Arrow icon
Network Effects: How They Work, What to Build, and Measuring Them
Arrow icon
Network Effects: How They Work, What to Build, and Measuring Them

Network Effects: How They Work, What to Build, and Measuring Them

Entrepreneurship
July 29, 2021
/
13 Jogs
Why it matters?

Network effects are a tricky thing to get right, but they're the holy grail of technology companies. They work in all sorts of different ways and have many different types, so this blog post is going to take you through what network effects are and how they work - as well as how you can build them into your company's product or service.

The idea of network effects is very simple. Every time someone joins your network, they are increasing the value of the whole system for everyone else in it. For example, the more people who use Facebook to connect with friends and family, the better it is for all of those other users. Or the more riders who use Uber, the more demand there is for drivers who in turn provide more supply, and shorter wait times, for riders thus increasing the user experience.

This blog post will explore how to build and measure them, as well as best practices of what to do to create a company with network effects from scratch.

The Content

Each link contains a summary produced by one of Joggo's geniuses so you can decide where to spend your time learning more

/
/
Lenny's Newsletter

Community Wisdom: Transitioning from IC to PM manager, visualizing your product flows, negotiating salary, effici…

Read more

The Summary

Transitioning from IC to manager of PMs

  • Might become less fun because you might spend all your time managing up/down/horizontal vs. actual product work
  • Might happen organically as the work scope is increased to manage multiple products

Visualizing your product flows

  • Tools
    • OmniGraffle - awesome flowcharting tools
    • Whimsical - fast to use
    • Figma - good for flows
    • Miro - high fidelity + flows
    • Lucidchart
  • Start with a vague wireframe for each screen first, then customized some of the screens to match up the UI
  • Screenshots instead of actual mockups/wireframes will save you time if you're just trying to explain data flows

Value of self-serve product demos

  • Always have either a demo space or preset examples
  • Focus on high intent customers and rework our onboarding to include elements from the demo

Negotiate salary

  • Think deeply about what are you truly willing to commit to
  • Nuance to one point
    • Add at least 10% to that amount when you approach your manager
    • The approach for the salary convo with your boss/skip should be: "I really do want to stay, but I can't ignore this offer of $X that I just got"

Conducting efficient customer interviews

  • Efficient interviews are not necessarily fast
  • Let go of the reins and be _with _the person you are interviewing
/
Packy McCormick
/
Not Boring

Compounding Crazy

Read more

The Summary

The past month there have been crazy events occurring around the world that make it feel like we are in a sci-fi universe. People are going space, and valuations are consistently in the billions. In another decade things today will seem normal and the new things will be crazier.

Die Progress Units and the Law of Accelerating Returns

  • Human progress is exponential 
  • The law of accelerating returns suggests that the rate of change of progress accelerates because humans can use technology at their disposal to progress faster than previous generations could without technology
  • We often think of history as straight line which is an incorrect assumption
  • The trajectory of a very recent history often tells a distorted story
  • Our own experience makes old men stubborn about the future
  • Human knowledge compounds

The Venture Funding Rorsach and $320T Global Markets

  • More VCs are investing in startups than ever
  • All of the unicorn startups in the world are now worth more than the total market cap of Apple
  • All of the world’s equities have 100x since 1970s in market cap

The Compounding Colossus

  • Every invention is apart of the latest chain of compounding inventions
  • AI and the future of scaling will be interesting
    • Starlink may allow everyone in the world to get wifi for $5 a month
  • Compounding happens extremely quickly
/
/
Stratechery

Olympic Ratings, Google Earnings, YouTube and Brand Advertising (Stratechery Daily Update 8-2-2021)

Read more

The Summary

NBC struggles to lift Olympic ratings

  • Viewership has declined by 42% compared to the 2016 Olympics.
    • This is partly due to the lack of in-person audience due to COVID and partly because of the time difference between the US and Japan.
    • Part of a larger trend of declining viewership in broadcast television.
  • NBC will still make a profit, though likely a very small one.
  • NBC still plans to cover the Olympics indefinitely, because they allow NBC to showcase the "strength of our platform."

