Marketplaces are often described as businesses that connect buyers and sellers. That is true in the same way that a bank is a place to keep money. Accurate, but increasingly incomplete.
The early marketplace problem is usually liquidity. Can you assemble enough supply to attract demand, and enough demand to keep suppliers engaged? If that flywheel starts turning, the business can look deceptively simple: more buyers attract more sellers, more sellers improve selection, and the resulting volume strengthens the platform.
But liquidity is rarely where the most interesting marketplace businesses stop. Once enough transactions begin flowing through the system, the marketplace starts accumulating something far more valuable than traffic. It learns how commerce actually works.
It sees what customers search for and what they ignore. It understands which sellers convert, where fulfilment breaks, how price affects demand, what causes a return, how promotions behave and which customers are worth acquiring. It begins to sit inside payments, logistics, advertising, customer service and increasingly the software sellers use to operate.
At that point, the marketplace is no longer simply facilitating commerce.
It is becoming part of the infrastructure underneath it.
Liquidity gets you into the game
The first generation of marketplace strategy was understandably obsessed with network effects. The basic economics were compelling: aggregate fragmented supply, remove discovery friction and make it easier for buyers and sellers to find one another.
Scale mattered because marketplaces are unpleasant places when they are empty. Customers want choice. Sellers want customers. Neither side is particularly interested in waiting around while the other arrives.
Once that liquidity exists, however, its value changes.
The marketplace now occupies a position between demand and supply that few individual participants can recreate. It sees the transaction from both sides. It understands what the customer wanted, what the seller offered, what price cleared the market and what happened after the purchase.
That creates a different kind of advantage.
A retailer sees its own customer behaviour. A brand sees its own products. A logistics provider sees movement. A payment provider sees money.
A scaled marketplace can begin to see all of them together.
This is where the strategic conversation should move beyond GMV.
Volume matters enormously, but the deeper question is what the platform learns from being in the middle of that volume and which capabilities become possible because of it.
The transaction is the beginning, not the end
Traditional commerce tends to treat the transaction as the outcome.
The customer converted. Revenue was generated. The order moves into fulfilment and the commercial funnel starts again.
For a marketplace, the transaction can be much more productive.
It creates another piece of demand intelligence. It generates payment information. It tests the quality of a seller. It creates fulfilment data, customer-service signals and evidence about whether the product was actually satisfactory once it arrived.
Over time, those signals begin to support adjacent businesses.
Payments are an obvious one. If the marketplace is already trusted by both sides of the transaction, handling the movement of money becomes a natural extension. The platform may offer faster settlement, fraud protection, credit, working capital or other financial services because it understands a seller’s commercial activity better than many traditional providers do.
Logistics follows a similar logic. Once the platform knows what has been sold, where inventory sits and where the customer lives, fulfilment becomes less of an external step and more of an optimisation problem.
The important point is not that marketplaces should launch every adjacent service they can imagine. Conglomeration for its own sake creates plenty of expensive distractions.
The point is that the transaction generates information and trust that can make certain adjacent services structurally easier for the marketplace to provide than for an outsider.
That is how the business gradually moves from earning economics on the transaction to earning economics around the transaction.
The strongest infrastructure becomes difficult to remove
There is a meaningful difference between being useful and being embedded.
A seller may use a marketplace because it provides incremental demand. That relationship can remain relatively transactional. If another platform produces better economics, the seller can move budget or assortment accordingly.
The relationship changes when the marketplace also manages payments, provides fulfilment, funds inventory, supplies advertising tools, generates customer insights and connects into the seller’s operating systems.
Removing the platform now creates work.
This is where marketplace power starts compounding.
The fulfilment service improves because there is more volume. Better fulfilment can improve conversion. Higher conversion attracts more sellers and products. More transactions create better demand information, which can improve advertising and inventory decisions. The platform becomes better not simply because it is larger, but because the capabilities around the marketplace reinforce one another.
There is a customer version of this as well.
Customers may initially arrive because the marketplace has selection. They remain because payment is easy, delivery is reliable, returns are familiar and the platform has accumulated enough behavioural understanding to reduce effort.
None of those individual capabilities is impossible to reproduce.
The difficulty lies in reproducing them together.
That is why I think the most defensible marketplaces eventually become less dependent on being the most exciting destination and more dependent on being the most useful system.
The interface remains visible. The infrastructure underneath becomes the moat.
Retail media is a clue to what comes next
Retail media is often discussed as a particularly attractive marketplace adjacency because the margins can be significantly better than the core transaction.
