Review

This is could be the best product book ever written. The author shares a unified theory of Network Effects; in doing so he gives us a common language to discuss them. Many product books make the mistake of going too broad, but this is a glorious deep dive into the mechanics of networked products.

This isn’t an empirical theory, instead it’s grounded in story telling. The author uses a lens of Network Effects to explain the strategies behind the valley’s most famous products and companies.

Despite having a narrow focus, this book taught me more about strategy, competition and product metrics than any other.

Please pass go and buy the book.

You Might Also Like…

Crossing the Chasm

This is a product management classic. The key insight is that you have to think about your Go To Market Plan differently depending on where you are in the technology adoption lifecycle. The hardest moment is crossing the chasm from early adopters to the early majority. You’ll likely have to change your approach to do it. It’s a great reminder that strategy should always be contextual. It’s worth reading Obviously Awesome by April Dunford too, as she has plenty to say about the positioning formula referenced in this book.

andrewclark.co.uk

The Innovator’s Dilemma

This book is a product management classic. A great framework for how disruptive technologies change market dynamics. It’s full of great insight for challengers and incumbents. The book is clearly well researched. The conclusions are both surprising and straight forward. A must for anyone trying to innovate and break into a market. Well worth the time.

andrewclark.co.uk

Book Review and Summary: Product-Led Growth by Wes Bush

Get a comprehensive review and summary of the best-selling book "Product-Led Growth" by Wes Bush, the product management and growth expert. Learn about the key principles and practices for driving growth through a product-led approach and maximizing the value of your product to customers

andrewclark.co.uk

Key Takeaways

The 20% that gave me 80% of the value.

• A Network Effect: describes what happens when products get more valuable as people use them.
• The Network → refers to the people who use the product to interact with each other
• The Effect → how the value increases as more people start using the product
• Metcalfe’s law doesn’t apply to most networked products → it doesn’t take into account the ‘cold start problem’
• Five Important Network Effect Concepts
1. The Cold Start Problem
2. The Tipping Point
3. Escape Velocity
4. The Ceiling
5. The Moat
• The Cold Start Problem - Networked products often suffer from anti-network effects in their early stages. Small networks want to self-destruct.
• The Cold Start Problem describes the vicious cycle where new users churn because they're aren't enough users yet. Also known as the chicken and egg problem.
• An Atomic Network is defined as the smallest network where there are enough people that everyone will stick around. Y
• If you can build the first atomic network → you can build a second adjacent to it → then you can get to 10 or 100 networks → soon you’ll have a huge interconnected network that spans the entire market
• Network size isn’t the most important thing Speed, Quality, Breadth, Stability and Density are all important
• Density is key → the interconnectedness of people on the network is key. You need to find the right people to start you Atomic Network
• Networks can be multi-sided (e.g. Creators or Consumers). There is a minority of users that create disproportionate value and as a result, have disproportionate power. This is the hard side of the network. Focus on attracting the hard side first. Have a hypothesis about how the product will cater to the hard side users from day one
• Magic Formula: Setup initial supply → bring demand → focus on supply, supply, supply
• Be simple to use, and easy to describe
• Network products love to be free - charging creates friction
• New technologies create mini-reset moments → opportunities to take on big companies
• Magic Moments: It's obvious when a product has solved the cold start problem - the experience starts to really work
• To consistently ensure that people don't experience zeros, the network needs to be built out substantially and needs to be active
• The Tipping Point - Once you discover a repeatable strategy to network building → you can execute until you tip over to the entire market
• Invite-Only: Allows you to control network quality and density by handpicking the initial network. You can capitalise on FOMO and Acquisition network effects. The connected bring more connected resulting in strong early engagement
• Come for the tool and stay for the network. Often tools are valuable for individuals, but more valuable with networks. It works because it's easier to spread a tool than a network. Utility then Network.
• If you have a chicken and egg problem, buy the chicken. Coca-Cola invented free drink coupons for retailers → grocers were the hard side of the network, once stocked, the product had a change (distribution is often the hard bit).
• Build your first atomic network without subsidising, get some product market fit
• Shared economic upside (is a great way to grow a network)
• Flintstoning → artificially propping up the hard side of the network with human effort. Manually fill in critical parts of the network, until it can stand on it's own (or you can automate it). Related strategy = first party content (Nintendo creating Mario and Zelda for the Switch).
• Escape Velocity - When products see success and start to scale
• The goal is to maintain fast growth by amplifying network effects AND get to a sustainable revenue generating business model
• The 'Network Effect' is actually three forces

Acquisition Effect	customers are acquired more easily as more people join (as the product uses its network to acquire customers) (viral growth, low CAC)
Engagement Effect	utility increases as more people join (density increases retention + usage)
Economic Effect	the business gets better as more people join (accelerate monetisation, reduce costs, better business models)

‣
The Engagement Effect: Utility increases as more people use the networked product
• Make products stickier over time and with more usage
• Social: Responses (payoff) to posts grow with connections and network size
• Marketplace: Seller more likely to sell a listed item if there are more buyers
• Collaboration: Share a project - coworkers engage and close the loop.
• If a network is too sparse, the loop is broken, the user doesn't get a payoff, they churn
• Users need to trust the loop to rely on it.
• in the negative, they churn and the network shrinks
• in the positive, they stay and the network grows, gets stickier, and denser
• The Growth Accounting Equation: Gain or Loss in active users = New + Reactivated - Churned
• Cohort Retention Curves - how many are still around 1 / 7 / 30 days after they sign up
• Rules of thumb for retention: 1 day: 60%, 7 days 30%, 30 days 15%
• Sometimes the curve smiles, as retention and engagement goes up over time as people reactivate. Smile curves are really rare, invest in those startups.
• Segment users by value to you, convert low value to high by getting them to try features
• LTV, Revenue, Frequency, Engagement, Use Cases
• Try to understand how needs and motivations are different. What would it take to move them into a higher group? what's the right lever?
• Escape velocity phase is about accelerating engagement loops. Make each stage of the loop perform better. Reduce friction of the action, increase the likelihood of a favourable response.
‣
Acquisition Effect: The ability for a network to attract new customers as it scales. This is the most magical and explosive forces.
• The power is in the ‘Product-Network’ duo. The product has the features to attract the user to the network, the network brings more value to the product.
• Viral Loop: Hear → Sign up → Find value → Shares product → They Sign Up (Repeats)
• it can be measured, tracked and optimised to be made more effective
• The ratio between each loop 1000 to 500, 500 to 250 is the viral factor
• 0.5 → each cohort generates 0.5 of the next
• starting with 1000, a viral factor of 0.5, leads to a total of 2000 users by the end of the amplification
• meaning an amplification rate of 2x
• Viral factors have a massive effect (0.5 = 2x, 0.6 = 2.5x, 0.7 = 3.3x, 1 = 20x)
• Strong retention has the biggest effect on the viral ratio
• Networks built through viral growth are healthier than those launched in the big bang fashion (like Google +) → they have low density and low engagement
‣
The Economic Effect: How a business model (inc. profitability and unit economics) improves over time as a network grows
• Overtime Bureaus tend to combine into larger ones → more data from more merchants and customers helps every merchant with better information
• buy content → win a niche audience → fund your own content
• Large networks with data advantages can personalise offers and subsidies
• Premium features can be designed in a way such that they're more useful as the network gets larger
• Strong economic effects allow you to maintain premium pricing → switching costs become higher for participants

• The Ceiling - As a product reaches scale, the growth curve teeters between expansion and contraction. Negative forces appear during the late stage of a networks cycle.