Google earnings see huge increases in the past two years

  • Google's parent company, Alphabet Inc. reported an increase of 62% this year. Last year Google was up by 55%.
    • The secular shift to digital advertising has worked in Google's favor, especially during COVID.
    • Google Shopping played a huge role in increasing their profits.
  • Google, like Facebook, is now focusing on conversions, not just advertising.
    • Google is selling ads using Machine Learning and AI, giving it the ability to work around iOS 14 changes.

YouTube benefits from TV viewership decline

  • YouTube advertising revenues were up by 84% this year.
  • YouTube has found through surveys that advertisers who shift a portion from TV to YouTube will have more success reaching their target audiences and will likely profit more.
/
Gigi Levy-Weiss, James Currier, Pete Flint, Josh Elman, Chris Anderson
/
NfX

Viral Effects Are Not Network Effects

Read more

The Summary

Virality and Network Effects are Different

  • Viral effects = growth of new users
  • Network effects = adding value and defensibility to your product
  • Can exist without each other and the playbook to build each is very different

Building Virality Into Products

  • Build the product around language, the rails on which products are shared
  • Get a user to feel the emotional content of your product (e.g. what will the user feel about themselves when they share?)

Building Network Effects Into Products

  • Create value for customers by getting other customers to put value in
  • Retention builds defensibility

Complementary Relationship Between Viral Effects and Network Effects

  • Network effects approach helps you achieve virality given its easier to include viral language hooks (e.g. “share”, “get”, and “see”)
  • Virality helps you achieve network effects by growing the network more quickly

Future of Virality and Network Effects

  • Hard to go viral today: must be high quality, shocking, or deliver 10x value
  • Network effects never go out of style
  • If you have to choose, choose network effects
/
/
NfX

An Epiphany about Network Effects

Read more

The Summary

Tickle was one of the first viral, user-generated content sites, but had near-zero retention.

Tickle began to build products that had network effects

  • Dating marketplace
  • Tickle Social Network
  • Grapevine: an advertising marketplace for ads on user-generated content sites

Despite network effects, these businesses weren't successful

  • Potentially due to lack of skill or running five different businesses inside one company

In 2004 Monster, a two-sided labor marketplace, bought Tickle

  • Monster was suffering from poor products, poor customer services, poor strategic decision making, and a lack of foresight.
  • Ultimately they succeeded because network effects can be all that matter
/
James Currier
/
NfX

What Are Network Effects?

Read more

The Summary

The phenomena where every new user of a products adds value to the product for other users

  • Think of telephones: the value increases when a new number is added because all the existing customers can call that number and the new user can call any of the existing numbers.
  • Where network effects apply, value grows exponentially with each user.  Network effects are particularly important because they are one of the four main ways to get ahead of your competitors. The three other defensibilities are: 
  • Scale
    • Large companies can generally offer more competitive, bulk pricing
  • Embedding
    • When companies embed themselves in users' lives, it can become costly to stop using that product and switch to another product
  • Brand
    • Users build habits around brands they trust Ultimately, the marginal benefit of increasing any of these three defensibilities begins to decrease and reach a plateau. Because network effects increase exponentially, they will not reach a plateau and are, therefore, the most effective defensibility.
December 13, 2018
/
D'Arcy Coolican, Li Jin
/
a16z

The Dynamics of Network Effects

Read more

The Summary

All of the most successful technology companies have relied on network effects. As platforms with network effects get bigger, they get better. Recently, network effects companies have been splitting markets despite big moats. Network effects are as dynamic as ever. 

Not all products are created equal

  1. Ridesharing 
    1. Driver supply and passenger demand reinforce each other 
      1. More drivers means lower wait times 
      2. More passenger demand means more drivers want to drive 
    2. Once there are enough drivers to satisfy less than 5 minute wait times, passengers are indifferent to increasing network effects
  2. Social lending 
    1. Changes in network effects can be driven by absolute liquidity 
    2. If Frank was lending his money to friends, more friends meant more demand 
    3. Once a group had more than 7 friends, they became less likely to lend or borrow since people only have approximately 7 friends they trust
  3. Social networks 
    1. Adding more people to Facebook turned it from social network into social media 
    2. Adding additional nodes in context of media discovery was less valuable as more people joined the app 

