Pattern/Economics/No. 0665
Network Effects
Network effects occur when a product’s value to a user changes because other people use it. Studied in economics by Jeffrey Rohlfs, they can arise within one user group or across groups, increasing its overall usefulness or reducing it through congestion.
- Evidence
- Well established
- Read
- 6 min
- Links
- 12 connections
01You've seen this when…
- in life
Your favorite messaging app has better privacy and cleaner design. You still keep another one installed because your family organizes everything there.
- at work
You list your handmade furniture on a new marketplace with lower fees. Weeks pass without an inquiry, so you return to the expensive site where the buyers already shop.
- out in the world
A ride service spreads across your city, and pickup times shrink as more drivers join. After a concert, thousands of riders open the app at once; waits and prices jump.
02The idea
A beautifully designed messaging app can be nearly useless if nobody you need to contact uses it. Its value comes partly from the product and partly from the people connected through it.
A network effect exists when a product’s value to a user changes because other people use it. The effect can come from their number, but also from who they are, where they are, and what they contribute.
A direct effect connects users of the same kind: another friend joining your messaging service gives you another person to reach. A cross-side effect connects different groups: more suitable sellers make a marketplace better for buyers, and more buyers make it better for sellers.
The direction isn’t always positive. More riders can attract more drivers, but more riders competing for the same drivers can also mean longer waits. Congestion is a negative network effect.
To understand a platform’s value, focus on the people behind its account count: which additional users would make this product more—or less—valuable to whom?
03Why it happens
- Other users supply part of the service. A phone network supplies connections; its subscribers supply the people worth calling. Adding someone you want to reach expands what you can do.
- Participation attracts complementary participation. Buyers bring sellers, and sellers bring buyers. Software users attract developers, whose applications attract more users. This can create reinforcing feedback.
- Compatibility concentrates the benefit. If two messaging services cannot exchange messages, joining one doesn’t let you reach the other. Users may cluster around an established service even when a rival has better features.
- The right participants matter more than raw totals. Ten nearby drivers can help you more than ten thousand drivers overseas. A specialist marketplace needs relevant buyers, not just visitors. A network can be large overall and thin where you need it.
- Participation can outrun capacity. More people compete for road space, server capacity, appointments, or attention. If capacity and matching don’t keep up, an additional user can make the experience worse for existing users.
These effects can produce a critical-mass problem: people won’t join until enough others participate, yet those others have the same reason to wait.
04A worked example
In the 1999 findings of fact in the U.S. Microsoft antitrust case, Judge Thomas Penfield Jackson examined why Windows was so difficult to challenge in desktop operating systems. One important obstacle was its large supply of compatible applications.
What it looks like Users keep choosing the familiar operating system, and developers keep writing software for it. A rival might seem to need only a better product or a lower price.
What’s actually going on The two choices reinforce each other. Developers have a stronger commercial reason to support an operating system with many users. Users have a stronger reason to choose a system that runs the applications they need. An entrant must persuade users to join before it has comparable application support, while persuading developers to invest before it has comparable demand. The court described this as an applications barrier to entry.
Windows runs at the same speed when another person buys it. A large user base helps sustain the surrounding software ecosystem. That is an indirect network effect.
What would have helped An entrant needed to address application availability, not just operating-system quality. Supporting existing applications or reducing the cost of developing for several systems would target that obstacle directly, though neither is easy. This diagnosis explains one barrier identified in the case; it doesn’t explain every reason for Microsoft’s position or every disputed business practice.
05How to spot it
06What to do about it
- Draw the value connection. Write down who benefits when each kind of participant joins. If you cannot name the benefit to an existing user, you may be looking at ordinary growth rather than a network effect.
- Build density before breadth. For a local marketplace, make one neighborhood useful before recruiting scattered users across a country. For a professional tool, connect one working team rather than thousands of isolated accounts.