Market Saturation	Regulatory action	Degradation of marketing channels
Churn from early adopters	Later mainstream users dilute quality of initial communities	Network Revolts - hard side becomes more concentrated
Bad behaviour: trolls, spam, fraud	Crowding - and degrading user experience (discovery, noise)

1. Features can raise the ceiling - but for only so long
2. A product can saturate it’s market or it’s niche.
3. Marketing channels become less effective over time (banner ads and email marketing)
4. Power becomes more concentrated on the hard side of the network. It becomes harder to keep everyone happy. A well organised revolt can kill a product completely. Often the 80/20 rule apples, so you're going to have imbalance in the importance of contributors. As a network scales, the hard side will professionalise.
5. As you reach the mainstream audience, more and more people arrive diluting the quality. Discovery becomes harder, and you get overcrowding
6. As the network gets more dense over time, its network effects become incrementally less powerful
• The 100th connection for any given participant is likely less impactful than the first few
7. To fight the forces, you have to evolve your product, market and feature set
• New Adjacent Networks: Figure out the adjacent set of users whose experience is subpar. continually evolve the offering to attract the next set of hard side users to your platform. Uber started to think about signing up people who didn't already have a car.
• New Formats: Ebay → adding 'buy it now' and stores
• New Geographies: harder than adjacent networks
• New Products: hard in existing companies → acquiring companies is the cheat code
8. The Law of Shitty Clickthroughs: Every marketing channel degrades over time. What worked before eventually stops scaling as fast as you need it to. People become skilled at ignoring advertising. Starts off highly efficient - pay back periods creep up over time. Embrace new marketing ideas early!
9. Tapping into the acquisition network effect - It's more efficient for networked-products to optimise viral loops vs traditional marketing spend. You can't buy 1 billion users, you need to have a viral loop.
10. Scale attracts bad behaviour: When successful networks grow large audiences they attract spam. Context collapse it what happens when too many networks are simultaneously brought together, and they collapse into one - on social networks, it inhibits the behaviour of content creators. Leverage networks themselves to flag bad behaviour and remove bad actors. Create features that nudge interactions in the right direction.
11. One hypothesis on why social networks lose heat at scale is that the 'old money' can't be cleared out, and new money loses the incentive to play the game
12. Data network effects are often invoked as a path for networks to solve relevance and overcrowding issues that emerge over time
• The Moat - the competitive advantage of any given company, and the durability of that advantage. For Networked Products... the moat is the effort, time and capital it would take a competitor to replicate the product and network.
• If your product has network effects your competitors likely have them too. Effective strategy is about who scales and leverages their network effects in the best way possible. Smaller players often upend larger ones.
• When markets mature, competition becomes zero sum
• New players therefore can't just do the same thing. New entrants have to either
• provide a much better experience
• or a differentiated experience
• Networked products lean toward 'winner takes all'
• If a company can win a series of networks faster than its competition, it develops an accumulating advantage

	Goliath	David
Goal	Fighting market saturation and growth slow down	Solving the cold start problem
Strategy	· add new use cases · introduce new audiences · generate profit	· starts with a niche · doesn't have to worry about profitability · focusing on top line growth
State	More resources, man power and existing products	Fewer resources, capital, employees and distribution
Play	· slower execution, risk aversion, strategy tax (new products have to align to existing business) · large companies introduce processes which slow down entrepreneurial risk taking	· but they have speed, and no sacred cows · trying & failing many times is normal · they might try different niches · discover 1 atomic market they'll get investment and resources to support them

• Cherry Picking - Each startup needs just one atomic network, yet each incumbent has to defend all of its networks. This is the asymmetry of network-based competition. Large networks made up of many communities, often leave some communities underserved.
• Network density beats total size. AirBnB could quickly create a dense network within a city. They would quickly have more listings than craigslist in a given city (together with a superior product).
• Upstarts benefit from cherrypicking because the incumbent has conveniently aggregated the network. large networks can't defend every inch of their product.
• Big Bang Failures - Wide launches create many weak networks which aren't stable on their own. Bottom-up networks are more likely to be densely interconnected, healthier and more engaged. Big bang launches also stop you doing things that don't scale
• Paradox of small markets - they seem to small to be noteworthy, until an airbed company ends up disrupting the hotel industry
• Compete over the hard side - Microsoft won developers

Deep Summary

Longer form notes, typically condensed, reworded and de-duplicated.

Part 1: Network Effects

• To present a unified theory of network effects and establish a common language to describe them across domains… and to increase the chance you’ll be able to leverage them by making better product decisions
• Technology is easier to build and more reliable than ever
• Open source · Cloud platforms · Targeted advertising · SaaS tooling · Appstore distribution
• Despite that, it's harder to grow and defend a product
• Constrained attention (zero-sum) · Fierce competition · Marketing channels are ineffective · Network-based competition · Unclear future application platforms
• Network effects are one of the only protective barriers in an industry with fierce competition
• Network gets more valuable with growth · Others face the cold start problem · You can weaponise your network

Defining a Network Effect

• A Network Effect: describes what happens when products get more valuable as people use them.
• The Network → refers to the people who use the product to interact with each other
• It’s the connection that matters and creates the value
• The Effect → how the value increases as more people start using the product
• E.G. Youtube becomes more compelling as videos, creators and viewers grow
• Many products that have more than 1B users have strong network effects

A telephone without a connection is useless. Its value depends on the connection with other telephones, and increases with the number of connections 1900 AT&T Annual Report

• Networked products are often at the heart of Network Effects. The product is ‘the Uber App’, the network is the ‘riders and drivers’
• How can you tell if a product has a network effect?
• Does it connect people with each other (for commerce, collaboration, communication) at the core of the experience?
• Does the ability to attract new users, or to become stickier, or to monetise, become even stronger as its network grows larger?
• Does the user face a Cold Start Problem where retention is low when there's no other users?

Dotcom Boom → Created New Vocabulary

• As millions of people began to use the internet in 1995 there was a wave of excitement and IPOs (Yahoo, Netscape, Ebay, Amazon, Priceline). The NASDAQ rose 400%. By 1996 there were 20 million internet users.
• A new vocabulary emerged (‘winner-take-all’, ‘first mover advantage’) to describe the common belief that the first networks to reach scale in a given domain would be unstoppable. Investors believed they would provide more value to users, buyout the competition and dominate the industry. AOL was valued at $224B at its peak. The ‘Dot-com boom’ became the ‘Dot-com bubble’ but that vocabulary lives on.

Things didn’t play out like people expected

• Turns out there’s little advantage to being first and the winner usually doesn't take all.
• Miss-application of Metcalfe's law helped justify big valuations in the boom.
‣
The systemic value of compatible communicating devices grows as a square of their number.
• Each time a user joins an app with a network behind it, the value of the app is increase to n^2
• If a network doubles users (from 100 to 200) it's value will quadruple (from 10 to 40)
• Defining the value of a network as a mathematical function based on the number of connected devices
1. BUT Metcalfe’s law was about communications devices, not websites and market places. It doesn’t take into account
• The very beginnings of the network - when there’s little value
• The quality of the user experience
• Some networks are multi-sided
• The difference between active users and inactive users
• Crowding out - if there are too many users

Meerkat’s Laws > Metcalfe’s Law

• Meerkats benefit from living together (in groups of 30-50) by coordinating hunting and spotting predators

• The ‘Allee threshold’ is the tipping point when the group dynamics become beneficial
• Populations under the Allee threshold quickly go to zero.
• Populations above the Allee threshold quickly grow. This is the ecological network effect.
• Population growth can't last forever though, the natural environment only has so many resources (called the carry capacity).

Humans are the networked species. Networks allow us to cooperate when we would otherwise go it alone. And networks allocate the fruits of our collaboration. Money is a network. Religion is a network. A corporation is a network. Roads are a network. Electricity is a network. Naval

Brief Intro to the 5 Stages of ‘Cold Start Theory’

• Cold Start theory is a network effects framework. 5 stages of network growth that cover creation, scaling and defence. Highlighting the challenges and goals of each stage.

The 5 stages of Network Growth

1. The Cold Start Problem
2. Tipping Point
3. Escape Velocity
4. Hitting the Ceiling
5. The Moat

1) The Cold Start Problem

• Most new networks fail, due to a self-reinforcing destructive loop. When users don't find who or what they want, they churn. These anti-network effects are more common than positive network effects.
• Overcoming the cold start problem requires getting conditions right (the right users, the right content, the right time) which is difficult to execute in a launch
• Key Levers:
• Build a single atomic network (the smallest possible stable network that can grow on its own)
• The product idea is at the heart of every network effect
• Identifying the first most important users to get on a nascent network going
• Seeding the initial network so it grows in the way that you want

2) The Tipping Point

• Building beyond the atomic network. Build many more networks to expand into a market
• each new network tips faster and faster - the entire market is more easily captured
• tipping over a row of dominos - unstoppable momentum, following a small win
• Network effects grow..
• city by city
• company by company
• campus by campus
• team to team (SaaS - lands and expands)

3) Escape Velocity

• Furiously working to strengthen the 3 network effects and sustain growth
• the acquisition effect - lets products tap into the network to drive low-cost highly efficient user acquisition
• the engagement effect - increases interaction between users as networks fill in
• the economic effect - improves monetisation levels and conversion rates as the network grows
• Understand how the forces work - accelerate systems that power them
• Acquisition effect powered by
• viral growth → referrals (Paypal), recommendations (linkedIn)
• a good first user experience
• Engagement effect
• increased engagement as the network grows
• move users up the engagement ladder (incentives, marketing, comms, new features)
• Uber got 'airport trip' drivers to do more than airport runs
• Economic effect
• increasing conversions in key monetisation flows and ramping up revenue per user, as the network grows
• convert paying to free
• upsell
• Together these are a flywheel that can power networks into the billions of users

