What Is a Network Effect? Value That Grows With the Number of Users

An ordinary product delivers the same value regardless of how many other people own one, and a product with this property does not. A telephone connected to nobody is useless and one connected to everybody is essential, with the value to each user rising as the network grows. That creates a feedback loop, since growth makes the product more attractive, which produces further growth, and the loop runs in reverse too, since decline makes it less attractive and accelerates the decline. The effect divides into direct forms, where users benefit from other users of the same product, and indirect forms, where users benefit because a larger user base attracts complementary products, which is how operating systems and games consoles work, since the software available depends on the installed base and the installed base depends on the software.

Markets with this property behave differently from ordinary ones:

• •Tipping, where a market with several competitors resolves rapidly to one dominant option once a lead becomes visible
• •Winner-take-most outcomes, since the largest network is the most valuable and attracts the remaining users
• •High barriers to entry, since a better product with no users is worth less than a worse product with many
• •Lock-in, since leaving costs a user access to everyone who stayed
• •Early competition on growth rather than on profit, which explains the sustained losses of companies pursuing scale
• •Sensitivity to expectations, since users join what they expect to win, which makes the belief self-fulfilling

The most interesting consequence is that the outcome depends on history rather than only on quality. If a slightly inferior option gains an early lead, the feedback can lock it in permanently, and later entrants cannot displace it however much better they are, which is path dependence. The standard illustration is the keyboard layout, whose history is more complicated than the popular telling and which remains a reasonable example of an arrangement persisting because everyone learned it. Videotape formats, electrical standards and railway gauges supply further cases. The general point is that markets with this property do not reliably select the best option, which is a limitation on the usual argument that competition produces good outcomes, and it is one of the better-established results in the economics of technology.

The concept is invoked far more often than it holds, and distinguishing it from ordinary advantages of size matters. Economies of scale mean a larger producer has lower unit costs, which is a supply-side advantage and is a different thing entirely, since it benefits the producer rather than each additional user. A brand that more people recognise is not thereby more useful to each customer. A marketplace with more sellers does benefit buyers and can reach a point where further sellers add nothing, so the effect saturates rather than compounding indefinitely. Local effects matter more than global ones in several services, since a taxi application needs density in one city rather than users worldwide, which is why such markets have supported multiple competitors regionally. Companies claim the property routinely because it justifies losses in pursuit of growth, and the claim deserves testing rather than acceptance.

Policy responses exist and each has difficulties. Interoperability requirements force a dominant network to connect with competitors, which dissolves the advantage of size, and this is how telephone networks were opened and how several recent regulations aim to treat messaging and digital platforms. Data portability lets users take their information elsewhere, reducing the cost of leaving. Open standards prevent any single firm owning the network. Competition enforcement can block acquisitions that consolidate networks, which is the main route regulators have used, with mixed success since the effect operates faster than enforcement. Breaking up a network is the strongest remedy and is hard to apply, since splitting a network destroys the value that made it worth using, which is the argument dominant firms make and which is partly correct.

Value rising with the number of users creates a feedback loop that runs in both directions, and indirect versions operate through complementary products rather than other users. Markets tip rapidly to one dominant option, and an early lead can lock in an inferior one permanently. Interoperability and data portability attack the advantage directly, and breaking up a network destroys the value that made it useful.