To scale a marketplace business, find the binding constraint on liquidity - supply coverage, supply quality, demand, or matching - and fix that one before you spend on anything else. Then concentrate liquidity in one market before you expand. At EnergySage, that meant seeding installer coverage in Boston first and driving homeowner demand into the same market. Marketplaces that expand into new geographies or verticals before local liquidity holds dilute the network effects they are trying to build.
I have worked this problem from inside product and growth at TripAdvisor, as CPO at EnergySage, and in fractional work for marketplaces in browser gaming, solar and education. The verticals have almost nothing in common, but the first question is always the same: which constraint is binding? The answer is different almost every time.
Fabio Rosati, former CEO of Upwork, gave me the framing I still use: liquidity is a product problem. You cannot buy your way to a liquid marketplace. That is why most generic growth playbooks miss it. A SaaS funnel has one customer. A marketplace has two, and satisfying one without the other produces nothing.
You cannot buy your way to a liquid marketplace. You have to design your way there.
Four constraints, four different fixes
Every marketplace I have worked on was stuck on one of four constraints. The job is to name the right one before the budget meeting names it for you, because the budget meeting always picks demand.
Demand that was really retention: Arkadium
At Arkadium, a browser gaming platform, daily active users had plateaued and the first instinct was to add more games. The behavioral data said otherwise. Most players arrived through organic search, played one game, and left. The gap between D1 and D7 retention carried the whole story. Nothing gave a player a reason to come back tomorrow. More games would have bought more one-visit players, so the work pointed at giving the players already arriving a reason to return.
Supply quality: Rhino Eco
At Rhino Eco, a solar marketplace, homeowner interest was high. Demand was fine. Too many low-quality installers sat in the network, and cancellation rates after first contact climbed with them. Growth required tightening supply standards before scaling demand acquisition. That runs against every standard growth playbook, which says to grow both sides as fast as the budget allows. A smaller network of installers who close the job beats a bigger one that burns the homeowner.
Supply coverage: EnergySage
At EnergySage, the chicken-and-egg problem was acute. Installers would not pay to be on the platform without qualified leads, and homeowners would not engage without credible installer options. We seeded the supply side in high-density geographies first - installer coverage in Boston, not everywhere - then drove demand acquisition into those same markets. Concentrated liquidity came before geographic expansion.
Matching: TripAdvisor
TripAdvisor's metasearch business was a marketplace connecting travelers with hotels and OTAs competing for the booking. Traffic was never the constraint. TripAdvisor had enormous organic reach. Monetization efficiency was: the right traveler, the right hotel offer, the right moment, so the click was worth something to both sides. The $100M+ of growth came from improving the signal between traveler intent and hotel inventory. Hotel partners got better return on their bids. Travelers got more relevant results. Liquidity improved on both sides at once. Instant Booking, built on the same traffic, is the cautionary half: $0 to $200M in 18 months, then wound down. I cover both halves in the Instant Booking story.
How do you scale a marketplace business?
The strategies that scale a marketplace treat supply and demand as two separate problems with one shared outcome, the completed transaction. Four moves show up again and again. Launch narrow, in one geography or one vertical, and build local liquidity before scaling. Seed supply to a minimum viable inventory before you open demand acquisition. Pay referrals to both sides of the transaction. Build trust infrastructure - reviews, verification, guarantees - that lowers the cost of transacting with a stranger.
The strategies that fail are just as consistent. Broad geographic coverage before liquidity is dense anywhere. Demand acquisition before supply is ready. Registration targets in place of completed transactions. A marketplace with 10,000 registered users and thin GMV has a liquidity problem, and more volume makes it worse.
The chicken-and-egg problem is a sequencing decision
Every marketplace has one side that will show up before the other one exists. Find that side and build for it first. At EnergySage it was installers in dense markets, because coverage in one metro was achievable and coverage nationally was not. In other categories it is the side that gets standalone value from the product, a tool or a listing they would use with zero counterparties on the other end.
I now run that diagnosis with data before anyone argues about strategy. The method is in how I use AI to diagnose marketplace liquidity: pull the funnel by side and by market, find where matches stall, and name the constraint in one sentence.
B2B marketplaces add a buying committee
In a B2B marketplace the demand side is a group of people with a procurement process, and trust comes from verification, payment terms and a supplier's track record more than from star ratings. Sales cycles are longer, order values are higher, and one lost account removes far more liquidity than one lost consumer.
BetterLesson, an edtech marketplace selling into schools, is where I worked this version of the problem; I wrote up what BetterLesson taught me about edtech marketplaces separately. The constraint question stays the same in B2B. The unit of demand changes from a person to a committee.
Trust is the product in a marketplace of strangers
A marketplace asks two strangers to transact, so trust is part of the product and has to be designed and measured like one. Reviews, verified profiles, guarantees and clear refund rules all lower the perceived risk of the first transaction. The Rhino Eco cancellations were a trust problem that showed up in the supply data. A homeowner who meets a weak installer does not blame the installer. They blame the marketplace, and they do not come back.
Measure transactions, not registrations
Registrations measure marketing. Liquidity shows up in a short list of numbers: the share of demand that finds a match, time to first transaction on each side, repeat transaction rate by cohort, and how those numbers differ by market. I track them by side and by geography, because a blended average hides the market that is working and the one that is not. My fuller list is in key product metrics, and the case for picking one north star metric applies twice over in a marketplace: pick one that moves only when both sides win.
Expand when the core market carries itself
The signal to expand is a core market that passes the two-sided test without help. New supply brings demand, new demand brings supply, and acquisition cost falls as the market fills in. Expand before that and every new market dilutes the one that was working. Comparison marketplaces face a newer version of the same question as AI agents start doing the comparing, and the answer still starts with which side you own.
Building or fixing a marketplace growth engine?
Marketplace growth runs on its own logic: supply and demand balance, liquidity management and network effects. I have grown marketplace businesses from zero to hundreds of millions in revenue, and I work with marketplace founders and product leaders on exactly these problems as a fractional CPO. Let's talk.