4) Hitting the ceiling

• Rapidly growing networks, both want to grow and tear themselves apart
• there are enormous forces in both directions
• Networks hit the ceiling when growth stalls. Due to things like...
• customer acquisition costs (rise @ saturation)
• viral growth slows
• law of shitty clickthroughs
• drives down performance of acquisition and engagement loops over time
• people tune out of stale marketing channels
• Fraudsters, over crowding, context collapses
• spam and trolls
• Products grow, and hit a ceiling, until the product team addresses problems - then get another growth spurt
• solutions are difficult
• sometimes networks are destroyed
• Some problems can't be solved, only managed (trolls and spam)

5) The Moat

• Using network effects as a moat to fend off competitors
• Other moats: product, brand, partnerships, technology
• It’s hard to use network effects against competition who are in the same industry
• you end up in network-based competition
• Airbnb fought off Wimdu by competing on the quality of the network, scaling its network effects (not by using pricing or features)
• Competition ends up being asymmetrical, while leveraging the same forces
• large and a small network in any given market have different strategies
• David Strategy - pick off niche segments within a larger network, build atomic networks that are highly defensible with key product features, and, when applicable, better economics and engagement
• Goliath Strategy - use larger size to drive higher monetisation and value for top users, and fast-following any niches that seem to be growing quickly

Unconnected Insights

Non-Core Insights

Part II · The Cold Start Problem

• Networked products often suffer from anti-network effects in their early stages. Small networks want to self-destruct.
• The Cold Start Problem describes the vicious cycle where new users churn because they're aren't enough users yet. Also known as the chicken and egg problem.
• Often products get a spike in new users → but the novelty quickly wears off
• If you don't solve the cold start problem, you will fail.
• The Atomic Network is the answer to the cold start problem.
• An Atomic Network is defined as the smallest network where there are enough people that everyone will stick around.
• If you can build the first atomic network → you can build a second adjacent to it → then you can get to 10 or 100 networks → soon you’ll have a huge interconnected network that spans the entire market
• Each one becomes easier, because each network can become intertwined with the next
• Clearly tangible IRL networks like Uber should be built out city by city BUT successful digital networked products often do the same (Tinder, Facebook, Slack).
• Often networks start small: a single city, campus, or group beta users.
• Slack → One team in the company → Organic growth then captures the company →
• When launching a network you need to quickly create sufficient density and breadth to break through early anti-network effects and grow on it's own
• Enough to meaningfully improve the user experience through connections
• Then there’s a gradual series of improvements in core metrics as the network fills in
• At Slack the key test was…. How likely is it that the colleague you want to message is registered and going to reply?
• Speed, Quality, Breadth, Stability and Density are all important
• Density is key → the interconnectedness of people on the network is key

• The Question: How can you add a small group of the right people? at the same time? That use the product in the right way?
• The Atomic Network → the smallest, stable network from which all other networks can be built
• Magic Moments → When the cold start problem is solved, a product is able to consistently create 'magic moments' - users open the product and find a network that is built out, meaning they can generally find or do what they want. The network effects kick in, and the market hits its tipping point as users start coming to you
• Tools for building atomic networks, some are counter intuitive
• Launch in the simplest possible form - not fully featured (dead simple value proposition)
• Target the smallest possible network - and density - ignore the objection of 'market size'
• Do whatever it takes attitude to executing the launch - even if unscalable or unprofitable - just get some momentum, don't worry about how to scale
‣
Growth Hacks that can help form your Atomic Network
• Slack → invite only launch
• Paypal → $5 referral fee
• Dropbox → Video demo on hacker news (drove many pre-registrations)
• Uber → Ice Cream OnDemand (drove free PR)
• Successful network products can look like toys that won’t serve mainstream needs
• BUT they can grow out and capture the market
• They are easily underestimated - at launch they undershoot the needs of the many (the first telephone only worked over a distance of a mile)
• In the beginning most people aren't in the target audience (so that’s OK)
• People underestimate how rapidly technology and infrastructure can improve and how quickly it can be adopted (non-linear network effects)
• Pick a small and targeted starting point
• the network needs to fill out to the point where the people and content are relevant to you
• Have a hypothesis about what your atomic network might look like
• Probably smaller and more specific than you think
• Think tiny - and at a specific moment in time
• The language of launching new networks, should be hyper local
• focusing on groupings of a handful of users with the right situation, at the right time
• Uber didn't start with NY → it started with 5PM at the Caltrain Station at 5th and King St
• The more users you need for your atomic network the harder it is to create
• Communications products have small minimum size requirements
• these are some of the stickiest fastest growing products too
• BUT what's easy for you, will be easy for your competitors too
• Enterprise products (like workday) are harder to get going
• require that the company implement them before there is any value
• viral growth is difficult - company wide coordination is required
• top-down enterprise sales work better
• Growing city by city, campus by campus or team by team is a powerful strategy → it leads to dense, organic connections of users that strengthen network effects across multiple dimensions
• engagement goes up - users are more likely to find other relevant users
• viral growth goes up - when users see their friends and colleagues are using the service
• Don't spread your effort too widely

The Hard Side of the Network

• Networks can be multi-sided (e.g. Creators or Consumers)
• There is a minority of users that create disproportionate value and as a result, have disproportionate power. This is the hard side of the network.
• They do more work. Contribute more to your network, but are harder to acquire and retain. Hard sides exist because there are tasks in networked products that require more work
• Social Networks → Content Creators
• App Stores → Developers
• Marketplaces → sellers or providers
• Job platforms → people hiring
• Generally one side of the network is easier to attract → therefore you should focus on attracting the hard side first. On the hard side often a small people that do the majority of the work in the community.
• for UBER that's drivers (5% of total users). Riders are numerous but engage less frequently and deeply.
• To attract the hard side, you need to solve a hard problem → design a product that is compelling to a key subset of your network
• Tinder did this for attractive people in its network
• Networks can be one-sided, it’s called ‘cross-side network effects’ when more users on one side of the network benefit the other side of the network
• Wikipedia: 500M visitors to 4000 editors
• Uber: 100M riders to 3M drivers
• Youtube - 2B viewers - 1M creators
• You need to understand, acquire and retain those on the hard side of the network
• Study and understand the motivations of the contributors → at wikipedia it’s about status, teamwork, mission,
• Hard side users have complex workflows, expect status benefits as well as financial outcomes and will try competitive products to compare
• their expectations are higher, and its difficult to engage and retain them
• BUT the hard side creates a ton more value consumers are typically cheaper and easier to engage
• Have a hypothesis about how the product will cater to the hard side users from day one
• Who is the hard side?
• How will they use the product?
• What is the unique value proposition to the hard side?
• How do they first hear about the app? and in what context?
• As the network grows, why will they come back more frequently and become more engaged?
• What makes them sticky to your network such that when a new network emerges, they will retain on your product?

On Social Networks

• On social 20% of influencers drive the majority of engagement

‣

Evan Spiegel → A Pyramid of internet technology or communications technology

• Social feedback is key. If a creator creates and posts something and gets no feedback, they're likely going to be disappointed.

• The Hard Side is Usually the Supply Side
• Supply side = provide time, products and effort - and are trying to generate income
• Marketplaces tend to revolve around their sellers → 20% of Ubers drivers do 60% of the trips
• Focus on appealing to the hard side of your atomic network from day one with your product.
• To solve a cold start problem for market places you need to bring a critical mass of suppliers on board
• Magic Formula: Setup initial supply → bring demand → focus on supply, supply, supply
• Initial supply might be easy to attract with subsidies - eventually it will become a bottleneck. The hard side of the network is harder to scale
• How do you find a problem, where the hard side of a network is engaged, but their needs are undressed? The answer is look at hobbies and side hustles
• Smart motivated early adopters who are finding opportunities to make themselves useful.
• What people do on nights and weekends represents under-utilised time and energy of the world
• Build an army with under-utilised time or under-utilised assets.
• Uber → under-utilisation of cars
• AirBnB → spare bedroom
• Ebay → stuff
• Often the hard side continue to use the platforms because of the positive network effects
• Segment the hard side and ask who is being underserved (sometimes this is a niche)
• Start with underserved segments - users may not be very attractive customers on their own
• New products often disrupt markets by starting on the low end, providing good enough functionality, and growing from there into the medium and eventually into the core market of the incumbents
• A recent trend is to start at the top of the market as a luxury product and work your way down (Uber and Superhuman (email))
• Combining disruption theory and network effects. Atomic networks often start at the low end in terms of functionality in a niche market.
• Establish an atomic network.
• Get the hard side of the network to expand their offerings to go into the next vertical
• This attracts an incrementally higher-end opposite side, which spurs the hard side to extend even further
• Airbnb: airbeds → rooms → apartments
• Network effects create momentum for a low-end atomic network to slowly build out into higher-end offerings over time.
• You need to get the value proposition for both the hard and the easy side of the network right. But the hard side is probably more important.

