Two Sided Networks Have No Value Until Both Sides Show Up

What Makes a Network Two Sided

a bilateral network It serves two different groups whose mutual value depends on the presence of the other. Merchants and cardholders. Riders and drivers. Buyers and sellers. Developers and users

The defining characteristic is cross-cutting network effects: Adding participants on one side increases value for the other side not your own. More drivers do not benefit drivers. They benefit passengers and only indirectly benefit drivers by attracting more passengers

That indirect direction is what makes these businesses difficult to start and extremely difficult to displace once they are up and running

It is worth noting that the cross-side effect can be negative within one side. More drivers means more competition for the same rates so from the driver's point of view each new driver is firstly a rival and secondly a contributor to the attractiveness of the platform. Therefore a platform recruitment offer is selling each new entrant something that slightly worsens the situation of the existing entrants which is a tougher talk than it sounds and largely determines the behavior of these companies

The Cold Start Problem

At the time of launch neither party has any reason to participate. Sellers will not post listings where there are no buyers. Buyers will not visit places where there is nothing to buy

Strategies that have worked share a structure: heavily subsidizing one party or manufacturing the supply directly

Why Density Beats Reach

That call is the least intuitive rule of the entire model so it is worth showing it rather than stating it. Illustrative and round

Suppose a marketplace has managed to register a thousand sellers. This is a genuine achievement and the same thousands can be deployed in two ways

Spread them across a hundred cities and each city will have ten. A buyer comes to their own city sees ten ads discovers that none of them are exactly what they wanted or are available when they need them and leaves. They don't come back and don't tell anyone. Each of the hundred cities offers that experience

Put the same thousand sellers in a city and a buyer will come to a real selection. They find something make transactions and come back. Sellers in that city receive orders so they stay and list more

Identical total offer. In one agreement the network works and in the other it does not exist and no amount of additional marketing solves the first case because the problem is not awareness. The problem is that the promise fails on arrival

Operating costs also go the other way. One hundred cities means one hundred local supply problems one hundred sets of regulations and support in one hundred places all incurred to produce zero liquidity anywhere

So the relevant measure is never total supply. It is supply within whatever limit a participant actually chooses and that limit is usually a city a category or a time window. Initial announcements about operations in forty markets typically describe one hundred cities with ten sellers each

Which Side to Subsidise

The economics of setting the price of a bilateral network is counterintuitive. The price is not set by the cost of each side. It looks at which side is more sensitive and which side attracts the other more strongly

The side that is harder to attract and more valuable to the other receives subsidies. The side that is captive or that benefits most from access pays

Payment networks illustrate this: cardholders receive rewards and in effect negative prices while merchants pay the fees. Cardholders are the side that must be attracted and whose presence forces the participation of merchants

The Subsidy Has a Threshold Below Which It Returns Nothing

Subsidizing a party sounds like normal customer acquisition and behaves completely differently in a way that determines how these companies should be funded

Ordinary acquisition is divisible. Spend to acquire a customer he uses the product and it is worth something whether he acquires another one or not. Stop spending tomorrow and you keep what you bought

The network subsidy is not divisible because what money buys is not a customer it is density. Spend half the way to a liquid market and you have not bought half a network. Buyers keep coming the supply is scarce and they leave sellers still do not receive orders and become agitated and the moment the subsidy stops both parties disperse. The spending returns nothing at all

So there is a threshold and the profit below it is close to zero while the profit above it is a self-sustaining business. That way has consequences that are worth stating clearly

It means that the funding requirement is irregular and not incremental. A company cannot start sensibly unless it can finance all the way to critical mass in its chosen market which is the structural reason why this category is funded by venture capital rather than direct startup

It means the failure mode is binary. Running out of money ninety percent of the way is not ninety percent of the result it's the same result as never starting which is unusual and brutal

And it means that a better-funded competitor can beat a better product simply by being able to keep both sides in place longer. This is not a market failure it's what makes a threshold and explains why these markets tend to resolve into a small number of winners rather than a spectrum of quality

Why They Are So Defensible

Once both sides are established a competitor must appeal to both simultaneously. Offering a better product to one side is insufficient because that side gets its value from the other side which is somewhere else

This is a substantially greater barrier than the advantage of a conventional product. A better restaurant can gain customers one at a time. A better restaurant market has to bring together restaurants and diners in the same city and at the same time

