The chicken-and-egg problem is the defining challenge of the first two years of every two-sided marketplace. Without sellers, buyers have no reason to show up; without buyers, sellers have no reason to list. The instinct is to solve both sides at once with two parallel marketing pushes. That strategy has almost never worked. The teams that actually got marketplaces off the ground — Airbnb, Uber, DoorDash, and a long list of less famous survivors — picked one side, seeded it manually, stayed geographically concentrated, and refused to expand until their first market was genuinely liquid. This piece is the 2026 version of that playbook, with the tactics that still work and the ones that no longer do.
Pick a side — supply, almost always
The first real decision is which side to seed. The usual answer is supply. There are fewer suppliers than buyers in most marketplaces, they are easier to identify, and they are more willing to take a meeting with a founder because the upside for them — new customers — is clear and immediate. Uber solved supply first by recruiting drivers in person and paying them to be on the road whether or not there were riders. DoorDash's three founders posted menus from nearby restaurants, took the orders themselves, picked up the food, and delivered it. The supply came before the platform, and the platform existed because the demand started to arrive.
Airbnb is the famous exception, and it's not really an exception once the details are examined. The team did solve demand first, but only by piggybacking on Craigslist's existing demand — running a script that contacted landlords posting on Craigslist and offered them a second listing on Airbnb. The underlying pattern is the same: find a side that can be acquired mechanically, get it to a point where the other side has a real reason to show up, and then stop.
The tactics that solved the dead start
| Tactic | Used by | When it works | When it fails |
|---|---|---|---|
| Manual supply recruitment | Uber, DoorDash, Airbnb | Supply side is local, identifiable, and willing to take a meeting | Supply is remote, fragmented, or low-trust by default |
| Fake-it-till-you-make-it supply | DoorDash, early Instacart | The service can be delivered by the founders for the first 100 orders | The operation is physically impossible to fake at any volume |
| Piggybacking on an existing network | Airbnb via Craigslist, PayPal via eBay | A larger platform has the demand and a loophole to reach it | The host platform closes the loophole and growth stops |
| Heavy supply-side subsidies | Uber, Lyft, every food-delivery market | Capital is available and unit economics improve with scale | Subsidies become the product — demand disappears when they end |
| Vertical niche focus | OpenTable, early Etsy, Zillow | A specific category has under-served demand and findable supply | The niche is too small to graduate beyond |
| Single-player mode | OpenTable, Calendly, early Alibaba | One side gets software value before the marketplace activates | The software is not good enough to stand on its own |
Every tactic in this table has a corresponding failure case. The point is not to pick one and ride it — it is to pick the one that matches the shape of the marketplace and abandon it the moment it stops producing real liquidity.
Single-player mode — the most underused pattern
Single-player mode is the tactic most teams overlook, and it may be the most robust. The idea is that the supply side gets real, standalone value from the product before the marketplace activates. OpenTable built reservation software for restaurants that was useful whether or not any diners used it — when enough restaurants adopted it, the diner-facing marketplace layered on top of an already-working SaaS product. Calendly is a later version of the same pattern: individuals used it to book meetings, and the marketplace of service providers emerged from a critical mass of calendars already in the system.
The test for whether single-player mode is viable is simple. Would a supplier pay for the software if the marketplace never launched? If yes, single-player mode is available and it's usually the lowest-risk path into a two-sided network. If no, the marketplace has to solve the cold start the hard way.
Geographic density — the most important rule
The rule that successful marketplaces followed and most failures violated is geographic density. DoorDash did not try to serve every city at once — the founders started in Palo Alto with a small set of restaurants, perfected the operation, and only then expanded. Uber launched San Francisco, worked out the mechanics, and then replicated the playbook city by city. A marketplace with thin coverage in twenty cities is worse than a marketplace with thick coverage in one, because thin coverage produces bad matches — a rider who can't get a ride in under ten minutes, a restaurant a buyer cannot actually order from — and bad matches produce churn that never gives the marketplace a chance to compound.