• Networked products often do one thing well

Networked products vs everything else

Simple products with clear value propositions are easy to describe - which makes them easier to spread from person to person

• be really basic, you take one or two main actions
• Snapchat - send photos to friends
• Dropbox - magical folder that syncs your files
• Uber - hit a button and get a ride
• Youtube - watch videos
• Be simple to use, and easy to describe
• Don’t worry about technology differentiation or defensibility.
• Create viral easy-to-use products works in both consumer and enterprise landscape.
• Dropbox → envisioned as consumer → became popular in enterprise → changed strategy

The ideal product to drive network effects combines both...

• Simple as possible product - easily understandable by anyone as soon as they encounter it
• Bring together a rich, complex, infinite network of users that is hard for competitors to copy
• Zoom is a good example

Networked products love to be free

• Freemium models (ad supported, pay to unlock, micro-transactions)
• Social networks + communication apps → free
• SaaS products → freemium
• Marketplace → free to browse (transaction cost)

• Charging creates friction for new users!
• It's hard enough to build a network, without adding friction
• Paying can slow or stop virality
• Short-term pain for longterm gain

Technology and Behaviour Shifts Create opportunity

• Video conferencing → Professional work + broadband + remote work
• New technologies → enable new behaviours (Tinder → swiping)
• New technologies create mini-reset moments → opportunities to take on big companies
• Cloud → Allowed G-Suite to take on Office
• Everyone needs to start over

Magic Moments

• It's obvious when a product has solved the cold start problem - the experience starts to really work
• Network fills out, people are connected in the right way, experience really starts to shine
• This is the magic moment when the product can deliver its core value
• a product that hasn't solved the Cold Start Problem won't deliver any magic
• The opposite of the magic moment:
• You can also measure the opposite of the magic moments →
• UBER called them zeros (Riders can’t find drivers, or drivers can’t find riders)
• No value given. Users will bounce and possibly never come back
• It's obvious when there are not enough users to make something engaging
• A zero is a terrible experience - its not easy to solve though
• To consistently ensure that people don't experience zeros, the network needs to be built out substantially and needs to be active
• people on the network have to be responsive
• Zeros cause churn

Part III · The Tipping Point

• You need to scale beyond your atomic network to really benefit from network effects
• When viral growth, stickiness and strong monetisation become repeatable, the network will hit the tipping point

• Once you discover a repeatable strategy to network building → you can execute until you tip over to the entire market
• Tipping Point Strategies
• Invite only - suck in a large network through viral growth
• Come for the tool and stay for the network
• Paying up for launch
• ‘Flint-stoning' - submitting links and content manually, before adding automation and community features for scale

Invite Mechanics

• Turning people away creates hype (but FOMO is not the biggest factor of invite only) - its the ability to control network density that’s key
‣
Examples (LinkedIn, Gmail, Facebook)
‣
Linkedin
• Top tier people (like Bill Gates) didn’t need early LinkedIn. People who are still building and hustling needed it
• People invited people they knew from their contact book
• This creates a denser network
• The network therefore wasn’t diluted by geography, industry or demographic
• True believers were highly engaged, and helped the product spread all over the world
• Tipped the market within a few weeks
‣
Facebook
• Required harvard.edu email address
• Defined an atomic network, everyone trusted each other, school-by-school launches could be controlled
‣
Gmail
• 1000 external invites, each could send 2 invites on
• storage was 100x what you could get elsewhere
• invite limitation was born of necessity, Google couldn't scale it quickly
• invitations became hot property - that was unintentional
• invites were sold for $150 on eBay
• gmail email addresses were a cool thing to have
• Limiting audience growth also allows you to fix the product before scaling.
• Invite-only mechanics provide a better welcome experience (Everyone is already connected to at least one person).
• Slack used corporate email addresses to determine what network you should join
• The most connected people tend to be invited earlier, and in turn they tend to invite other highly connected people. The connected bring more connected resulting in strong early engagement
• Often positioned as 'find friends', early users have larger contact lists and invite people with larger lists too
• LinkedIn refined invites over time, from ‘import contacts’ to "People you may know" (ML powered)
• Invites help fill out networks, build density and new users get a better experience
• Scarcity dynamics (Hype and Exclusivity)
• people like to share when they're included + people without an invite will ask for one
• Creating attention and engagement.
• Early users get valuable @handles
• Risks of invite only
• limits topline growth
• requires throw away functionality/code
• turning people away isn't a great experience
• you might limit your network too much and amplify the cold start problem
• There are huge benefits for networked products
• early network can gel as a community
• develop high density of connections
• and grow organically via virality

Curating a high-quality network

• Handpick and curate the initial network → with the right people (Uber hand picked initial drivers)
• High touch onboarding → teach them how to use the product
• Enforce the rules

Start with a deliberate point of view on who's best for your network - will define it's magnetism, culture and ultimate trajectory

Come for the Tool and Stay for the Network

• Come for the tool and stay for the network is a famous strategy for launching and scaling networks
• Start with a great tool
• Then get the users to to do things that tap into a network
• collaborate, share, communicate or otherwise interact with users

‣

Instagram: Filters were the tool → social network kept people engaged

• Burbn (later instagram) pivoted to photo sharing, liking and commenting from something that was initially more like Forsquare.
• Instagram had networks from day one
• profiles, feeds, friend requests, invitations (many social features), discovery feed, free, filters
• 100k downloads in first week. Initial engagement was for photo-editing, not social engagement.
• 2.2 million users uploaded 3.6 million new photos a week
• Celebrities showed up on instagram. Influencers would join to create content.
• 18 months after launch, Facebook acquired instagram for $1bn
• Now, 82% of photos are posted without use of a filter
• network effects had taken over the utility of the initial tool
• Now has >1bn users and generates $20b revenue

How great tools help tip entire markets

• Chris Dixon 2015 Essay "Come for the tool, stay for the network"
• Attract users with a single player tool
• Then get them to participate in a network
• Tool gets you to critical mass
• Network creates long term value for users, and stability for the makers
• Other examples
• GSuite - comments, chat and video
• Minecraft + roblox
• Yelp
• LinkedIn (put your resume online)

Come for the tool circumvents the cold start problem

• easier to launch into an entire network
• minimises the size of the atomic network

The tool props up the value of the network when the network is small.

• The tool acts as a backstop in the value of the product
• Making it useful even when nobody is on it

• Eventually a pivot has to happen
• this can be changing what you emphasise in the app
• notification emphasis
• recommendation prominence
• The tool-to-network shift is a specialist strategy
• many products have no single player mode (communication and marketplace tools)
• Tinder
• Slack
• WhatsApp
• eBay
• Deliveroo

• Underlying patterns for tools and networks
• Tools are often around content editing and hosting
• Networks: allowing people to interact with the content by extension

• Often tools are valuable for individuals, but more valuable with networks
• Pivoting users from tool to network can be hard
• sometimes only a small % transition, they have to change their behaviour, not every feature can be a social network
• A tight coupling of the tool and the network matters → users need to want the network piece, and drive the product towards it
• It works because it's easier to spread a tool than a network. Utility then Network.

Cost of growing a network

• Is this ever going to be profitable? Is a common criticism of startups (Uber)
• standing up a network can be expensive
• If often makes sense to spend widely to establish a network → the goal is to get the market to hit the tipping point, driving toward strong positive network effects, and then pull back the subsidies
• hoping you're left with a fast growing, monetisation product

Coupons

• Invented by Coca-Cola → a free glass, redeemable at any dispenser, 8.5m free drinks given
• Helps penetrate grocery stores by solving the ‘chicken an egg’ product stocking issue
• Retailers get the retail price, customers get something for free, retailers have to stock the product to take advantage, if they don't their competitors will
• grocers were the hard side of the network, once stocked, the product had a change (distribution is often the hard bit)
• coupons for condensed milk achieved 97% distribution
• bootstrapping the hard side of a multi-side network
• Uber guaranteed drivers $30 an hour to bootstrap supply
• IF YOU HAVE A CHICKEN AND EGG PROBLEM, BUY THE CHICKEN
• Uber burned cash too quickly using this model, so they monitored and challenged city operations teams to move to commission based models as quickly as possible
• Then you use that network to grow
• with referral bonuses → Give $200 and get $200 when a friend signs up to drive
• and word of mouth
• Then do more jobs
• surge pricing
• do 10 trips and get an extra $1 per trip