Users and inventory

  1. Type of incremental user 
    1. Not all members of a given network are equal 
      1. Popular restaurants add more value to OpenTable than a restaurant nobody wants to go to 
    2. Making sure you incentivize the users you want is key 
  2. Commoditized vs differentiated supply 
    1. Platforms with more differentiated inventory have stronger network effects 
    2. The more differentiated inventory, the more the platform needs to a good job of curation and matching

Competition

  1. Network overlap 
    1. Think about which competitors have a network overlap and could potentially move into your market 
      1. DoorDash and Uber Eats 
  2. Switching costs 
    1. Low switching costs can also lower network effects 
  3. Multi-tenanting to meet demand 
    1. Network effects are weakened when users are unable to use a single platform to accomplish their goal 
/
James Currier
/
NfX

Reid Hoffman’s Playbook for Growth and Network Effects

Read more

The Summary

Network effect businesses are hard to execute

Building Innovative Business Models

Business model should minimize two growth limiters 1. Product market fit 2. Operational scalability Maximize four key growth factors 1. Market size 2. Distribution 3. High growth margins 4. Network effects

Network Effects in LinkedIn

  • Leveraged direct and two sided network effects
  • Direct: more users make the network more valuable
  • Two sided: more users attract more corporate employers which increases the value of the network

Distribution

  • A good product with great distribution beats a great product with poor distribution
  • Network effects produce and require aggressive growth
July 29, 2019
/
D'Arcy Coolican
/
a16z

Hidden Networks: Network Effects That Don’t Look Like Network Effects

Read more

The Summary

Hidden networks, networks that don’t appear to have network effects but do, have an advantage over their competitors. Types of hidden networks:

Slow networks

  • These have a longer time horizon until the network reaches a critical mass, when it then grows quickly.
  • These often have long product loops or insufficient saturation.

Unfinished networks

  • The product feature or a strategic decision leaves the network temporarily incomplete in some material way.
  • The emergent behavior of users is primarily on the incomplete side of the network.

Throttled network

  • The product feature, or a strategic decision, limits the size or engagement of the network.
  • There are positive or neutral effects, when one or more of these constraints are relaxed.

Latent network

  • Companies build a network before they build the actual product or tool
  • The networks users are engaged primarily with just the central node While hidden networks have their own challenges, being "hidden" can be an advantage to an entrepreneur hoping to reach critical mass before their competitors.
/
/
NfX

The Network Effects Bible

Read more

The Summary

Why Network Effects Are Important

  • Every new user makes the product/service/experience more valuable to every other user.
  • Network effects are the best form of defensibility, and thus value creation, in the digital world.

Nodes and Links

  • Nodes are network participants: consumers, devices, customers, buyers, sellers, brokers, etc.
  • Central nodes: nodes with a high number of links, often more valuable. 
  • Marginal nodes: relatively few links and less value
  • Network size: the total number of nodes in a network.
  • Links: connections between nodes or groups of nodes. 

Network Density

  • The density of a network is its ratio of links to nodes. The higher the density, the more powerful network effects are.

Directionality

  • The direction of a link between nodes in a network is determined by which way the interaction flows.

One-to-One vs One-to-Many

  • One-to-many connections: unidirectional.
  • One-to-one relationships: usually reciprocal.

Clustering

  • Nodes tend to not be dispersed evenly.
  • Bridge: when two clusters are connected by only one link.
  • Higher degrees of clustering leads to more powerful network effects.

Critical Mass

  • This is when the value produced by the network exceeds the value of the product.

Metcalfe’s Law

  • The value of a communications network grows in proportion to the square of the number of users on the network.
  • It holds because the number of links between nodes on a network increase at a rate of N^2.

Reed's Law

  • Reed suggests using 2^N to account for clustering.

Homogeneous vs. Heterogeneous Networks

  • Homogeneous networks: all the nodes have the same function in the network.
  • Heterogeneous networks: there are two or more classes of nodes categorized by both function and utility. 

Asymptotic Network Effects

  • A network is Asymptotic when network effects have diminishing returns.

Same-Side Network Effects

  • These occur on the same side of a multi-sided network.