That is true, but it can obscure the more interesting shift taking place.
Advertising becomes valuable because the marketplace possesses commercial information suppliers want access to. It knows what customers are searching for, what they buy, what converts and, crucially, whether an advertisement eventually produced a transaction.
The ad unit is therefore only the visible monetisation layer.
The underlying asset is commerce intelligence.
That matters because the same information can support much more than advertising. It can inform assortment, pricing, new-product development, inventory placement, demand forecasting and supplier negotiations.
A manufacturer may begin by buying sponsored placement and eventually depend on the platform to understand how its category is moving.
That is a very different relationship from renting advertising inventory.
It moves the marketplace closer to becoming an intelligence layer for the ecosystem around it.
There is an obvious tension here. A marketplace that extracts too aggressively from sellers can weaken the ecosystem it relies on. If every useful surface becomes paid placement and every piece of insight becomes another toll, suppliers eventually notice that the platform designed to help them reach customers has also become remarkably good at charging them for the privilege.
The best marketplace economics therefore require some restraint.
Infrastructure becomes powerful when participants depend on it because it makes them better, not simply because they have nowhere else to go.
That distinction matters more as the platform expands.
AI makes the plumbing more important, not less
Agentic commerce creates a temptation to think that the next marketplace battle will happen mainly at the interface.
Customers may increasingly ask an AI assistant to find a product rather than visiting a marketplace directly. An agent could compare products across multiple merchants, assemble a basket and choose where to transact based on price, availability, delivery or trust.
On the surface, that appears threatening to destination marketplaces. If the customer relationship moves into an AI interface, perhaps the traditional marketplace loses its privileged position.
Some of that will happen.
But the transaction still needs somewhere to land.
The agent needs reliable product information. It needs inventory, pricing, identity, payment, fulfilment, returns and customer-service capability. It needs confidence that the merchant can actually do what the interface promises.
In other words, AI can weaken the value of owning discovery while increasing the value of owning infrastructure.
This is why the marketplaces that have spent years building payments, logistics, identity and seller tooling may prove more resilient than those whose primary advantage was traffic.
The customer may never consciously visit the marketplace.
The marketplace can still participate in the transaction.
That is a significant strategic change.
The visible surface of commerce may become more fragmented while the infrastructure underneath becomes more concentrated.
Marketplaces may increasingly sell capabilities, not just access
This creates an interesting question for retailers and other commerce businesses as well.
If you become genuinely excellent at a capability required by the market, should it remain internal?
A retailer with unusual logistics density might eventually provide fulfilment to third parties. A business with sophisticated payments infrastructure might monetise it outside its own checkout. A company with valuable customer intelligence could create supplier services that look increasingly like software or analytics products.
Most organisations should be cautious here.
A capability that works adequately inside one enterprise is not automatically a platform. Serving external customers requires reliability, productisation, support, governance and a willingness to expose whether the capability is actually as good as internal presentations claim.
But where the capability is strong enough, externalisation changes its economics.
Cost centres can become revenue streams.
Operational scale can improve unit economics.
Internal infrastructure can become ecosystem infrastructure.
This is one reason the boundary between retailer, marketplace, technology company, logistics business and advertising platform keeps getting less tidy.
The strongest commerce businesses increasingly occupy several of those roles at once.
Not because they decided to diversify randomly.
Because enough transaction volume eventually creates capabilities that can be monetised beyond the original transaction.
The endgame is not invisibility. It is indispensability.
There will always be value in the customer-facing marketplace.
Brand matters. Discovery matters. Trust matters. A destination with enormous consumer habit remains a powerful commercial asset.
But the endgame becomes more interesting when the business can create value even when the customer is not consciously thinking about the marketplace.
The payment still runs through it.
The parcel still moves through it.
The seller still uses its software.
The brand still relies on its demand intelligence.
The AI agent still queries its inventory.
The transaction may happen somewhere else on the surface while the marketplace remains deeply involved underneath.
That is when the business begins to look less like a channel and more like infrastructure.
And infrastructure has a different competitive advantage.
Customers do not wake up excited about infrastructure. Suppliers rarely put it in their advertising. When it works, much of it is invisible.
But removing it becomes painful.
That is why the most powerful marketplace businesses may eventually be judged less by how much commerce happens on them and more by how much commerce becomes difficult to conduct without them.
The marketplace may begin as the place where the transaction happens.
Its endgame is becoming part of the reason the transaction can happen at all.