Financial Levers for growth

• Subsidise the hard side of the network → e.g Guarantee payments (twitch did to creators)
• B2B Freemium model helps get you usage
• Build your first atomic network without subsidising, get some product market fit
• Once you can reliably launch new networks, use financial levers to accelerate growth and the speed at which you reach the tipping point
• Use to encourage more people, or to encourage people to spend more time on the network
• Bitcoin - shared economic upside (is a great way to grow a network)
• particularly effective for influencers, creators and developers
• if the network grows and succeeds its individual participants also win
• Microsoft - partnerships with larger companies
• you build them something in exchange for distribution or revenue
• IBM paid for DOS → giving Microsoft distribution
• but Microsoft retained the ability to sell DOS to others
• so if IBM were successful, they'd be in a great position
• IBMs hardware side of the business was competed away
• Microsoft retained software dominance as it had a critical mass of users, developers and software
• Why unprofitability is sometimes smart
• In many cases positive unit economics in early network growth isn't possible. Or it's too slow
• Once you have a few atomic networks, you may want to buy your way to an entire market before you work out profitability
• What looks like unprofitability in the short term might lead to dominance in the longterm if the market reaches a tipping point in your favour

Flintstoning

• Flinstoning → artificially propping up the hard side of the network with human effort
• Named after the leg powered Flinstone car
• People do things initially → then they automate it once the volume grows
• Manually fill in critical parts of the network, until it can stand on it's own
• imagine a 3 sided network, users, business and platform. You could fake the platform and focus on building the network
• Enables earlier feedback, and you might not build features that you don't need
• Reddit founders created dozens of dummy accounts to post links and make the homepage look complete else - the community might have dried up. They automated the fake users, by scraping news sites for popular stories
• Paypal created a bot to buy and sell items on Ebay, but it insisted on using paypal only - that grew the user base (convincing Ebay sellers to sign up)
• Related strategy = first party content (Nintendo creating Mario and Zelda for the Switch)
• Cyborg startups → combine humans with a team of software engineers
• hybrid models are smart (more robust than an early algorithm, more scalable than people alone)
• You must phase out flint-stoning over time. You need a plan to switch from manual to automated.
• Combine with ‘come for the tool’ , THEN prop up the hard side of the network.
• you need a plan to switch from single player to multi-player

16 Always be Hustlin'

• The architects of Ubers’ growth were the local operations teams. Travis: ‘Product can solve problems, but it's slow. Ops can do it fast’.
• Uber was an ops led company.
• Creativity is key to growing and reaching the tipping point
• Twitter launching at SXSW.
• AirBnB bootstrapping on events (make $1000 renting during Octoberfest)
• Creativity works when cities are autonomous and decentralised
‣
Uber created a city launch playbook
• Rider Zero (celebrity in a new town)
• Uber Puppies & Kittens
• Uber Ice Cream
• Uber Mariachi
• Targeted drivers of limo companies
• Push notifications
• Referral programs aligned around local events
• Uber Moto, Helicopter, boat (Apps could add new ride types)
1. Each new market became easier and easier

Hustle and creativity help tip over markets, each atomic network is different

• GMs were CEOs of their cities (creativity and decentralisation were key)
‣
Uber 1.0 Values
• Make magic
• superpumped
• inside out
• be an owner, not a renter
• optimistic leadership
• be yourself
• big bold bets
• customer obsession
• always be hustling
• let builders build
• winning: champions mindset
• principled confrontation
• meritocracy and toe-stepping
• celebrate cities

Part IV · Escape Velocity

Introducing Escape Velocity

• When products see success and start to scale, the product begins to hockey stick.
• The challenge is twofold:
• Maintaining fast growth and amplifying successful network effects
• To continue scaling the business whilst building a revenue generating business
• It takes a huge amount of energy and effort to scale. It doesn't just happen when you reach escape velocity. You can't coast.
• play defence - to counteract market saturation, competition, spam, and other forces
• play offence - building up network effects to amplify their strengths over time
• The 'Network Effect' is actually three forces

Acquisition Effect	customers are acquired more easily as more people join (as the product uses its network to acquire customers) (viral growth, low CAC)
Engagement Effect	utility increases as more people join (density increases retention + usage)
Economic Effect	the business gets better as more people join (accelerate monetisation, reduce costs, better business models)

• The Growth Accounting Equation
• Engagement, Acquisition and Economic Effects map to the key outputs that product teams care most about... active users and revenue.
• Active users → sign ups, engagement and retention
• Revenue → due to active users and in increase in average revenue per user
• Growth rate → the ability to repeatedly scale these network effects consistently over time
• Growth Accounting Equation:
• Gain or Loss in active users = New + Reactivated - Churned
• This months actives = last months actives + gain or loss
• Active users is relevant for social or messaging apps. Could be active subs for a SaaS product.
• Active users * ARPU (average revenue per user)
• Density of the network…
• More new users appear through viral growth
• Product becomes stickier - less churn
• Conversion and revenue per user increase

The Engagement Effect

• 1753 - British Navy looking into Scurvy divided the ship into cohorts to test the effectiveness of fruit on fighting it.
• Cohort Retention Curves - how many are still around 1 / 7 / 30 days after they sign up
• Retention is the most critical metric. Ugly data though.
• Users don't stick to their apps. 25% abandon after one use.
• 70% aren't active the next day
• 96% aren't active over 90 days
• Where the curve levels determines how many people are retained
• the average app has a curve that falls whittling itself to zero
• the brutal result for most apps is failure
• of the millions of apps, only a few have large audiences
• people spend 80% of their time in 3 apps
• Investment Retention Rules of Thumb

Days	Retained
1	60%
7	30%
30	15%

• Hopefully the curve levels out by and after 30 days.

Sometimes the curve smiles, as retention and engagement goes up over time as people reactivate. Smile curves are really rare, invest in those startups.

How New Use Cases Drive More Engagement

• Engagement effect can raise retention by layering on use cases
• Slack - more channels as adoption increases, once everyone is on there you can poll etc
• turns infrequent and noncommittal usage into daily usage
• Segment users by value to you, convert low value to high by getting them to try features
• Revenue
• LTV
• Frequency
• Engagement
• Use cases
• LinkedIn segmented users by Active 7/7 days last week, 6/7 days last week etc
• try to understand how needs and motivations are different. What would it take to move them into a higher group? what's the right lever?
• levers are different for early users and power users
• You can do cohort analysis to confirm your hunches and drive strategy
• When a user connects with 5 people in the first week, they're more likely to become a high value user later on
• How to get users to take the actions that are more likely to make them higher value?
• education
• promote new features
• free subscription if they complete certain actions
• Dropbox - users who instal across multiple devices are more valuable than those on a single device → even better if they share folders and collaborate with other users. Dropbox could offer more storage if you performed certain actions.

Engagement Loops

• Making products stickier over time - describes how users derive value from the network
• Social: Responses (payoff) to posts grow with connections and network size
• Marketplace: Seller more likely to sell a listed item if there are more buyers
• Collaboration: Share a project - coworkers engage and close the loop.
• If a network is too sparse, the loop is broken, the user doesn't get a payoff, they churn
• Users need to trust the loop to rely on it.
• in the negative, they churn and the network shrinks
• in the positive, they stay and the network grows, gets stickier, and denser
• Escape velocity phase is about accelerating these loops
• make each stage of the loop perform better
• Reduce friction of the action
• Increase the likelihood of a favourable response
• this should give you a number of ideas and experiments to try

Reactivation

• The engagement network effect can help reactivate churned users - growing active user count
• Typically only 20-25% of signed up users are 30 day active. 75% of users are inactive at any point.
• Reactivation is a the counter weight to churn. Traditional products struggle with reactivation, but it's easier for networked products.
• Reactivate users by enlisting active users to bring them back
• being notified of people interacting with you (sharing, liking, commenting)
• Churned users are 'dark nodes'
• even if inactive for months they can flipped back to an active
• tipping an inactive user to an active user
• The bigger the network, the more likely an infrequent user continues to get reengaged and over time that might make all the difference
• What is the experience of the churned user?
• What notifications are you getting from other users?
• Are they compelling enough to bring you back?
• Try sending a weekly digest
• Make it as easy as possible to reactivate (password recovery)
• Reengagement can be a big growth leaver if you have a bunch of churned users

The impact of the engagement effect

• Retention can be improved systematically by looking at engagement loops
• create user cohorts by level of engagement and analyse what differentiates high values from low value users
• use A/B testing to prove causality
• once the best levers are found, test many variations of these ideas
• Rinse and repeat, to systematically strengthen the engagement network effect
• Engagement effect will automatically kick in as colleagues, influencers and others join.