Where They Are Vulnerable

Three weaknesses are repeated

Multiple reference. If participants can easily use multiple platforms at once the network's advantage is dramatically weakened. Drivers running two ride apps simultaneously or sellers listing on multiple marketplaces prevent any one platform from achieving exclusivity over supply. Defenses are exclusivity terms loyalty structures and tools that make single-sourcing more convenient

Disintermediation. Participants who meet on the platform can transact outside of it which is the same threat that markets face. It bites harder here because of the subsidy described above. The platform spent real money to manufacture the density that produced the introduction and if the two parties then trade directly the platform funded a match from which it gets nothing. Exposure is worse in categories where the relationship is repeated since an introduction can be worth years of transactions that occur elsewhere

Local rather than global effects. Many networks are only valuable within one geography or category. A dominant transportation platform in one city has no advantage in another which is why those markets are contested city by city and why the national scale confers less protection than it seems

Multi Homing Is the One That Decides It

Of those three multihoming is the one worth taking a closer look at because it determines whether or not the defense described above exists

Explain again why these companies are difficult to attack. A competitor must bring both sides together at once. That argument is only valid if joining the competitor means abandoning the incumbent. If a seller can quote in both places on a Tuesday afternoon at no cost the challenger does not need to win the supply side. He borrows it

At that point the crossover effect stops being a moat because both platforms use the same group of participants and competition collapses again in prices and products like any normal business

Therefore the strategic question is never whether multihoming is possible in principle. It is what a single route is worth to a participant and the answers form a ladder

Contractual exclusivity is the most forceful and the least durable as it attracts regulatory attention and is resented by participants

Loyalty and status structures work by making volume splitting costly rather than prohibited. A seller who loses a status level by sending half of his orders elsewhere is not prohibited from making multiple visits he is simply paying for it

Integrated tools are stronger again. A platform that becomes the system on which a seller runs their business handling inventory scheduling payments and analytics has become expensive to duplicate rather than simply expensive to abandon

The strongest of all is everything that doesn't travel. Ratings review history and ranking accumulate on one platform and start at zero on the next so a seller with years of experience is not choosing between two equivalent options. That's why the platforms that last tend to be those on which a participant's reputation resides and why reputation portability is such a contentious issue whenever regulators look at these markets

Reading the Metrics

Informational measures are liquidity indicators rather than user counts. The proportion of listings transacting time to match repeat participation rate on each side and the growth balance between the two sides

A platform that grows on one side much faster than the other is building up a problem. Oversupply frustrates sellers who don't receive orders. Excess demand frustrates buyers who can't be served. Balanced growth is harder to achieve and considerably healthier

Why Growth on One Side Can Be a Bad Quarter

That last point deserves numbers because it produces the most counterintuitive result of the model: a metric that increases sharply as business worsens. Illustrative and round

Start with a market that has one hundred buyers and one hundred sellers roughly evenly matched where most buyers find something and most sellers receive orders. Both parties are having a decent experience and both are returning

Now the sellers are multiplying by ten to a thousand while the buyers remain at a hundred. The supply has increased enormously and the platform can announce it. But there are still only a hundred orders to distribute so the vast majority of sellers receive nothing at all. They did the work of listing they waited and they got silence and most of them will not bother again

Run it the other way around and it fails differently. Buyers increase tenfold versus a hundred sellers and most buyers arrive to find that what they want is already taken or unavailable which is the same failed promise described above arriving through an abundance of demand rather than a shortage of supply

In both cases the number of holders increased and retention on the other side decreased and retention is what makes it worse. A cohort of sellers burned by a month of empty inboxes will not only be absent next quarter they will be unavailable because they have already concluded that the platform is not working

That's why growth on one side is only good news when compared to the other. The honest measure is the proportion of participants on each side who actually transacted and a platform that reports user growth without it is reporting input rather than output

The Bottom Line

Bilateral networks derive value from cross-spill effects making them nearly useless at launch and extremely defensible at scale. Bridging that gap means narrowing the market until density can be achieved and subsidizing the side that is hardest to attract. The enduring vulnerabilities are multi-direction disintermediation and network effects that turn out to be local rather than global. And since the subsidy pays nothing until density is reached the money required is a threshold rather than a budget which is whywhich these markets end up with few winners instead of many competitors