Network effects
The economics that made the platform era
Description
A telephone that exists in only one household is useless. A telephone that exists in every household is so essential that its absence becomes a handicap. The difference is what economists call a network effect — the value of the product depends on how many other people use it, which means that as adoption grows, the product becomes more valuable, which accelerates further adoption. For most of the twentieth century, network effects were a minor curiosity in economics textbooks. In the past thirty years, they have become the dominant mechanism through which the largest companies in the world have captured their markets and defended their positions.
The platform economy is almost entirely built on network effects. Facebook is valuable because your friends use it. eBay is valuable because it has the most buyers, which attracts more sellers, which attracts more buyers. Uber is valuable because it has the most drivers, which attracts more riders. Visa is valuable because it is accepted at the most merchants. The dominant firms in every major platform market — social media, search, marketplaces, payments, operating systems — have reached their positions through network-effect dynamics that produced winner-takes-most outcomes the earlier economic theory did not anticipate.
What network effects are, how they work, and why they produced the industry structures of the contemporary economy are questions economic theory and business strategy have been refining since the 1980s. The concept has gone from academic obscurity to mainstream business vocabulary, with consequences for antitrust policy, investment, and competition the regulatory systems are still working out. Understanding network effects is prerequisite to understanding why certain companies have the market positions they do and whether those positions will persist.
● The question we're asking: what are network effects, how did they come to define the digital economy, and what are their limits?
● What we'll see: the underlying economics, the types of network effects, the platform companies they built, and the cases where the theory breaks down.
Table of contents
01The underlying economics
The formal study of network effects began in the 1980s with Michael Katz, Carl Shapiro, and Hal Varian. Their framework built on an observation going back to Theodore Vail and Bell's telephone monopoly: some products become more valuable as more people use them. Katz and Shapiro formalized this in 1985, distinguishing direct network effects (users benefit from other users of the same product) from indirect network effects (users benefit from a growing ecosystem of complementary products). The framework provided the foundation platform strategy would build on two decades later.
Metcalfe's Law, from Robert Metcalfe of Ethernet fame, was the popular shorthand. It stated that network value scales with the square of the number of users. A network of ten users has a hundred potential connections; a network of a hundred has ten thousand. The specific form has been challenged — recent work suggests value scales more slowly, perhaps with n log n rather than n squared — but the qualitative point is correct. Network value grows faster than linearly with size, which produces the accelerating adoption dynamics characteristic of these markets.
02The types of network effects
Not all network effects are the same. The literature distinguishes between several types that have different implications for strategy. Direct network effects, the type the telephone exhibits, mean that the product becomes more valuable as more people who use it directly interact with each other. The more people on a social network, the more of your friends are on it, the more valuable it is to you. The more people speaking a language, the more useful the language is. Direct network effects are the strongest and produce the most extreme winner-takes-all dynamics.
Indirect network effects, sometimes called cross-side effects, operate between different groups. A marketplace like eBay benefits from many buyers because sellers want buyers, and from many sellers because buyers want sellers. Each side becomes more valuable to the other as it grows. Indirect effects are the basis of two-sided and multi-sided platforms, which now include most of the major platform businesses — marketplaces, operating systems, payment networks, advertising platforms, ride-share services.
03The platform companies they built
The first generation of network-effect platforms emerged in the 1990s with eBay, Amazon, and the early online marketplaces. eBay was the cleanest case — a pure marketplace whose value depended almost entirely on the number of other users. Amazon started as a catalog retailer but progressively added marketplace functionality, third-party sellers, and complementary services. These companies grew into dominant positions that, twenty years later, remain largely intact. The network effects provided durable advantages ordinary industrial competition could not displace.
The second generation, emerging in the mid-2000s with Facebook, Google's ad business, and social platforms, took the logic further. These were attention and advertising platforms whose value depended on the audience they assembled. The bigger the audience, the more valuable to advertisers; the more advertiser money, the more reinvestment that attracted more audience. The advertising-supported platforms became the most valuable network-effect businesses in history, with Google and Facebook reaching trillions in market cap on network-effect dynamics that locked in their positions.
04Where the theory down
Network effects are not invincible. Several cases demonstrate the limits of the framework and the conditions under which network-effect advantages erode. MySpace was displaced by Facebook despite having a substantial early network effect lead, because Facebook's product was enough better that users were willing to migrate despite the network cost of leaving their MySpace connections behind. AOL was displaced by the open web despite its dominant position in dial-up internet access, because the broader web provided more value than AOL's walled garden could. Yahoo was displaced by Google in search despite its early lead. The pattern suggests that network effects alone do not produce permanent dominance if product quality differences become large enough or if underlying technology shifts.
Multi-homing is another limitation. The theoretical winner-takes-all prediction assumes that users will not use multiple competing platforms simultaneously. In practice, many platform markets have substantial multi-homing — users use both Uber and Lyft, both Visa and Mastercard, multiple social media platforms. When multi-homing is easy and the costs of participating in multiple platforms are low, the tipping effect is weaker. Several platform markets that theorists expected would consolidate into single winners have remained competitive because multi-homing preserved demand-side competition between platforms. The extent of multi-homing is one of the key variables determining how strongly network effects produce concentration.
05Why it still matters
Network effects matter as a subject because they are the main explanation for why the contemporary technology economy is dominated by a small number of enormous firms rather than competed over by many smaller ones. The specific concentration of the digital economy — the handful of companies that control search, social media, e-commerce, mobile operating systems, and cloud infrastructure — is a direct consequence of network-effect dynamics that the earlier industrial economy did not share. Whether this concentration is desirable, tolerable, or a problem requiring regulatory intervention is one of the major policy arguments of the current period, and the argument runs on network-effect economics at its core.