The Acquisition Effect

• Acquisition Effect: The ability for a network to attract new customers as it scales
• This is the most magical and explosive forces
• Viral Growth Examples
• Youtube → make it easy to embed youtube videos on other sites
• LinkedIn → import contacts
• Eventbrite → email invitations
• Paypal → Sign up to accept your payment (the best value proposition ever)
‣
Paypal Viral Growth Tactics
• Ebay seller asked to display "We accept paypal" image on a listing. They realised how important eBay sellers were. They automated this, allowing sellers to enter Ebay credentials and they'd insert the button into their auctions.
• $10 to you and the friend you refer. The motion to signup was strong, but so was the incentive to continue inviting.
• As users started using the service more, they were able to recoup though transaction fees (it was more cash efficient than you'd think)
• WhatsApp got to 1M installs a day, using viral growth, without paid marketing
• Viral growth builds on the power of networks to acquire users, often free of charge

Product Driven Viral Growth

• Viral growth is not 'viral marketing' (creating sharable adverts for non-network products)
• Network driven viral growth is more powerful
• Network products can embed their viral growth into the product experience itself
• Dropbox → folder sharing
• Paypal → signup to receive a payment
• Slack → invite your colleagues to this chat
• Instagram → invite and connect your facebook friends
• The power is in the ‘Product - Network’ duo
• Product → has features to attract the user to the network
• Network → brings more value to the product
• This loop is fundamentally created by the product
• The Viral Loop
• New user hears about service
• Signs up
• Finds value in it
• Shares the product with their friends
• Some of whom sign up
• → Repeat
• it can be measured, tracked and optimised to be made more effective
• making the acquisition effect a potent force
• To make loops actionable you can...break them into granular steps, A/B test them, brainstorm and test 100s of ideas
• Making 5-10% gains can have a big impact on viral growth
• Hundreds of A/B tests later, the millions of dollars you're spending on acquiring customers is going to be more efficient

Measuring the acquisition network effect

• You have to be able to measure it to increase it effectively
• Viral growth can be rolled up into one number
• Productivity tool example
• 1000 users download
• x% of those invite (over the month)
• 500 of those download
• They invite, and 250 download
• and so on
• The ratio between each loop 1000 to 500, 500 to 250 is the viral factor
• 0.5 → each cohort generates 0.5 of the next
• starting with 1000, a viral factor of 0.5, leads to a total of 2000 users by the end of the amplification
• meaning an amplification rate of 2x
• Higher ratios are better - each cohort is more efficiently bringing on the next batch of users
• Use A/B testing and experimentation to increase your viral factor
• Sharing, referrals, invites,
• Viral factors have a massive effect (0.5 = 2x, 0.6 = 2.5x, 0.7 = 3.3x, 1 = 20x)
• Viral factors can't stay above 1 for long, you get market saturation and changing user demographics
• What levers effect the viral ratio?
• Strong retention is the usually the biggest
• Paypal: if people come back to send payments regularly, then there's a big chance they'll suck other people into the ecosystem
• big virality projects (referrals, contacts)
• lots of little optimisations
• You’ll need your best copywriting, psychology and product design to move the needle.
• Psychological elements + value proposition of a product = the best viral growth strategies
• Dropbox → folder sharing
• Zoom → meeting invites have Zoom invite links
• Often the loops are specific to the product. Traditional online advertising can be bought by anyone, naturally driving up costs and lowering effectiveness over time

Acquisition isn’t powerful without engagement

• The acquisition effect is independent of the engagement or economic effect
• You can acquire customers virally, but still miss on stickiness.
• Chain letters and ponzi schemes collapse when the supply of new, novelty-seeking recipients dries up

The Impact of the Acquisition Effect

• Cornerstone → understand how one group of users taps into their respective networks to bring in the next group of users
• Networks then attract other atomic networks - and so on
• Land and Expand
• Dropbox → a company sharing a folder with another sucks them into the network
• Landing = start new atomic networks
• Expanding = increasing the density of a network as coworkers join
• Networks built through viral growth are healthier than those launched in the big bang fashion (like Google +)
• big bang launches are great at landing, but less good at expanding
• low density leads to low engagement
• Increasing density and engagement isn't just easier new user acquisition, but also stronger engagement and economic network effects.

The Economic Effect

• Economic Effect: How a business model (inc. profitability and unit economics) improves over time as a network grows
• Driven by 'data network effects' - the ability to better understand the value and costs of a customer as a network gets larger
• drives higher efficiency in promotions, incentives and subsidies into a network
• Also grows revenue by increasing conversion rates (by building features for the network not the tool)
• By understanding these systems, product teams can strengthen this important force

The Network Effects of Lending

• Some of the oldest writings known are to record debts. (1754 BC, Code of Hammurabi - 33% interest rates for the ancient Babylonian)

• Overtime Bureaus tend to combine into larger ones → more data from more merchants and customers helps every merchant with better information
• As networks grow, they get stronger and develop advantages of premium pricing and higher conversion rates and cost improvements (like incentives and risk-taking)
• Efficiency over subsidy
• Launching new networks often requires subsidies for the hard side - paid back over time if successful
• Exclusive content wars (Netflix, Amazon, Apple and others)
• buy content → win a niche audience → fund your own content
• owning your own content can be a huge advantage over time as the network gets bigger

• Larger (denser) networks have higher efficiency than a smaller ones → therefore they can offer bigger incentives
• As networks grow → so does their ability to subsidise the ecosystem
• Marketplace companies are built on under-utilised assets
• Large networks with data advantages can personalise offers and subsidies
• Youtube pays more for finance, tech and fashion content

Higher Conversion Rates as the Network Grows

• Conversion is at the heart of many business models
• Economic network effect → conversion can go up over time as the network grows
• Premium features can be designed in a way such that they're more useful as the network gets larger
• so the larger network, the greater the incentive to convert to premium
‣
Slacks premium features become more useful with larger teams
• Voice calling, searchable message history
• the more people use slack, the more likely somebody will pull out their credit card and pay for it
• Market Place → more range, availability, reviews and ratings
• Social → social status

The impact of the economic effect

• Strong defence against new entrants
• The larger the network, the better the competitor has to be over the status quo to attract everyone across
• Strong economic effects allow you to maintain premium pricing
• switching costs become higher for participants
• Google charges high fees - because its network of advertisers, publishers and consumers is unrivalled
• Price becomes less relevant as the network grows - good luck moving your company off of Slack
• Competitors can copy features, you can have lower pricing. But the network is hard to acquire
• Economic effect is a powerful force that strengthens the business model over time
• more efficient subsidy
• increase conversion rates
• maintain premium pricing

Part V · The Ceiling

• As a product reaches scale, the growth curve teeters between expansion and contraction
• with phases of both extraction and contraction
• users leave at the same rate as new users sign up growth slows.
• Negative forces appear during the late stage of a networks cycle

Market Saturation	Regulatory action	Degradation of marketing channels
Churn from early adopters	Later mainstream users dilute quality of initial communities	Network Revolts - hard side becomes more concentrated
Bad behaviour: trolls, spam, fraud	Crowding - and degrading user experience (discovery, noise)

• Growth curves are rarely smooth, they grow in fits and starts. When a ceiling is hit, product teams scramble to address underlying causes. Features can raise the ceiling - but for only so long
• Network effects can unravel just as fast as they gathered, pulling down acquisition, engagement, and monetisation all at once. Hitting the ceiling hurts.

• Facebook - had to to build a growth team to get past 90m users - they plateaued. Creating friend recommendations.
• They Typical SaaS growth story
• Early success with startups adopting tools - high churn though
• Larger enterprise customers are slower to adopt but more stable and profitable
• often require enterprise sales team to break through the startup ceiling
• Introducing the ceiling
• bumping up against a range of problems - growth stalls, network effects weaken
• hard decisions need to be made

Saturation

• Two ways a networked product can hit a ceiling (growth plateauing) due to saturation:
• of a market niche
• of the entire market
• Marketing Channel Degradation (the law of shitty click throughs)
• marketing channels become less effective over time (banner ads and email marketing)
• if your product's network effects depend on these channels, growth will decline over time
• The network changes over time
• power becomes more concentrated on the hard side of the network
• act accordingly and revolt
• hard to keep everyone happy
• Early communities are often special, curated and moulded to share attractive norms and qualities
• As you reach the mainstream audience, more and more people arrive diluting the quality
• Discovery becomes harder - Overcrowding
• algorithms, discovery, search
• else people prefer smaller more curated products
• To overcome these forces - firms often need to
• make major product changes
• launch new products or innovations

• RocketShip Trajectory is tough
• It's hard to hit every year - triple or double
• there are plenty of headwinds (mentioned above)
• growth rates tend to drop over time (despite investment, engineers and product)
• the vision has to become bigger (requesting a limo becomes global transportation)
• expectations rise quickly too
• hitting ceilings therefore is really dangerous - everyone knows which are on the rise and which are stuck
• colleagues and investors quickly defect to higher growth companies
• Teams use all the obvious growth levers in the early years
• All the best ideas are taken and executed
• Network effect products have more tools to counteract the effect
• tapping into the network effects to fight the slowdown
• When network products work, they typically work for a long time

Saturation

• Success brings the market saturation problem
• To grow a network, you add more nodes. Eventually everyone in the target market has joined the network
• shift from acquiring customers to layering on more services and revenue opportunities for your existing customers
• You have to innovate out of saturation

• Underneath exponential growth curves are often layers of different networks, revenue lines

Network saturation vs market saturation

• the 100th connection for any given participant is likely less impactful than the first few
• as the network gets more dense over time, its network effects become incrementally less powerful
• the 100th rolex seller, uber driver, Paris apartment or friend added to a network add less value than the previous ones
• your top friend contributes 25% of Snap send volume. By the time you get to 18 friends, each incremental friend contributes less than 1% of snap send volume
‣
Facebooks 7 friends in 10 days
• is 14 better, yes but not 2x better.
• is 10,000 better, no won't get 1000x engagement you'll likely get a worse experience and see overcrowding
• Market saturation caps the total number of people on the network
• Network saturation caps the effectiveness of your engagement over time as interconnections slowly diminish in incremental value
• To fight the forces, you have to evolve your product, market and feature set.