The density trap is the mistake of measuring traction in total signups rather than in per-market liquidity. A marketplace with 50,000 users across 50 cities is almost always worse off than one with 20,000 users in a single city, because the 50-city version fails to deliver a usable experience anywhere. Vanity totals hide the fact that no individual user is having a good time.
The ordered playbook
- Pick the first market narrowly. One city, one campus, one neighborhood — narrow enough that the founders can see every transaction for the first six months.
- Pick the seeded side. Usually supply, occasionally demand if a piggyback opportunity is available. Document why, in writing, so the decision isn't revisited weekly under pressure.
- Hand-recruit the first 20 suppliers. Not through ads; through meetings. The goal is to learn what the supply side actually needs, not to optimize a funnel.
- Fake what cannot be bought. If the platform's product depends on a workflow that doesn't exist yet, the founders run the workflow manually for the first 100 transactions. DoorDash's founders delivered food themselves. That is the playbook, not an anecdote.
- Measure per-market liquidity, not global signups. The metric is the percentage of requests that find a match within the SLA — rides found in under 10 minutes, bookings confirmed in under 24 hours, listings that produce an inquiry in under a week.
- Stay in the first market until liquidity is stable. Stable means the curve has flattened at an acceptable level for at least a quarter with no founder intervention on any individual transaction.
- Expand to adjacent geography before adjacent niches. The second city is a replay of the first-market playbook with faster learnings. The second vertical is an entirely new cold start and should be treated as such.
- Subsidize only with a plan to stop. Every dollar of supply subsidy should have a trigger condition for when it ends. Markets that never graduate off the subsidy are not really marketplaces.
Niche expansion — the second cold start
Once the first niche and the first geography are working, the question becomes where to expand. The usual instinct is horizontal expansion — same geography, more verticals. The usual outcome is that the second vertical is an entirely new cold start and the team underestimates it because the first one succeeded. A good heuristic is that adjacent verticals should only be added when the first vertical requires less than 20% of leadership attention and the infrastructure — identity, payments, reviews, support — can be reused without modification. If the second vertical requires rebuilding the trust layer, it is not an expansion; it is a second product.
Seed tactics that no longer work in 2026
A few classic tactics have aged badly. Craigslist piggybacking is gone — platforms that had open APIs have closed them, and the loopholes that let Airbnb scrape listings are not available to anyone today. Pure supply subsidy without a path to profitability is harder to fund in a 2026 capital environment than it was in 2015. Growth-at-all-costs hiring that throws contractors at supply recruitment works for a quarter and then collapses when the cost per acquired supplier fails to amortize against lifetime value. The tactics that still work — manual recruitment, single-player mode, geographic density, fake-it-till-you-make-it operations — are the slower and more boring ones. That's the pattern.
The liquidity metric — what to actually watch
Total signups is the wrong metric. Gross transaction volume is also the wrong metric, because it can be inflated by subsidies and founder-run transactions. The right metric is match rate within a service-level window — the percentage of demand-side requests that find a supply-side match within the timeframe users expect. Ride within 10 minutes. Booking confirmed within 24 hours. Freelance brief matched within a week. When that number is stable and above the user's tolerance threshold, the marketplace is liquid. When it is not, no amount of marketing spend will fix it.
The single most useful dashboard a marketplace team can build early is the per-market match-rate chart. It shows where expansion is premature, where density is holding up, and where supply or demand has started to erode before any other metric catches it.
Key takeaways
- Pick one side of the marketplace and seed it manually. Supply is usually the right side because it's identifiable, recruitable, and has a clear reason to show up early.
- Single-player mode is the most underused tactic. If the supply side would pay for the software without the marketplace, that is the safest cold start available.
- Geographic density beats geographic coverage for the first two years. Twenty cities with thin liquidity is worse than one city that genuinely works.
- Measure match rate within an SLA window. Total signups and gross volume are lagging and easy to fake; match rate is the first number that reflects a working marketplace.
- Expand only when the current market does not need daily leadership attention. Adjacent verticals are second cold starts, not extensions — treat them as such.
- The tactics that still work in 2026 are the boring ones. Manual recruitment, density, and patient expansion — the playbook has not changed, even if the capital environment has.