Cross-Side Network Effects

  • Cross-side effects are direct network effects from complementary goods or services in a network with more than one side.

Indirect Network Effects

  • These occur when the value of a network increases as a result of one type of node benefitting another type of node directly, but not directly benefiting other types of nodes.

Negative Network Effects

  • Negative network effects are usually network congestion (increased usage) and network pollution (increased size).

Multiplayer vs. Single-Player Mode

  • Single-player products: help the user alone, and can be used without other users.
  • Multiplayer products: use the presence and impact of the other users in the product.

Switching Costs

  • These are the costs in time, effort, or money of switching from one product to another.

Chicken or Egg Problem (Cold Start Problem)

  • This is the problem of initially reaching critical mass to trigger a positive feedback loop.

Multi-Tenanting

  • Multi-tenanting: when there are low costs to participate in competing networks at the same time.

Disintermediation

  • After initially connecting through a market network product, users transact directly off the product. 
January 9, 2018
/
James Currier
/
NfX

The Network Effects Manual: 13 Different Network Effects

Read more

The Summary

Why are Facebook, Uber, Twitter and Microsoft some of the most successful companies in the world? What is behind their success?

They all share one property: network effects

  • Network effects is one of four remaining defensibilities, the others being brand, embedding, and scale.
  • 70 percent of value in tech is driven by network effects, and there have been 13 distinct types of network effects identified so far.

These 13 network effect types can be grouped into five broader categories:

Direct Network Effects

  1. Physical
  2. Protocol
  3. Personal Utility
  4. Personal
  5. Market Networks

2-Sided Network Effects

  1. Marketplace
  2. Platform
  3. Asymptotic Marketplace

Data Network Effects, Tech Performance Effects, "Social" Network Effects

  1. Language
  2. Belief
  3. Bandwagon
/
James Currier
/
NfX

Foundations for Mastering Network Effects

Read more

The Summary

Network effects have become instrumental to value creation in a networked economy, so NFX has summarized key articles under three main headings (Network Science, Network Effects, and Defensibility) to help people master how they work.

Network Science

  • Definition: the academic study of complex networked systems and their real-world properties.

The Small-World Problem

  • The first empirical study of personal direct networks that verified the existence of the “small-world phenomenon,” popularized as six degrees of separation.

The Strength of Weak Ties

  • Weak ties or acquaintances connect nodes in different network cliques, reduce the overall average degree of separation, and are crucial to the cohesion of a network.

Social Capital in the Creation of Human Capital

  • Social capital arises from a mix of network structure and its strength of ties which is important to the formation of norms and trustworthiness within a network.

Collective Dynamics of ‘Small-World’ Networks

  • The “small-world network” thesis shows a larger network can still be fairly cohesive (without risking network pollution) if a few “shot-cut” connections exist.

Emergence of Scaling in Random Networks

  • One of the first to use real-world data to evaluate network topology, it shows that a small number of “rich” nodes accrue the most links via preferential attachment.

The Structure and Function of Complex Networks

  • An all-inclusive resource, it examines the foundations of network science and how network properties are statistically studied with real-world implications.

Four Degrees of Separation

  • A 2012 Facebook study observed 3.74 degrees of separation on its network which shows how network effects can be applied to technology companies.

Three and a Half Degrees of Separation

  • A 2016 Facebook study observed fewer degrees of separation on its network which shows the increasingly networked nature of the world due to the internet.

Network Science

  • A foundational resource on network science as well as graph science, preferential attachment, and other fundamentals to study and apply to companies.

The Square and the Tower

  • An evolution of social and political networks and hierarchies throughout history, it explains network science in an accessible way using real-world context.

Network Effects

  • Network effects have become the most important source of value creation in a networked economy.

Increasing Returns and the New World of Business

  • A look at diminishing returns markets and increasing returns markets, it explains how to actively manage increasing returns to be successful.

A Theory of Interdependent Demand for a Communication Service

  • An examination of the “dynamic demand" for communications services and “demand-side economies of scale," it features how to navigate pricing models.

Platform Competition in Two-Sided Markets

  • A pricing structure formula for 2-sided markets, it is based on an understanding of network effects to account for the complexities of multi-sided networks.