New, Adjacent Networks

• In a network of networks, some are more engaged than others.
• the further reaches are likely to be less healthy

• Understanding the adjacent networks is key, so they can be targeted one by one to expand and fight saturation

• Figure out the adjacent set of users whose experience is subpar
• there might be multiple sets of nonfunctional adjacent networks at any given time, that require different approaches to fix each one
• low-end Android apps
• Ex. Women 35-45 in the US, had facebook but not instagram
• then it became women in Jakarta on older Android phones with a pre-paid mobile plan
• be nimble - build recommendations that utilise their connections
• market place products tend to become seller constrained over time
• continually evolve the offering to attract the next set of hard side users to your platform
• Uber started to think about signing up people who didn't already have a car
• Teams have to think about new markets - rather than listening to the vocal users in the core markets

New Geographies

• Adding international regions
• Obvious for hyper-regional products like OpenTable
• may need to solve the cold start problem in each region
• languages and currencies
• Adjacent networks are easier (some interconnection and similarity of preferences)

Eventually you have to add new products

• harder in existing companies (politics, distractions, lack of resources, adverse selection of talent)
• teams have an incremental mindset at this point, not what's needed for new products
• a product success rate outside in a funded company is 50/50. Exceptional outcomes happen 1 in 20 times
• Therefore
• buying companies is a cheat code - integrate them into your network
• ebay did that with paypal
• acquisitions are hard and expensive

The Law of Shitty Clickthroughs

• Banner Ads
• Every marketing channel degrades over time
• lower click through
• lower engagement
• lower conversion
• email, paid, social, or video
• A core reason why products hit ceilings
• What worked before eventually stops scaling as fast as you need it to
• when products hit 200% growth, acquisition channels need to grow that fast too
• Typical banner ad clickthrough rates (0.3-1%)
• the first banner ads had clickthrough rates of +70%, but they've degraded over time
• email effectiveness has degraded over time too (30% to 13%)
• Why drop-off? Consumers acclimate to specific brands, marketing. techniques and messaging and then tune them out.
• People become skilled at ignoring advertising

Degrading the network

• Acquisition effect is a series of steps where users encounter a product through an invite, use the product, and invite others.
• if you send too many emails, people start to ignore them
• Small changes in viral factors have amplified outcomes on conversion of the viral loop
• New users are often the most engaged in welcoming others to network. Remove the stream of new users, and engagement within the platform from more tenured users can decline too

Layering on New Growth Strategies

• Embrace it's inevitability
• New products there are usually one or two acquisition channels that work
• but they might not scale
• Dropbox → the viral video + hacker news (wasn't going to continue working
• Pouring more money into marketing creates problems
• starts off highly efficient - pay back periods creep up over time
• the more you spend the worse it performs
• You need to constantly lay on new channels
• paid - youtube, snap, instagram and other ad platforms
• viral loops
• engage content creators
• content marketing - SEO
• direct sales (for SaaS) → look for email domains as clues for companies (start asking customers for name, team size, company)
• add contact us (on your pricing page)
• Understand what channels are a best fit, hire somebody who's already worked on those channels (or get advisors)
• Embrace new marketing ideas early
• New media formats and platforms all the time
• TickTok, Twitch, Instagram
• Referal programs, memes, emojis, video clips and other tactics

Tapping into the acquisition network effect

• It's more efficient for networked-products to optimise viral loops vs traditional marketing spend
• Twitch focused on making streamers more satisfied, as that ultimately was more effective than trying to drive users to the existing base
• You can't buy 1 billion users, you need to have a viral loop
• Marketing cost to get an app installed $10 per user, 3x more for financial apps

When the network revolts

• Uber would pay $20-50 for an active rider. $200-2000 for a driver. 5% of the network were drivers.
• A small number of power drivers became really important to UBER - important and scarce. 15% of drivers doing 40% of the trips.
• A well organised revolt can kill a product completely
‣
Vine was killed off by stars leaving the platform
• 18 creators demanded $1.2m and some feature changes for 12 posts a month, else they'd leave
• they left and Vine died
• Often the 80/20 rule apples, so you're going to have imbalance in the importance of contributors
• Concentration is the result of healthy loops that driver a network toward higher quality.
• Bad creators churn
• Good creators earn, reinvest and get better
• Networked products generally want to nudge the ecosystem toward professionalism because it helps scale the hard side
• training - incentives for the best on the hard side
• As a network scales, the hard side will professionalise. Quality and consistency will increase, the most sophisticated players will be able to do it at scale.
• This dynamic misaligns incentives - they might protest - they might complain, quit or compete. SaaS partners might negotiate pricing, request feature or threaten to quit

How professionalism happens

• Homegrown professionals - becoming a power user over time. Easier to get big on youtube than AirBnb (you don't need the $ms of real estate)
• Larger more stable networks become platforms or economies that people rely on and invest in (youtube and App Store)
• a rich eco-system of players emerges.
• Different sets of platforms concentrate to different degrees
• AirBnB and Open Table - less concentration
• Social - more concentration

No choice but to scale

• You can saturate the hard side of the market quickly
• it starts to get really expensive to recruit more people
• So you move from acquisition to increasing engagement and scale of your existing suppliers
• BUT - That creates concentration
• Uber runs out of professional drivers, so had to go after those without a car, or a job
• Whilst focusing on getting more trips from existing drivers
• You don't want the hard side to churn, so you should invest in training
• Embrace the professionalisation of the hard side- it is needed to scale, but then you're concentrating power too.
• Eternal September - Usenet students joined in September, each time degrading the network, September 1993 was the September that never ended

Scale attracts bad behaviour

• When successful networks grow large audiences they attract spam
• Godwins law - every heated online argument devolves into comparisons with Nazis
• Smaller communities police themselves, but that doesn't scale
• core culture is diluted as new members join
• porn, piracy etc

Context Collapse

• Early focused atomic networks have concepts like 'netiquette'. Shared context.
• As social networks grow, what starts with your close friends ends up attracting interest from your parents and your boss.
• What you say in one context might be different from another
• what you post, how you interact, what constitutes an appropriate comment
• Context collapse it what happens when too many networks are simultaneously brought together, and they collapse into one
• on social networks, it inhibits the behaviour of content creators
• Context collapse
• An infinite number of contexts collapsing upon one another into that single moment of recording. The images, actions and words captured by the lens at any moment can be transported to anywhere on the planet and preserved for all time.
• Making people face a crisis of self-presentation
• The bigger the network, the more people might see your content
• Tension:
• Product team wants and needs growth
• Growing slowly degrades the experience
• When the anti-network effects become strong enough, the network hits a ceiling

How do you prevent a context collapse?

• Provide smaller spaces and pockets for communities to exist
• permissions and privacy features
• Split things up too much though, and you'll end up with many small inactive networks
• Not all networks need experience context collapse as soon as others

The power of the downvote

• Spam, trolls, scams, fraud
• They affect both retention and acquisition
• Leverage networks themselves to flag bad behaviour and remove bad actors
• give users the ability to report spam, accounts, block content
• allow users to customise their experience then mine the data
• Reddit downvote is a form of the community policing itself
• Large scale communities - standards and self governance can't be maintained by people simply running around and talking to each other.
• Create features that nudge interactions in the right direction
• Working hours in Google Calendar
• Report a twitter account as hacked
• Machine learning and automation can help detect and block scammers.

