Network Effects are a virtuous circle - the gift that keeps giving.
And in my opinion, Network Effects are the most valuable and most defensible competitive moat a company can develop.
Network Effects come in different flavors. And not all are created equal.
As such, there’s an art and a science in identifying which you should put your resources towards trying to harness. Some take capital, some take content, and all require users.
This week we’ll dive into the subtle differences that define these flywheels:
Physical Network Effects
Direct Network Effects
Two Sided Network Effects
Local Network Effects
Indirect Network Effects
Hub and Spoke Network Effects
Data Network Effects
Belief Network Effects
Tech Performance Network Effects
Asymptotic Network Effects
Pony Express Recruiting Poster, 1860
(The Pony Express was an OG Physical Network Effect. Killer advertising, BTW.)
Physical Network Effects:
The network becomes more valuable as more physical nodes join the system. The network is dependent on physical, tangible objects, and they need to be local.
Pony Express: The more riders, the faster you could get mail somewhere. Each new rider who signs his life away makes the nascent postal service more valuable.
Railroads: Another OG physical network, which replaced the Pony Express… The more railroad ties laid and stations built, the further you could transport goods and people.
Cable TV: The fact that crazy people laid physical cables across the country, hanging from telephone poles and digging under lakes and oceans, makes this one even more impressive.
Physical networks are less impressive for their technology (e.g., horses) and more impressive for their sheer scale, breadth, and capital intensity. That’s why people stick with Comcast even though their customer service sucks - the physical network they’ve built makes them the only show in town (until streaming came along).
Direct Network Effects:
The value of the network simply increases with more users. The most obvious example: the phone. I imagine Thomas Edison had a tough first sale - it was entirely useless without anyone else to call.
Facebook: As more users join, there are more people to interact with and share your crappy photos (Annual PSA - please do not share your 4th of July firework photos this year. I do not care…)
Twitter: Each new user adds value by contributing content, which in turn attracts more users seeking information and interaction (read: bullying)
WhatsApp: The utility of the messaging app increases as more people in one’s personal network join… you get the point.
Two-Sided Network Effects:
More supply attracts more buyers, and more buyers attract more suppliers. Think of a farmer’s market - more families make it a weekend activity as more local vendors hawk their jams (variety!). And more local vendors show up because they hear about the increase in foot traffic, envisioning all the money they could make. Bingo bango.
Airbnb: More listings attract more travelers, and more travelers encourage more hosts to list their properties.
LinkedIn: The platform becomes more valuable for candidates as more companies use it to recruit, and with more fish in the pond, it becomes more attractive for recruiters.
DoorDash: A growing number of restaurants on the platform attracts more customers, and more customers means more orders for restaurants, drawing in even more dining establishments who want to share in the network activity.
Local Network Effects:
Babysitters in the Chicago area are useless to parents in Boston. Speaking of that. Edwin Dorsey told a hilarious story about exposing Care.com on the Run the Numbers Podcast (Apple | Spotify | YouTube)
Grubhub: Restaurants and diners need to be located in proximity of one another for it to be useful - nachos don’t travel well.
Lyft: The availability of drivers in a specific area improves wait times and reliability, making the ride share service more attractive to people in the area.
Nextdoor: You don’t want advice on a retirement community in Florida if you live in a bustling suburb of Nevada.
Indirect Network Effects:
Remember when you were a kid and wanted to play Halo, but had a PlayStation, not a Xbox? This one was always a hard one for me to wrap my head around. I’ll try to make it simple…
This is when a tangential and complimentary product increases demand for another product.
The more video games that are built for Xbox, the more attractive the system becomes.
Electric Charging Stations and Electric Cars: The more electric charging stations that are built along a country’s highway systems, the more appealing electric cars are to buyers, as they have more access to charging.
Google Chrome Plugins: The more tools you have to connect to your browser, the stickier Chrome becomes.
Garmin Watches and Lulu Lemon: As more people want to track their fitness, more athletic wear is purchased.
Source: NFX
Hub and Spoke Network Effects
Substack is an excellent example - writers (spokes) push their content to a hub, where it’s distributed back out on the Substack homepage and through authors who recommend one another.
TikTok: Influencers post their content to the TikTok platform, and it gets distributed via an algorithm to people all over the globe, where some go viral.
Book Publishing: Author send their books to Penguin Random House, and it’s distributed to books stores (Barnes and Nobles was a VIBE back in the day).
HackerNews: Tech workers submit links and they get upvoted, climb a list, and get more eyeballs.
The key here is building an algorithm that finds consumers where and how they want to receive the content being distributed. As NFX points out, the Hub and Spoke model can go wrong when the content being distributed alienates the audiences who came to the Hub in the first place.
Data Network Effects:
This is when data from users leads to a better experience, attracting even more users. Think of it as using the “exhaust” from the original system to make it even better.
Amazon: User behavior data helps improve product recommendations, enhancing shopping experiences and attracting more shoppers.
Netflix: Viewing habits help to improve movie recommendations, increasing user satisfaction and retention, which encourages more subscriptions.
Spotify: Listening data is used to refine music recommendations, and serve up those BANGER Daily Mix Playlists
Belief network effects
These occur when the value of a product or service increases as more users adopt the same belief system or ideology.
Religion: Belief in a particular religion is strengthened by the collective belief of others who follow the same faith. It’s a snowball effect. The more people who believe in a religion, the more socially accepted and reinforced that belief becomes, creating a network effect (or cult).
Fiat Currencies. The belief in the stability and strength of a currency is only as good as the collective belief of the population using it. The more people who trust and use a specific currency, the more valuable and widely accepted it becomes, creating a belief network effect.
Bitcoin: The value of the currency increases as more people believe in its potential (even if there aren’t a ton of real use cases yet, lol).
Tech performance network effects
The more nodes on the network, the faster it moves.
LimeWire: Remember illegally downloading files as a kid (hypothetically speaking, of course)? When there were more people willing to share the same file, the download went faster. Peer to peer file sharing runs on network effects.
Bitcoin Network: How congested is the network? How fast can you get a transaction processed? This can be measured in transactions per second.
ReCAPTCHA: Solving those annoying word puzzles actually digitizes physical books and newspapers (seriously!). The speed at which people authenticate therefore impacts the speed of translation on a New York Times article from 1923.
Asymptotic Network Effects
OK, this is our first less positive Network Effect. This is when there are diminishing (or even destructive) returns as more users are added.
Uber: Eventually the difference between a driver getting to you in three minutes vs four is negligible, and actually panic inducing, since you don’t have time to grab your stuff.
Tours: Once you have four dolphin watch boats in Tampa on your marketplace for activities, the value of the fifth is smaller than the first.
Dating App Mix: If Tinder or Bumble are flooded with too many males, they’ll inundate the number of women with connection requests, and actually degrade their experience. Supply isn’t always additive past a certain point, especially if it’s not the right supply the opposite side wants to see.
Here’s an excellent video NFX made on the 9 types of network effects they identified:
