Network Effects in Platforms: The Hardest Asset to Build and the Most Valuable Once You Have It

The conclusion first: network effects in platforms are the structural advantage that lets a small number of businesses dominate the categories they operate in — not because they outspent the competition, but because the value of their product to each new user is determined by how many other users are already on the platform. This is a fundamentally different commercial model from traditional businesses, and the strategic implications run deep. Every platform business eventually has to solve the chicken-and-egg problem of bootstrapping its network from zero. The ones that succeed build assets that compound for decades. The ones that fail typically did not understand what they were trying to build in the first place. This post is a working primer for anyone — operator, investor, executive — trying to think clearly about this category.

What network effects actually are

A network effect is, in its simplest definition, the property of a product or service whose value to each individual user increases as more users join. A telephone is useless if only one person owns it. A telephone is enormously valuable if everyone you know owns one. The underlying technology has not changed between the two situations. The user base has, and the user base is what creates the value.

This dynamic was first formalised in commercial contexts by Robert Metcalfe, co-inventor of Ethernet, who proposed in the 1980s that the value of a network grows in proportion to the square of the number of connected users. Metcalfe's Law, as it became known, was an early attempt to describe the non-linear way that user-base growth produces value in networked systems. Later economists — most notably Michael Katz and Carl Shapiro in the 1980s, then later Jean Tirole and Jean-Charles Rochet on multi-sided markets in the 2000s — refined the analysis substantially, but the core insight has held up: in certain product categories, the user base is not just a customer count. It is the asset itself.

The 2016 book Platform Revolution by Geoffrey Parker, Marshall Van Alstyne, and Sangeet Paul Choudary remains the most influential mainstream treatment of how this dynamic plays out across modern digital businesses. It distinguishes carefully between traditional pipeline businesses — which create value by controlling a linear sequence of activities from raw input to finished product — and platform businesses, which create value by enabling exchanges between participants on different sides of the platform. The strategic considerations for the two types of business are not the same, and treating a platform business as if it were a pipeline business is one of the most common reasons platforms fail to reach the scale they were structurally capable of.

Network effects in platforms are therefore not a marketing observation. They are the architectural reason that platform businesses, when they work, produce returns that traditional businesses structurally cannot match.

The two-sided (and multi-sided) market problem

Most modern platforms are not single-sided networks where users simply benefit from each other being there. They are multi-sided markets that connect distinct groups whose value to each other compounds in both directions.

Visa connects cardholders and merchants. The more cardholders carry Visa, the more useful Visa acceptance is to merchants. The more merchants accept Visa, the more useful a Visa card is to cardholders. The platform is valuable to each side specifically because the other side is well-populated.

Uber connects riders and drivers. The more drivers are available, the shorter the wait time for riders, which makes the platform more useful to riders. The more riders are active, the higher the utilisation for drivers, which makes the platform more economically attractive for them to participate in.

Airbnb connects hosts and travellers. Each side's value to the other compounds with their respective scale.

This is the structural picture of network effects in platforms as Parker, Van Alstyne, and Choudary describe it: not a single network growing, but multiple sides of a market reinforcing each other through the platform's mediation. Once both sides reach critical mass, the platform becomes increasingly difficult for competitors to displace because the cost of replicating both sides simultaneously is much higher than the cost of replicating either one in isolation.

This is also where the chicken-and-egg problem lives. A platform starting from zero has no users on either side. The first hosts on Airbnb had no travellers to host. The first travellers had no hosts to stay with. The first drivers on Uber had no riders to drive. The first riders had no drivers to call. Solving this bootstrap problem — getting both sides to critical mass simultaneously — is the central operational challenge of platform building, and the reason most platforms never get past their first year.

How the major platforms actually solved the chicken-and-egg problem

The platforms that successfully built strong network effects in platforms did not solve the bootstrap problem the same way. Each one developed a strategy specific to the structure of its market, and the patterns are worth studying because they illuminate the constraints any new platform faces.

Visa built its network by leveraging the existing infrastructure of member banks, which already had cardholders and merchant relationships. The platform did not start from zero — it inherited critical mass from its founding member institutions. This is the "piggyback" strategy: launching on top of an existing network that already has both sides populated.

Microsoft Windows built developer-side network effects by paying developers, providing tools, and creating the conditions under which application development would happen on the platform, while simultaneously selling Windows into the consumer and business markets. The two sides reinforced each other once each reached scale, but Microsoft actively invested in the developer side until consumer adoption made developer participation self-sustaining. This is the "subsidise one side" strategy.

Uber subsidised both sides aggressively in early markets — driver bonuses to attract supply, rider discounts to attract demand — until the operational density in each city was high enough that the platform's two sides became self-reinforcing. The capital intensity of this strategy was enormous, but it solved the bootstrap problem at speed. This is the "buy both sides" strategy.

Airbnb focused on quality and trust on the host side, knowing that travellers would not return after a single bad experience. The "professional photography" programme that Brian Chesky famously rolled out himself in early years was, in network effect terms, an investment in supply-side quality that paid back in demand-side retention. This is the "make one side disproportionately good" strategy.

Amazon Marketplace opened the platform to third-party sellers after Amazon's own retail operation had already built substantial demand-side scale. The marketplace inherited the demand side and only had to recruit sellers. This is, again, a variant of piggybacking — building one side as a traditional business first, then opening the other side once the asset is mature.

Stripe built developer-side network effects through extraordinary product quality and documentation, in a market (payment processing) where the "demand side" (merchants paying for payment services) was already well-defined and willing to switch given enough technical advantage. The network effects compounded as more developers built integrations, plugins, and extensions on Stripe, making it the easier choice for the next developer evaluating the category.

Shopify built its position as a merchant-side platform by combining excellent product design with a partner ecosystem of theme designers, app developers, and agency partners — each of whom had a commercial incentive to bring more merchants onto the platform. The platform's growth was distributed across the partner network rather than driven entirely by Shopify's own marketing.

These are not the same strategy. They are the same problem solved with different tools, depending on the structural characteristics of each market.

Why network effects compound and become defensible

Once a platform reaches critical mass on its various sides, network effects in platforms produce a defensive moat that is genuinely difficult for competitors to overcome.

The reason is that a new entrant faces the same chicken-and-egg problem the incumbent faced — but now with the additional challenge that the incumbent has already captured both sides. The new platform cannot offer participants more value, because value in this category is a function of who else is on the platform, and the new platform has no one. Participants will not switch unless the new platform offers them substantially better economics or a substantially better product, sustained over a long enough period that they are willing to take the risk of moving to a smaller network.

This is why dominant platforms in any given category — Visa in payment cards, Facebook in personal social networking, Uber in ride-hailing in many markets, Airbnb in short-stay rentals, LinkedIn in professional networking — have proven remarkably difficult to displace despite well-funded competitors. The competition can outspend the incumbent on a per-user acquisition basis without making meaningful progress, because acquisition alone does not solve the network value problem. Both sides have to grow together, both have to reach critical mass, and the incumbent's network is providing value to its existing participants the whole time the competitor is trying to bootstrap.

This is the mathematical expression of why network effects in platforms become structural advantages once they are established: each side's participation is rational given the other side's scale, and the rationality of participation is what defends the network from displacement.

What makes network effects fragile

The picture above is, however, incomplete. Network effects are not magic. They have failure modes, and Parker, Van Alstyne, and Choudary spend considerable time in Platform Revolution on the conditions under which they break down.

Multi-homing. When participants can be present on multiple competing platforms simultaneously without significant cost, the network effects of any single platform are weaker. A driver active on both Uber and Lyft, a host listing on both Airbnb and Vrbo, a developer publishing to both iOS and Android — each of these is participating in multiple networks rather than committing to one. The defensive moat created by network effects shrinks when multi-homing is easy and grows when it is structurally difficult.

Quality degradation. A network whose participants become lower quality on average — whether because of scaling moderation problems, fraud, or low-quality entrants overwhelming the platform's curation — loses value to existing participants even as the user count grows. Several major platforms have experienced quality collapses that destroyed network value despite raw user numbers remaining stable. Network effects only produce value when each new participant is, on average, additive to the value of existing participants. When average quality declines, the dynamic can reverse.

Disintermediation. When the platform's value is to introduce two sides to each other but those two sides can then transact directly, the platform can be cut out of the relationship over time. This is a particular risk in services platforms where repeat relationships develop between consumers and providers.

Side switching. Some platforms see participants migrate from one side of the market to the other — buyers becoming sellers, content consumers becoming content creators — in ways that change the structural dynamics of the network. Whether this strengthens or weakens the platform depends on the specific commercial design.

Negative network effects. In some categories, additional users actively reduce value to existing users — congestion on a transport network, content fatigue on a social network, spam on a communication network. Platforms in these categories have to actively manage the conditions under which growth produces positive rather than negative network value.

