Marketplaces connecting buyers and sellers are attractive because they scale without holding inventory, and they face a starting problem that most businesses do not.
The chicken and egg problem
Buyers arrive when there is supply. Sellers arrive when there is demand.
At the start there is neither, and the platform has no value to either side.
Which is the defining difficulty, and it is why marketplaces are hard to start and defensible once established.
The approaches that work
Subsidising one side, generally the harder one to attract, until the other follows.
Starting with a narrow niche or geography, so that liquidity is achievable within it before expanding.
Providing standalone value to one side, so that they use the product even without the other side present.
Seeding supply directly, by aggregating listings from elsewhere or by acting as the supplier initially.
Each has been used successfully and each has failure modes.
Liquidity
The metric that determines whether a marketplace works.
Defined as the probability that a listing sells, or that a search finds something suitable, within a reasonable period.
Which is measurable, and it is more informative than transaction volume, since a large marketplace with poor liquidity fails to serve either side.
Liquidity is local — a national marketplace with no depth in a particular city or category does not work there regardless of aggregate scale.
Disintermediation
Where the parties meet through the platform and then transact directly, avoiding the fee.
Which is the fundamental threat to marketplaces in categories involving repeat transactions between the same parties.
Responses include withholding contact details until payment, providing services that make the platform worth using — payment protection, dispute resolution, scheduling, insurance — and pricing so that avoiding the fee is not worth the effort.
The durable answer is providing genuine ongoing value rather than blocking contact, since blocking eventually fails.
Trust
Strangers transacting requires mechanisms that substitute for prior relationship.
Reviews and ratings, identity verification, escrow payment, guarantees and insurance.
Which are expensive to build and are the actual product, more than the matching function is.
Rating inflation is a common problem, where nearly everyone receives top ratings, which destroys the signal — and it is generally caused by reciprocal rating systems.
Take rate
The percentage of transaction value the platform retains.
Which varies enormously by category, driven by how much value the platform adds and by how easily it can be bypassed.
High take rates attract competitors and encourage disintermediation, which is why rates tend to compress as categories mature.
Regulation
Marketplaces have attracted regulatory attention in several areas.
Worker classification, where platforms coordinating labour have faced litigation and legislation about employment status.
Product safety and liability for third-party sellers.
Competition concerns where a platform also sells in competition with its sellers.
And tax collection obligations, which have been extended to platforms in many jurisdictions.
All of which mean the operating model carries obligations that a simple matching service did not anticipate.
The honest assessment
Marketplaces are among the most valuable business models and among the hardest to start.
Most attempts fail at the liquidity stage, having spent capital acquiring both sides without either reaching sufficient density.
Narrowing the initial scope aggressively is the most consistent lesson from those that succeeded.
Search and matching
The core product function once liquidity exists.
Poor matching wastes liquidity — supply exists that buyers cannot find, which produces the experience of an empty marketplace despite adequate inventory.
Which makes search quality, filtering and ranking central rather than peripheral, and it is where established marketplaces defend their position.
Supply quality
Marketplaces must decide how much to curate.
Open supply grows faster and admits poor quality, which damages buyer trust.
Curated supply grows slowly and maintains standards.
Most successful marketplaces started curated and opened gradually as reputation mechanisms matured, which is the reverse of the intuitive sequence.
Unit economics
Take rate multiplied by transaction volume must exceed the cost of serving both sides.
Which is frequently not the case in categories with high support requirements or low transaction values.
Adding services — payments, logistics, financing — increases the take rate and the operational complexity, which is the path most large marketplaces have followed.
Network effects and their limits
More participants makes a marketplace more valuable, up to a point.
Beyond that point, more supply can reduce value by making search harder and by reducing the probability that any individual seller transacts.
Which means unlimited growth is not obviously desirable, and marketplaces that grew supply aggressively have sometimes damaged the seller experience enough to lose them.
Multi-homing
Participants using several marketplaces simultaneously.
Which weakens network effects, since the alternative is always available, and it is common where switching is cheap.
Reducing it requires making the platform genuinely better rather than making leaving harder, since the latter is generally circumvented.