The Network Effects Bible

www.nfx.com @nfxguild

Roadmap Part I - Why Network Effects Are Important Part II - How Networks Work Part III - Network Properties Part IV - Building and Maintaining Network Effects Part V - Related Concepts 3

Part I Why Network Effects Are Important

70% of the value created in technology since 1994 has been driven by network effects. Still, too few Founders understand network effects well, nor know how to design them into products. 5

There are 4 main defensibilities in the digital world: Scale, Embedding, Brand, and Network Effects. Of these four, network effects are the most powerful. 6

The 13 Network Effects (and counting…) 8) Asymptotic Marketplace (Uber, Lyft) 1) Physical (landline telephones) 9) Data (Waze, Yelp!) 2) Protocol (ethernet) 10) Tech Performance (BitTorrent) 3) Personal Utility (iMessage, WhatsApp) 11) Language (Google, Xerox) 4) Personal (Facebook) 12) Belief (currencies, religions) 5) Market Network (HoneyBook, AngelList) 13) Bandwagon (Slack, Apple) 6) Marketplace (eBay, Craigslist) 7) Platform (2-sided) (Windows, iOS) 7

Network Effects Case Studies The NFX map applied to category-leading companies

We don’t think most people really understand Facebook yet. It is the most powerful network effects company we’ve ever seen, with no less than six of the thirteen known network effects currently working in its favor. On top of that, Facebook has fortified itself with all three other defensibilities: brand, scale, and embedding. Network effects have defined Facebook throughout their history. 10 years ago we wrote that Google should buy Facebook at any cost because of their powerful network effects — and time has more than vindicated the argument. More than any other factor, Network Effects predict the future of Facebook. Read the full case study here. 9

Trulia revolutionized the US real estate market by successfully creating an online marketplace for real estate. Trulia’s Founder & former CEO Pete Flint (currently a managing partner at NFX) says Trulia built three distinct types of network effects into the core of its business — data, marketplace, and market network — in order to achieve success, leading to a massive $3.5 billion exit and an industry-shaping merger. Read the full case study here. 10

Most people think of Uber as a true 2-Sided Marketplace. As shown in the NFX Map, true 2-Sided Marketplaces have some of the most powerful network effects and therefore strong defensibility. But, in reality, Uber is an asymptotic marketplace which has much weaker network effects. For this and other reasons, Uber is more vulnerable than most realize. Read the full case study here. 11

Part II How Networks Work

Networks are any set of nodes and links Nodes are network participants. Not all nodes are equal in function and importance. Central nodes have a high relative number of links, and marginal nodes have fewer links. Network size can be measured by the total number of nodes in a network, and degree is a measure of the number of links each node has. Links are the connections between nodes in a network. Links vary in terms of directionality, the strength of the connection, and the frequency of activity. Network density is the proportion of links to nodes within a network. Read more about Nodes and Links 13

Network density is the ratio of its links to nodes. The higher the ratio, the greater the density of the network. Network Typically, the higher the density of the network, the more powerful its network effects are. Density Founders should build products to promote higher network density. Within a network, density is usually distributed unevenly. At the beginning, Founders should typically focus on the “white-hot center” of your network, where the density and activity are highest, and let the network grow from there. Read more about Network Density 14

Directionality A link between nodes can be either directed or undirected. Directed links denote one-way relationships. When the interaction between one node and another isn’t reciprocated, it’s a directed link. Undirected links are two-way, reciprocal relationships. The interaction between two nodes on a network with undirected links flows both ways. Read more about Directionality 15

Relationships between nodes in a network can be one-to-one, or they can be one-to-many. One-to-One vs. YouTube is an example of a network with one-to-many links (accounts with large One-to-Many followings that they don’t follow back). Read more about One-to-One Facebook has one-to-one connections at its foundation. vs. One-to-Many 16

Clustering Within real-world networks, nodes are unlikely to be dispersed evenly. They tend to cluster or form local groupings that are more tightly knit than the network at large. If you look at how you use services like Slack or Facebook Messenger, where you form subgroupings that are more tightly-knit and active than the broader network, you can see some examples of clustering. A “clustering coefficient” measures the degree of clustering in a network. Often, the higher the clustering coefficient, the faster the value of the network increases as usage grows. Read more about Clustering 17

Critical Mass The critical mass of a network refers to the point at which the value produced by the network exceeds the value of the product itself and of competing products. Most products with network effects must ultimately reach critical mass in order to fully take advantage of the defensibility provided by their network effects. Before the size of the network reaches critical mass, the product remains quite vulnerable and may not have much value to users. Read more about Critical Mass 18

These laws describe the rate at which network value increases as they scale. The Network Sarnoff’s Law describes broadcast networks, which increase in value in proportion to N (with N being the number of nodes). “Laws” Metcalfe’s Law describes networks without clustering, which grow in value at a geometric rate (N^2). Read more about the Network Laws Reed’s Law applies to networks with clusters, which increase in value at an exponential rate (2^N) 19

