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 Silicon 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 go and buy the book.

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 capitalize 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 subsidizing, 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 monetization, reduce costs, better business models)

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 optimized 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 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 favorable response.

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 personalize 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 behavior: trolls, spam, fraud	Crowding – and degrading user experience (discovery, noise)

• Features can raise the ceiling – but for only so long
• A product can saturate it’s market or it’s niche.
• Marketing channels become less effective over time (banner ads and email marketing)
• Power becomes more concentrated on the hard side of the network. It becomes harder to keep everyone happy. A well organized 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 professionalize.
• As you reach the mainstream audience, more and more people arrive diluting the quality. Discovery becomes harder, and you get overcrowding
• 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
• 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
• 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!
• Tapping into the acquisition network effect – It’s more efficient for networked-products to optimize viral loops vs traditional marketing spend. You can’t buy 1 billion users, you need to have a viral loop.
• Scale attracts bad behavior: 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 behavior of content creators. Leverage networks themselves to flag bad behavior and remove bad actors. Create features that nudge interactions in the right direction.
• 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
• 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 undeserved.
  • 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 cherry picking 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