- Solve the missing-side problem. In a two-sided market, identify which group is holding back the other. Recruit or support that group deliberately instead of treating all signups as equally valuable.
- Give early users something useful immediately. A scheduling tool might help someone organize their own calendar before coworkers join. Standalone value reduces dependence on everyone adopting together.
- Check whether expansion improves the experience. Track successful matches, relevant choices, wait times, spam, and retention. Growth that adds irrelevant listings or overloads providers can weaken the network.
- Examine your exit options. As a buyer, check data export, compatibility, and whether you can use several services at once. A valuable network can become lock-in when leaving means losing access to people or tools you need.
07When it isn’t a network effect
A company becoming cheaper to operate as it grows is economies of scale, not necessarily a network effect. A factory can spread its equipment costs over more units without making each unit more useful because other people bought one. Both mechanisms can exist together.
Popularity can also act as social proof: you infer that a restaurant must be good because it’s busy. That’s different from its patrons actually improving your meal. A packed restaurant might instead make you wait longer.
Leaving a service can involve costs tied to the software itself. Learning a new interface or transferring files creates switching costs, even for software you use alone. Losing access to your collaborators is the network-related part.
Finally, network effects don’t guarantee monopoly or endless improvement. People can use competing services, compatibility can connect separate networks, and congestion can limit growth. More possible connections do not mean every connection is valuable.
08Roots
At Bell Laboratories, economist Jeffrey Rohlfs studied a peculiar demand problem. A communications service could have an attractive price and still struggle to recruit subscribers: people needed other subscribers worth contacting. In his 1974 paper, he formalized how each person’s willingness to join depended on others joining too.
Telephone networks had made the pattern visible long before the modern label became familiar. A telephone connected to nobody offered little. A connection to a useful community offered much more. Selling individual subscriptions therefore required building a connected population.
Michael Katz and Carl Shapiro brought this interdependence into the economics of competition in their influential 1985 paper. Their question concerned competing technologies: when customers value compatibility and a large installed base, how do firms compete, and what happens to market outcomes? The existing network could shape a purchase as strongly as the product’s own features.
The idea later became central to software and online platforms. Research on two-sided markets examined businesses that must bring different groups together, such as merchants and cardholders. What began with communications became a way to understand products whose users provide part of one another’s value.
09How solid is this?
Theory and empirical research support network effects in communications, software, and platforms. Their strength depends on compatibility, relevant participation, and congestion; establishing the effect requires evidence of how other users change a product’s value, beyond user-count growth.
10Connections
- Often confused with Bandwagon Effect, Economies of Scale, Social Proof
- Can lead toLock-In
- Part of Emergence, Reinforcing Feedback, Complementarity
- IncludesCongestion Externality, Cross-Side Network Effect
- See alsoTwo-Sided Market, Critical Mass, Switching Costs
+ 2 more in the list
11Origin and sources
Recognized in communications networks and formalized by Jeffrey Rohlfs (1974). Michael Katz and Carl Shapiro’s 1985 work on network externalities, competition, and compatibility helped establish the modern economic framework.
- [1]Rohlfs, J. (1974). A Theory of Interdependent Demand for a Communications Service. The Bell Journal of Economics and Management Science, 5(1), 16–37.
- [2]Katz, M. L., & Shapiro, C. (1985). Network Externalities, Competition, and Compatibility. The American Economic Review, 75(3), 424–440.
- [3]Rochet, J.-C., & Tirole, J. (2003). Platform Competition in Two-Sided Markets. Journal of the European Economic Association, 1(4), 990–1029.
- [4]United States v. Microsoft Corp., 84 F. Supp. 2d 9 (D.D.C. 1999). Findings of Fact, paragraphs 36–52.
- [5]Tucker, C. (2008). Identifying Formal and Informal Influence in Technology Adoption with Network Externalities. Management Science, 54(12), 2024–2038.
Suggest an edit· Updated 2026-10-02