Network Externalities (Effects)

  • An overview of the early literature of network effects, it breaks down network effects into their basic elements and looks at network externalities versus network effects.

Metcalfe’s Law is Wrong

  • An arguably failed attempt to disprove Metcalfe’s Law, it suggests an alternative formula for valuing network effects using logarithmic rates as network scales.

Metcalfe’s Law Recurses Down the Long Tail of Social Networking

  • Metcalfe’s revision of his own Law, it debunks counterarguments and further refines some areas to strengthen his Law and discuss newer phenomena.

Metcalfe’s Law After 40 Years of the Ethernet

  • A follow-up, more formalized rebuttal, it proves the validity of his Law using real-world data and anecdotal pieces to show how network effects help sell vision.

Data Network Effects in SaaS Enabled Marketplaces

  • Software-enabled marketplaces have four key advantages that allow them to collect a high volume of valuable data and get a data network effect going.

A16z Podcast: Not All Network Effects are Created Equal

  • A recap of the history, evolution, and taxonomy of network effects, it provides valuable insight.

Strategies for Two-Sided Markets

  • An analysis of two-sided markets, it shows how to optimally navigate the complexity of multi-sidedness and handle supply and demand dynamics.

Required Reading for Marketplace Startups: The 20 Best Essays

  • An anthology resource, it features real-world, applied approaches to build marketplaces.

Pipes, Platforms, and the New Rules of Strategy

  • A framework for understanding a platform ecosystem and its four types of players, it makes it clear how dominant network effects are as digital-age defensibility.

Other Defensibilities

  • Network effects are the most important defensibility since they are native to the internet.

Defensibility Creates the Most Value for Founders

  • A look at why defensibilities create the most value and competitive advantages can only get you so far, it discusses core concepts and strategies to implement.

Economies of the Moat and the Castle

  • An interesting early formulation of the defensibility theory by Warren Buffet.

How Competitive Forces Shape Strategy

  • Also known as the “Five Forces Model,” it outlines competitive forces that shape industries and gives insight into company defensibility and profitability.

All Revenue is Not Created Equal

  • A look at how companies are valued against their revenue, it explains how to understand how investors value companies and what draws their interest.

Reinforcement: The Hidden Key to Building Iconic Tech Companies

  • The more defensibilities you add to your company, the more powerful all your other defensibilities become, and core network effects are easiest to reinforce.

The Dentist Office Software Story

  • A fictional anecdote, it shows it is lethal to not have a defensibility strategy.

Amazon Go and the Future

  • A case study of Amazon, it looks at the mechanics of scale effects, explains tech companies’ unique scale advantages, and why scale is an effective defensibility.

The Scale of Tech Winners

  • A discussion of modern tech giants, it explains how scale enabled them to build secondary competitive advantage.

Beyond Network Effects: Digging Moats in Non-Networked Products

  • A discussion of the other three defensibilities (scale, brand, and embedding), it also breaks down three reasons for why scale makes companies more defensible.
December 13, 2018
/
Li Jin, D'Arcy Coolican
/
a16z

16 Ways to Measure Network Effects

Read more

The Summary

Network effects are crucial to building a new business. While many people think a company either has network effects or they don’t, it is more complicated than that because network effects grow and change over time.   There are 5 main categories to measure network effects. In each category is a list of questions that can help a company determine if their company possesses network effects.  Acquistion-related metrics  1. What percentage of new users are organic?  2. How much traffic is generated internally vs. externally?  3. How much do you need to spend to acquire customers  Competitor related metrics  1. How many of your users use/ are active on similar services?  2. How easy/much value can a new user gain from joining the network?  Engagement-related metrics  1. Is user retention increasing for new users?   2. Is retention increasing for newer users using a core service?  3. Are newer cohorts retaining better on a dollar basis?  4. Are participants in the oldest market better retained than those in newer market?  5. Are users becoming more engaged over time?  Marketplace metrics  1. How successfully can the two sides of the marketplace find each other?  2. Is there enough supply and does it fit the users’ needs?  3. How long does it take for supply and demand to match?  4. How concentrated is the marketplace?  Economic metrics  1. How much are you able to charge/ how much are customers willing to pay?  2. What do the unit economics look like?