a16z Podcast: Not all Network Effects Are Created Equal

Episode Summary

Executive Summary: The episode traces the history of network effects from AT&T and Ethernet to modern software platforms, then breaks down a taxonomy of nine network-effect types and related defensibilities. James Currier and Anu Hariharan emphasize that not all network effects are equal: real-name personal utility, marketplaces with liquidity, platform ecosystems, and bandwagon effects each behave differently and require different growth playbooks. The key takeaway is that language matters because it helps founders and investors identify the right moat, growth levers, and timing.

Main Topics: History and origin of network effects (Priority: 5/5): The conversation opens with the early use of the term by AT&T in 1907, then connects it to Metcalfe’s law and the Ethernet era, showing how network effects first emerged in hardware before becoming central to software. Network effects as a source of defensibility (Priority: 5/5): The speakers argue that in software, traditional moats have weakened, leaving network effects as one of the most powerful native forms of defensibility alongside brand, scale, and embedding. Taxonomy of nine network-effect types (Priority: 5/5): James Currier explains the need for precise language and introduces a breakdown including direct, personal utility, market network, platform, two-sided marketplace, asymptoting marketplace, and bandwagon effects. Product-market fit and liquidity as prerequisites (Priority: 4/5): The discussion stresses that network effects do not work without strong product-market fit and, for marketplaces, sufficient liquidity on both sides of the market. Why some networks win and others fail (Priority: 5/5): Examples like MySpace, Facebook, WhatsApp, Path, and Airbnb illustrate how real names, utility, timing, and competition determine whether a network effect becomes durable. Virality versus network effects (Priority: 4/5): The speakers distinguish virality, which helps acquire users, from network effects, which help retain users and create defensibility; the two often coexist but are not the same. Bandwagon and social adoption dynamics (Priority: 3/5): The episode highlights popularity-driven adoption effects, such as Slack versus HipChat, and argues that social momentum can precede and reinforce true network effects.

Key Arguments: Network effects were first explicitly identified in AT&T’s 1907 annual report, showing that the concept predates modern software and began with telecom hardware. Metcalfe’s law became a useful shorthand for the increasing value of connected users, but it is descriptive rather than a strict law. In software, many old defensibilities disappeared with the removal of distance and time, making network effects especially valuable as a moat. Not all marketplaces inherently have network effects; they need sufficient supply-demand liquidity before the effect meaningfully compounds. Personal utility network effects are stronger and longer-lasting when the product helps users accomplish real tasks, not just socialize. Real-name identity and utility were critical to Facebook’s success and were missing in MySpace, which felt more like entertainment than infrastructure. WhatsApp succeeded by delivering clear utility, especially for international communication where SMS was expensive, and by keeping the product simple. Virality brings users in, while network effects keep them; founders and investors should analyze both separately. Bandwagon effects can accelerate adoption through social signaling, but they are distinct from durable product-based network effects. The taxonomy matters because different network-effect types call for different growth playbooks, feature priorities, and investment judgments.

Data Points: Year network effects first used in AT&T report: 1907 - The term was first used in an AT&T annual report describing how adding customers made the network more valuable. Year academics began studying network effects: 1974 - The speakers note that formal academic attention began decades after AT&T’s early usage. Period of Microsoft DOJ scrutiny: late 1990s - Microsoft’s OS and application ecosystem was cited as a classic network-effect case that drew antitrust attention. Number of major defensibility types discussed: 4 - Brand, scale, embedding, and network effects were presented as the main remaining moats in software. Number of network-effect types identified: 9 - Currier says the team has identified nine distinct types of network effects. Network-effect type clusters: 4 - The taxonomy includes direct effects, personal utility effects, two-sided effects, and bandwagon effects. Airbnb early growth period: 3 years - Airbnb’s marketplace struggled for roughly three years before liquidity made growth accelerate. WhatsApp signup price: $1 per year - Mentioned as the low-cost model during early adoption. WhatsApp communication cost advantage: free - WhatsApp was attractive to international users compared with expensive SMS and international calling. Facebook product timeline reference: year 3 - Mark Zuckerberg reportedly changed the homepage to emphasize Facebook as a social utility around year three.

Pivotal Quotes: "network effects are really the only one that's native to what we're doing, and therefore, it's the most powerful." — James Currier: Explaining why network effects are especially valuable in software compared with older forms of defensibility. "A network effect is essentially for defensibility and retention. ... the viral effect is about getting new users and the network effect is about keeping those users." — Anu Hariharan: Clarifying the distinction between virality and network effects. "Facebook is not valuable if you're the only user on the platform." — Sonal / panel discussion: Illustrating how direct network effects depend on other users to create value.

Implications: Founders should diagnose the exact type of network effect they are building, because each requires different product and growth choices. Investors should separate hype, virality, and true defensibility when evaluating companies. The taxonomy helps identify which businesses can become durable category winners.