Overcrowding - Youtube

• As youtube grew - discoverability became difficult
• Initially, there's very little content to organise. So just sorted by recently uploaded
• Viral loop:
• Video uploaded
• Shared to 10 people
• Viewed by 5 people
• 1 or more would upload
• Video embedding and real-time transcoding sped things up
• More videos required a redesign to help people discover the best videos
• Top 100 (day, week, month)
• Then by Country
• Homepage - 10 chosen videos
• 100x more viewers than creators - so we added comments
• helped creators get some feedback
• Overcrowding solutions - manual curation < popularity ranking < algorithmic
• Manual Curation
• Community Curation (wishlists on Amazon)
• Data (popularity, groupings, trending)

The Rich get Richer

• Already popular get more popular, how do more people break in?
• One hypothesis on why social networks lose heat at scale is that the 'old money' can't be cleared out, and new money loses the incentive to play the game
• encourages users to leave for a new network where they don't face that problem
• The more connected a node is, the more likely it is to receive new links

The Power of Data and Algorithms

• Youtube after the acquisition - "We were just trying to keep up with all the traffic"
• Relevance, search and algorithmic recommendations were the only new features added
• Better matching between content creators and viewers stops the overcrowding issue
• Take signals from user interaction to surface more relevant results

• Data network effects are often invoked as a path for networks to solve relevance and overcrowding issues that emerge over time
• individual interactions, plus that of the entire audience
• The fight against overcrowding never ends - search, discovery, algorithms, sub-communities
• Be cautious about selecting content that is just about engagement as that might not result in the highest quality experience (click-bait, rage inducing content)
• Youtube adds 600 hours of content every minute.

Part VI · The Moat

• If your product has network effects your competitors likely have them too

Introducing the Moat

• Warren Buffet explaining the moat → the competitive advantage of any given company, and the durability of that advantage
• For Networked Products... the moat is the effort, time and capital it would take a competitor to replicate the product and network.
• Creating AirBnBs product. Getting all the listings in a city. Getting customers onto the app
• Once you've overcome the Cold Start Problem - it creates the barriers to new entrants
• Overcoming the cold start problem is hard
• Overcoming the cold start problem + a competitor with a strong product and network is really hard
• New players therefore can't just do the same thing. New entrants have to either
• provide a much better experience
• or a differentiated experience
• AirBnBs global network is more robust, where as Ubers city specific networks are more challengeable

The Battle of Networks

• Stakes are high - complete dominance or complete annihilation are probable outcomes
• networked products lean toward 'winner takes all'
• this happens because atomic networks tend to standardise on a single product for convenience
• If a company can win a series of networks faster than its competition, it develops an accumulating advantage
• acquisition, engagement and monetisation advantages also happen
• same effects in reverse unravel smaller networks
• What is the competitive playbook?
• Doesn't matter → who can ship the most features
• Matters → network dynamics (like density and quality)
• Matters → harnessing network effects and building products that reinforce those advantages
• Not all about → size
• Not all about → first mover advantage
• Matters → cherry picking the high quality networks and winning them over

Your Competition Has Network Effects Too

• Network effects don't magically help you fend off competition
• If your product has them, your competitor does too
• Effective strategy is about who scales and leverages their network effects in the best way possible
• Smaller players often upend larger ones
• Who is doing the best job amplifying and scaling their Acquisition, Engagement and Economic Effects
• Myspace < Facebook
• HipChat < Slack
• Uber Eats < DoorDash

Network Collapse

• Maturity of the market dictates the nature of competition
• Early on - every networked product can gain traction
• When markets mature, competition becomes zero sum
• Vicious cycle created alongside a Virtuous Cycle
• Network effects boost the winner, and create strong negative network effects for the loser
• If the value of network grows quickly as people join, the opposite is often true too
• acquisition, engagement and economics all move in the wrong direction
• Sometimes networks go to zero (or nearly so), or retracts to a niche
• Competition can force your network to fully unravel and collapse
• Atomic networks that collapse can pull down adjacent networks too

David vs Goliath Strategies

• Asymmetry lies at the heart of a network-based competition
• The larger or smaller network will be different stages of a Cold Start framework
• gravitate toward a different set of levers
• Competitive moves of David and Goliath represent their maturity and the maturity of the market. They have have different goals and resources

	Goliath	David
Goal	Fighting market saturation and growth slow down	Solving the cold start problem
Strategy	◦ add new use cases ◦ introduce new audiences ◦ generate profit	◦ starts with a niche ◦ doesn't have to worry about profitability ◦ focusing on top line growth
State	More resources, man power and existing products	Fewer resources, capital, employees and distribution
Play	• slower execution, risk aversion, strategy tax (new products have to align to existing business) • large companies introduce processes which slow down entrepreneurial risk taking	◦ but they have speed, and no sacred cows ◦ trying & failing many times is normal ◦ they might try different niches ◦ if they discover one atomic market they'll be able to get investment and resources to support them

Cherry Picking

• Startups have consistently unbundled Craigslist, by cherry picking its most valuable audiences
• Tinder, AirBnB, Etsy, Right-move
• Craigslist is a network of smaller networks that can be picked off
• Startups can cherrypick the network that has a valuable use case and is poorly defended by the incumbent
• Each startup needs just one atomic network, yet each incumbent has to defend all of its networks. This is the asymmetry of network-based competition
• Large networks made up of many communities, often leave some communities underserved
• It makes sense for Craigslist to focus on features that help horizontally across the sight, vs helping just a small vertical (like room renting)

Finding the soft spot

• Innovator's Dilemma. New players can start in a seemingly undesirable niche which is ignored by the incumbents, while they are focusing on the most profitable segments and use cases
• Incumbents start over-serving their customers, as there's diminishing returns to new features
• Upstarts can dominate a niche then go after the mainstream.
• Where disruption theory meets network effects
• Network density beats total size. AirBnB could quickly create a dense network within a city. They would quickly have more listings than craigslist in a given city (together with a superior product).
• Pick your cherry smartly → look for things with network effects, high order values, engagement rates etc

Switching over entire networks

• Upstarts benefit from cherrypicking because the incumbent has conveniently aggregated the network
• AirBnB created a bot that copied AirBnB listing to craigslist, that all but killed that community off
• Lost networks are unlikely to be regained (as they now face the cold start problem)
• Decline in market share makes it harder to raise money

The Danger of platform dependence

• Be wary of platform dependence - integrating too closely with a preexisting network, allowing them to control your distribution, engagement and business model you become a feature of their network.
• Goliath might just choose to duplicate your functionality if you become successful.
• Cherry picking is powerful, because by competing and focusing on a single point, building an atomic network - large networks can't defend every inch of their product.

Big Bang Failures

• Big bang launches are often a trap for networked products.
• Wide launches create many weak networks which aren't stable on their own
• built on broadcast channels - large spikes in users, but they are untargeted
• get a small number of users from different networks
• density is too low
• aggregate numbers are vanity metrics!
• Google +
• Bottom-up networks are more likely to be densely interconnected, healthier and more engaged.
• Word of mouth is powerful, as you're likely to at least know one user on the network
• Apple does big bang though?
• Stand alone as premium, high-utility products that don't need networks to function

Paradox of small markets

• Start with small markets, grow atomic networks, grow into larger ones.
• The first network looks like a tiny market that doesn't deserve the attention. Sounds niche.
• Total addressable market might looks small.
• Use the success of the first network to tip over others
• An airbed company ends up disrupting the hotel industry

Allure of the big bang launch

• Big companies like big launches. Try to jump from zero to the Tipping Point in one bang, because of the internal pressures of starting and building a new product internally
• Why should we care about this its too small?
• Why are you aiming small? You should be thinking big
• It seems trivial to succeed in one school, or on one campus.
• Hyper involved CEOs will make sure its a big bet
• To startups, everything is better than zero
• Big bang launches also stop you doing things that don't scale

Competing over the hard side

• It's often not winner takes all
• Larger products have to spend huge amounts of energy defending against smaller players, sometimes they still lose
• If network effects are powerful, why are large networks vulnerable?

• Collect metrics to figure out the comparative position of all the players in the market
• this allows experimentation and execution
• set goals agains their numbers and that of competitors too

• Uber - focus on the hard side of a network, use financial incentives with product, support teams with sophisticated dashboards
• Uber also estimated how much runway competitors had, as they started to run dry Uber accelerated its attack

Bundling

• Bigger networks can expand into new categories and markets
• leveraging existing networks and capabilities
• Uber → Uber Eats
• Microsoft Windows → Azure, Office, Teams etc
• Distribution doesn't win when the product is inferior
• Bundling might get people to try, but doesn't mean they'll stay
• Microsoft bundled word, powerpoint and excel into office. Everything worked together making them all more powerful

Competing with a network not just features

• Cross-Selling = Market your new product at existing customers
• Unless they stick it doesn't help much

Locking in the hard side

• Microsoft won developers → with developer tools. Visual basic was key.
• Visual Basic allowed many companies to automate things
• Backwards compatibility (popular among developers)

Bundling drawbacks

• Adds clutter to designs, dilutes your value proposition
• can make your existing products worse