A well-built platform manages each of these failure modes deliberately. Most attempted platforms underestimate at least one of them, which is why the literature is full of detailed strategic frameworks for handling each case and why platforms that survive long enough to dominate their categories are, almost without exception, run by leadership teams that have internalised these dynamics deeply.

What network effects actually require: coordination at scale

The dimension of network effects in platforms that the popular literature tends to underplay is how operationally difficult it is to keep all the sides of a multi-sided market coordinated as the platform grows.

Each side of the platform has different incentives. Each has different pricing sensitivities, different quality expectations, different operational rhythms, different regulatory contexts. The platform has to design a commercial model that makes participation rational for every side simultaneously, holds together as both sides scale, and continues to produce value as the composition of participants on each side changes over time.

This is why a platform that worked in one geography frequently struggles in a new one — the coordination dynamics are different in different markets. It is why a platform that worked at a small scale sometimes fails to scale — the dynamics that made coordination tractable at 10,000 participants on each side become intractable at 10 million. It is why platforms that look similar from the outside often have very different operational realities — the commercial design choices that affect coordination are often invisible to anyone not directly involved in running them.

A platform that has actually solved coordination at scale is enormously valuable. A platform that has not, even if it has accumulated participants on both sides, is structurally fragile and tends to either dominate its category or collapse within a relatively narrow window of years.

What this means for new platform businesses

For founders, operators, and investors evaluating platform-shaped businesses, the implications of the literature on network effects in platforms are reasonably clear.

Solve for both sides early. A platform with one strong side and one weak side does not have network effects. It has a directory. The competitive position only emerges when both (or all) sides are at scale.

Choose a bootstrap strategy and execute it with conviction. The viable strategies — piggyback, subsidise one side, buy both sides, make one side disproportionately good, build a partner ecosystem — each work in specific market conditions. Doing all of them simultaneously is generally not affordable. Picking one and committing to it for long enough to reach critical mass is usually the right move.

Design the commercial model to align all sides' incentives. A platform whose participants earn when the platform earns, in proportion to the value they contribute, with transparent mechanics that every side can understand, holds together as it scales. A platform whose participants suspect they are being extracted from holds together until they find an alternative.

Manage the failure modes proactively. Multi-homing, quality degradation, disintermediation, side switching, and negative network effects each need active management. Platforms that defer thinking about these until they become problems typically discover the problems too late.

Understand that coordination is the actual work. The platform's job is not to "build a marketplace" in a narrow technical sense. It is to design and operate the commercial, technical, and operational architecture that lets all sides coordinate at scale, profitably, indefinitely. This is enormously harder than it looks from outside.

What NetworkCore is doing

This post is not primarily about NetworkCore, but the broader analysis explains, in part, why the platform was designed the way it was.

NetworkCore is a multi-sided platform connecting Charge Point Operators, private charging hosts, and Distribution Partners — businesses with EV-driving users such as fleets, OEMs, fintechs, wallets, insurers, and mobility apps. Every new CPO or private host that joins the network increases the value of the platform to every Distribution Partner already connected, because their users gain access to more chargers. Every new Distribution Partner increases the value to every CPO already connected, because their stations gain access to more demand. The commercial design — transparent public pricing as the baseline, per-session economics, settlement on a short cycle, optional bilateral arrangements layered on top of public price for specific pairs of participants — is structured to make participation rational for every side simultaneously, which is the condition under which network effects in platforms actually produce defensible value.

The platform's value compounds with each participant added on either side. The integration is single-API or iframe for Distribution Partners, with the choice of staying inside their existing PSP ecosystem or running autonomously on NetworkCore's infrastructure. For CPOs, it adds a distribution channel alongside their existing operations without disrupting their CSMS or roaming relationships. The architectural decisions reflect the platform-design literature directly: solve coordination at scale, make participation incentive-aligned for every side, and let the network effects compound from there.

This is the larger context in which platform-shaped businesses get built. The work for NetworkCore is the same work every platform that has ever produced durable network effects in platforms has had to do — solve coordination, align incentives, deliver value to every side, manage the failure modes, and keep the architecture clean as the scale grows. The reward, when it works, is what every successful platform in history has eventually accumulated: an asset whose value to each new participant is determined by everyone else already on it, and which becomes increasingly difficult for any competitor to replicate the longer the network has been operating.

If you are building, advising, or investing in something that looks like a platform, the strategic framework laid out in Platform Revolution and the broader academic literature on multi-sided markets is genuinely worth the time. The patterns are consistent. The mistakes are predictable. The opportunity, when the architecture is right, is structurally larger than anything traditional business design can produce.