campus. Density beats reach at the cold start, and the instinct to launch nationally is the fastest way to end up 1% full everywhere and 100% full nowhere.

Poshmark shows how narrow this can get. Its first phase focused on a small group of fashion-obsessed users for six months, ending with 500 users total, 300 of them active with a closet listed, buying and selling. Five hundred users is a failure by any growth dashboard. It was the right number, because those 300 people made the marketplace feel full.

Whatnot shows the same logic at scale. Its fastest-growing categories in 2025 were beauty at +791% year over year, electronics at +444%, jewellery at +259% and women's fashion at +223%, each built as its own dense pocket rather than a general "sell anything here" pitch. The platform crossed $8 billion in live GMV that year, roughly double the prior year's $3 billion.

Pick your pocket using three filters: transaction frequency high enough that users come back, a supply side you can reach without paid acquisition, and a boundary you can credibly claim to have covered. If you can't say "we have every good X in Y," you haven't picked narrow enough.

What is a liquidity hack and how do you design one?

A liquidity hack is the specific, usually uneconomic thing you do to make transactions happen before the market can produce them on its own. Alex Taussig of Lightspeed, who evaluates early-stage marketplaces primarily on this question, describes the prime directive of a marketplace as generating liquidity where none existed before, and calls the liquidity hack the founding insight of most great marketplace businesses.

His examples are each a different shape of the same move:

• Uber paid drivers to circle key neighbourhoods with no passengers in the car.
• Airbnb paid for professional photography so guests could judge quality.
• Faire guaranteed items would sell and offered retailers net 60 payment terms.
• thredUP processed the merchandise on the seller's behalf.

Two are subsidies, one is risk transfer, one is doing the work for the user. Each costs money per transaction and none scales as written. That's fine. The point is to buy enough completed transactions that both sides learn the marketplace works, then remove the subsidy as density replaces it.

To design yours, ask which step in the transaction is failing. Sellers list and nobody buys: trust or presentation, so Airbnb's answer. Buyers search and find nothing: availability, so Uber's. Suppliers won't join because the downside is unclear: transfer the risk, so Faire's. Suppliers join and then don't do the work: do it for them, so thredUP's.

What should you measure before you have a real marketplace?

Fill rate, not GMV. Fill rate is the percentage of sessions where a user showed real intent and ended up transacting. At Airbnb, intent meant searching with specific dates. At Uber, entering a destination. At Etsy, searching a keyword.

Fill rate beats volume because it bakes in supply quality, availability and funnel conversion at once. A16z calls the same idea match rate. Sarah Tavel of Benchmark calls a version of it happy GMV: the share of transaction volume that reflects an experience good enough to bring the user back. If a fifth of your rides end in fast pickup and a good review, you're at 20% happy GMV, and total GMV tells you almost nothing by comparison.

Lenny Rachitsky, who spent years on both sides of Airbnb's marketplace, lists four metrics worth optimising early:

1. Fill rate, the percentage of intentful sessions that convert
2. Bookings, completed transactions per week or month
3. Supply growth, new active supply per week or month
4. GMV growth, dollars flowing through the system

"Active" is doing real work in metric three. Registered suppliers are a vanity number. Suppliers with live, available, priced inventory are the number.

Don't benchmark fill rate against anyone else's. It ranges from under 5% in e-commerce marketplaces to over 80% at the bottom of a narrow funnel, so the absolute figure is close to meaningless. What matters is the direction it moves as you add supply. If fill rate is flat while supply grows, you're adding the wrong supply.

Why do most marketplace startups fail at the cold start?

They build both sides at once, at the same rate, in a market too big to fill. The failure modes, in rough order of how often they prove fatal:

1. Launching wide. Supply spread across 30 cities means no city has enough to convert a buyer. One city with everything beats thirty cities with something.
2. Recruiting supply that isn't supply. Signups are not inventory. A seller who lists once and never updates availability hurts fill rate by producing failed transactions.
3. Spending on demand too early. Paid acquisition into an empty marketplace burns cash and burns the users, who don't come back after the first empty search.
4. Refusing to do unscalable work. The photography, the manual matching, the phone calls. Founders who skip this stage never reach the stage where they don't have to.
5. Optimising GMV before fill rate. Volume can rise while the experience gets worse, producing a marketplace that looks healthy right up until retention collapses.
6. No answer to "why now, why here." If your supply side can't say what they get today, before the network exists, you don't have a seeding strategy. You have a hope.

The through line: the cold start is a period where you deliberately do things that don't scale and measure things that aren't impressive. If your plan for months one through six looks like a scaled marketplace with smaller numbers, it won't work.

Key takeaways

• The chicken and egg problem is structural, not a product bug. Neither side has a reason to arrive first, so you have to manufacture one.
• Build the constrained side first. Usually supply, but test both before committing.
• Single player mode is the most capital-efficient seeding strategy on record: 34% of the 100 largest marketplaces used it, at roughly 10x the revenue per dollar raised of the next strategy.
• If you can't build a standalone tool, fill empty seats. Find businesses with idle inventory and offer incremental revenue at zero upfront cost.
• Go narrow enough to be complete. Poshmark won with 500 users because 300 were active.
• Design an explicit liquidity hack aimed at the exact step that's failing, and accept that it won't scale.
• Track fill rate, bookings, active supply growth and GMV growth, in that order. Fill rate is the health metric. GMV is the vanity metric.
• The unscalable phase isn't a detour before the real business. It is the real business, for longer than you'd like.

If you're mapping a marketplace and want the seeding strategy, unit economics and competitive position in one place before writing code, Foundra walks first-time founders through that as a structured sequence rather than a blank document. A spreadsheet and a clear head work too. More frameworks like this one are at foundra.ai/key-reads/.

FAQ

How long does it take to solve the chicken and egg problem?
Plan for six to twelve months of manual seeding before the marketplace sustains itself in your first market. Poshmark spent six months getting to 300 active users. If you've budgeted three months, you've budgeted for failure.

Should I fake supply or demand to get started?
Fake listings destroy trust and produce failed transactions, which is worse than an empty marketplace. Doing the supply side's work yourself is different and legitimate: DoorDash's founders delivering food themselves was real supply, performed manually. Be the supply. Don't invent it.

How many suppliers do I need before opening to buyers?
Enough that a typical buyer search returns a usable result, which depends on your category. For a local services marketplace that might be 20 providers. For a fashion resale app it's thousands of listings. Work backwards from fill rate: run 20 realistic searches yourself and count how many would end in a transaction.

Can I start a marketplace without funding?
Yes, and single player mode is how. A tool you charge for from day one funds the seeding period out of revenue, which is why it outperforms subsidy-driven strategies on capital efficiency. Paying drivers to idle requires capital by definition.

Commission or subscription at the start?
Commission at zero upfront cost lowers the barrier to supply joining, which matters more than revenue in month one. Subscription works when you're leading with a tool that has standalone value. Match pricing to the seeding strategy, not to what competitors charge at scale.

What if a big competitor already owns my market?
Pick a segment they serve badly and be complete inside it. DoorDash launched in suburbs precisely because incumbents were fighting over dense urban markets. Density in a segment nobody's defending beats thin presence in the segment everyone wants.