Part III Network Properties

Irregularity Real-world networks tend to have clusters, hot spots, and dead spots. This is because complex systems have irregularities. The nature and configuration of nodes and links can vary widely throughout a single network. For Founders, it’s important to recognize these irregularities and go after the densest, most active “white-hot center” of the network. Read more about Irregularity 21

Many network effect businesses require users to create a profile that’s visible to other nodes in Real Identity, the network. Networks with profiles tied to a node’s real identity, like your real personal name or real company name, are typically more effective at building network effects than networks with Pseudonymity, pseudonymous profiles (e.g. user-generated handles like “Tiger123”) or completely anonymous networks. & Anonymity Read more about network identity 22

Asymmetry This term relates mostly to marketplaces, whether that marketplace is 1, 2, 3, or N-sided. In nearly every marketplace, one side - or one type of node - is harder to acquire than the other. In a “demand-side marketplace,” the demand side (buyers) are harder to acquire. In a “supply-side marketplace,” the sellers are harder. Look for these various asymmetries in a marketplace and prioritize which types of demand or supply will be the best to attract first, second, and third. Then focus on developing tactics to crack the highest-value target first. Read more about network asymmetry 23

Homogeneous networks are networks where all the nodes have the same function in the Homogeneous vs network. One user is interchangeable with the next in the basic function they perform. In a landline telephone network, for example, each node (telephone) performs basically the same Heterogeneous function as any other. Networks Heterogeneous networks are networks where there are two or more classes of nodes categorized Read more about Homogeneous by both function and utility. Buyer nodes on eBay, for example, are on the network for vs. Heterogeneous Networks fundamentally different reasons from seller nodes. 24

Asymptotic Network Effects Asymptotic network effects are network effects with diminishing returns. Recall the basic definition of network effects: as usage of a product grows, its value to each user also grows. In some cases, however, network effects can start to weaken after certain point in the growth of the network. Growth in an asymptotic network, after a certain size, no longer benefits the existing users. Examples include Lyft/Uber and most data network effects. Read more about Asymptotic Network Effects 25

Same-side network effects are direct network effects that occur on the same side of a multi-sided Same-Side (2-sided or N-sided) network. A platform like Microsoft’s OS, for instance, has a same-side network effect because Microsoft users directly benefit from an increase in other same-side users Network Effects (on the demand side of the platform) since they can interchange Word, Excel, and other files with a greater number of people. This makes both the platform itself and applications developed on Read more about Same-Side the platform more valuable for users. Network Effects 26

Cross-Side Network Effects Cross-side network effects are direct network effects that arise from complementary goods or services in a network with more than one side. As opposed to indirect network effects, cross-side network effects refers specifically to the direct increase in value to users on one side of a network by the addition of users to another side. More supply is better for the demand. More demand is better for the Supply. Read more about cross-side network effects 27

Indirect network effects occur when the value of a network increases as a result of one type of Indirect node benefitting another type of node directly, but not directly benefiting the other nodes of its same type. Network Effects Same-side nodes indirectly benefit each other because they create an increased incentive for complementary users on the other side of the network to use the network, which in turn benefits Read more about Indirect Network Effects all the nodes on the same side. 28

Negative Network Effects In some situations, more network usage or greater network size can actually decrease the value of the network, leading to negative network effects. Negative network effects can happen in two ways. First, network congestion, where the network becomes less valuable as a result of increased usage-- as with traffic in a road network. Second, network pollution, where the network grows too large and is polluted with unwanted content -- like your grandmother commenting on your party photos on Facebook. Read more about negative network effects 29

Part IV Building and Maintaining Network Effects

“Multiplayer mode” and “single-player mode” are terms that came from the gaming industry, but are Multiplayer vs useful for discussing network effects. Single-Player With single-player mode products, you get the full value of the product no matter who else is using it. For example, filing taxes via TurboTax. With multiplayer mode products, some of the value of Mode using a product (and often the majority of it) comes from other users. A multi-player product lets you feel and benefit from the activity of other users. Read more about Multiplayer vs. Single-Player Mode Products can have both single-player and multiplayer value. Watching YouTube videos is a single-player experience. YouTube, with view count and comments, is also a multiplayer experience. 31

Switching costs refer to the costs in time, effort, or money that arise when you switch Switching from using one product to another incompatible product. When switching costs are high, it tends to create customer lock-in because the customer has more of an incentive to stick Costs with the same supplier throughout their life cycle. Any product that has a lot of defensibility will have high switching costs. Read more about Switching Costs 32

Chicken or Egg Problem (Cold-Start Problem) When you first launch a two sided network like a marketplace or a platform, there is often no value to one side of the network until the other side is there too. So which comes first? There are at least 19 tactics to help solve the chicken or the egg problem. Several of them involve creating a single-player mode product where one side (either the supply side or the demand side, depending on the type of marketplace) can derive value from the software without a network in place. Others involve attracting one side of the marketplace with compensation, such as cash or leads. Read more about the Chicken or Egg Problem 33

Multi-tenanting occurs when there are low costs or no costs to simultaneously Multi- participating in competing networks at the same time. When the two sides of a marketplace switch between services to get to the other side of the marketplace at no cost, Tenanting that’s multi-tenanting. For example, both riders and passengers frequently switch between Uber and Lyft to find rides/passengers. Read more about Multi-Tenanting 34

Disintermediation Disintermediation is a vulnerability that mostly applies to Marketplaces and Market Networks. It happens when, after initially connecting through a marketplace or market network product, users take future transactions off the product and transact directly, without paying a fee to the marketplace. For example, you might hire someone to build IKEA furniture using Taskrabbit, and then exchange numbers so that you can hire them for future jobs directly. This is a significant problem for marketplaces and market networks because repeat purchase is critical to most transactional networks. To help prevent disintermediation, you can provide tools, reputation, insurance, compliance, leads and other incentives. Read more about Disintermediation 35

Retention is about how often your users return to use your product. This can make a big Retention difference in how powerful the network effects of a product are. Recall that network effects happen when increased usage of a network leads to greater value. Read more about Retention Usage can increase in one of two ways: either new users join the network, or existing users increase their usage. Any drop off in usage from existing users will weigh against the growth of the network from new users. That’s why retention is so important for building network effects. 36

Part V Related Concepts

Geometric Growth vs. Linear Growth Businesses with neither viral effects nor network effects tend to produce linear, straight-line growth. Linear growth can produce a business, but not the kind of high impact, exciting growth we’re all looking for. We all want the green-line growth curve, but it is not a network effect. It’s a growth curve. Founders should push themselves and their teams to reach geometric growth. Network effect businesses, once they hit the tipping point, typically show this type of geometric growth. Read more about Geometric vs. Linear Growth 38

Viral effects are different than network effects, but many people confuse the two because they are Viral Effects both positive feedback loops, and because they were experienced together in famous companies like Facebook, Twitter, and WhatsApp over the last 15 years. & Virality Viral effects are about growth -- when existing users bring you more new users for free. Read more about viral Network effects are about defensibility: a product gets more valuable when more people use it. A effects & virality product or service with viral effects has a “viral coefficient”, which is the number of new users that join as a result of each user. New Users New Users 39

“Platform Business Model” This is a recent term that is commonly used to talk about ALL companies with network effects. In our opinion, it’s too generic and conflates too many concepts together. We think that one network effect is a “2-sided Not all network effects companies are platform network effect.” We prefer to break platforms - platform out the 13 different network effects because we nfx are just one of 13 think it produces more a precise and revealing network effect types analysis. To raise money today, it’s popular to say “make your product a platform.” But we think this refers to a specific type of approach where other independent companies are building their own products on your product. Read more about the platform vs. network effect business model 40

Reinforcement An important thing to know about network effects is that once you have one going, it’s much easier to build all the other defensibilities on top of it, including other network effects. This “reinforcement effect” is often underappreciated by Founders, especially when they’re preoccupied with growth. Continually looking for and launching new defensibilities is important because they reinforce each other. The more you have going the better, as long as you don’t lose operational focus in the process. Read more about Reinforcement 41

Scale Effects Scale effects are easy to confuse with network effects because they both become stronger in relation to the number of users. Network effects cause products to become more valuable to users as the network grows, increasing defensibility by increasing product value. Scale effects, on the other hand, lead to lower per-unit production costs as the fixed costs of production are spread out among a greater number of units — meaning a higher profit margin which can increase defensibility against competitors with less favorable margins. Read more about Scale Effects 42

Brand defensibility is different from network effects. Like scale, the strength of your brand is correlated Brand with the growth and usage of your product, so it’s easy to confuse brand with network effects. However, they work very differently. Brand arises when people know who you are and what you do. A well-established brand identity comes with psychological switching costs. People are less likely to switch to an unknown or lesser-known brand Read more about Brand from yours because psychologically they will default toward what’s familiar. 3743

Embedding is an effective defensibility strategy that involves integrating your product directly Embedding into a customer’s operations so the customer can’t rip you out and replace you with a competitor without incurring significant cost in terms of time, energy, or both. Embedding can work with network effects to make your business more defensible, but they are Read more about Embedding separate concepts. Business models based on embedding include Oracle and WorkDay. 3744

NFX essays bring to light what most never see about network effects, growth, and more Subscribe here

www.nfx.com @nfxguild