The Flight of Networks and their Rebirth at Scale
The forces at play behind Network Effects. A retelling of the Cold-Start Problem.
Listen to this article:
Networks are a strange, intricate and interconnected web of objects1. They inform our lives deeply, yet we haven’t tried to understand them very well.
Languages are networks of words. Culture is networks of narratives. WhatsApp is networks of friends and families. And then there are religions— the ultimate networks—self-perpetuating myth machines that are generationally sustained.
On the surface, the value of networks seems obvious. More people that speak a particular language, more valuable it is to speak that language. Double the nodes, quadruple the value. This is called the Metcalfe’s Law.
Metcalfe's Law however is inadequate and incomplete; often misused. It assumes that every node in a network is created equal. Assumes that all network-life-cycles are created equal. Assumes that all networks are created equal. When in reality, we can see that the growth of networks is sinuous and fragile. Buddhism didn’t grow the same way that Christianity did. What made it collapse?
Why do some networks succeed while others fail? How do you create a thriving network? How do you keep it growing?
To answer these questions, we must first understand how networks form, wow they evolve, and how they die.
Andrew Chan has tried to account for Metcalfe’s misgivings. By making a more useful law with his experience at Uber. He calls it the Meerkat’s Law.
Meerkats (furry little African mongoose) stay protected from predators, in groups. They have strong and valuable network effects. But only at the right, stable population. Or the population dies.
Like the meerkats, there seems to be a threshold for networks to take hold. A tipping point. Once you pass this tipping point, the network still needs reinforcement, with active and engaged network participants to propagate the flywheel- or again the network can collapse. Beyond this lies the carrying capacity, where your network can get overcrowded and also collapse. Do all this well, then you can enjoy a strong and beautiful network moat. One that protects you from prey.
Part I. The Birth.
The early stages of labour are difficult. The birth of a Network too is a difficult strange loop paradox. Much like the chicken or the egg. A causality dilemma.
Why would riders join an Uber network if there are no drivers, and why would drivers join if there are no riders? Why would you use WhatsApp if none of your friends are on it?
This leads to the classic empty-state problem in UX. Andrew Chen likes to call it the Cold-Start Problem2.
The best way to solve for this?
By Building Atomic Networks. The smallest segment where networks can be formed and stabilised.
The science or art here is choosing what features make up the right atomic network for your domain. It’s probably much smaller and more specific than you think. Something tiny, like 5pm at Block 12 in Indiranagar for Uber. A university for Tinder. A city for Airbnb. Just 2 people was enough for Zoom. A single small tribe, fulfilled with a few pizzas. Not thousands of users. Finding the network-market fit is tricky. Much like the product-market fit.
The Hard Sides to Cold Starts.
Often there are tasks that need more work in a networked product. This forms the hard side to multi-sided networks. Creators for Twitch. Missionaries for Christianity. The supply side for Amazon. The banks for UPI. Writers for Wikipedia. It’s crucial to cater to the needs and workflows of your hard-side to solve the cold-start problem.
Other Strategies That Help.
Come for the Tool, Stay for the Network. This is one popular strategy. Build the single-player tool. Sell the tool. This lets you build a distribution. Now you can sell the distribution. This is what Instagram did initially.
Flintstoning. If the hard side of the network isn’t yet activated, a team can fill in the gaps themselves. Do it yourself if the automation doesn’t exist. Reddit co-founders submitted links and content themselves. Until they added automation and community features for scale.
Growth Hacks. There are some shortcuts to help with the Cold Start Problem. Using marketing and growth hacks. These are helpful but not enough. Thes best way is to get your own network to spread the word (more on this later). Another way is to bundle or leverage others’ networks with partnerships. Microsoft was able to use their impeccable distribution to have Teams overtake Zoom even with an inferior product. Know your innovators and early adopters very well. Get intimate with them. They will help solve you the cold-start problem.
The Tipping Point.
Each new addition of a network would become easier with each new network. The most successful networks are grown on a network-by-network basis. Uber is not one global network of drivers and riders but it is a network of networks. This seems to be a universal truth.
Interlude 1. The Magic of Open Networks like UPI.
Interoperability is a magic adhesive that binds disparate atomic networks.
The adoption curve for UPI (Unified Payments Interface) in India has been a hockey stick. Doing almost 6 billion transactions in a single month.
It leveraged the network of networks through interoperability3. Before UPI, PayTM and other wallet apps had their own individual atomic networks. UPI managed to engulf each of these networks. UPI is voluntary. Yet its network became so valuable that FOMO of not being in the network drove all the other banks to the network.
An RBI endorsement solved the hard side of the UPI cold-start problem of banks. Companies like PayTM and PhonePe solved for the other sides. For consumers and shopkeepers. One city, shop and household at a time.
Part II. The Flight of the Flywheel.
Laying the atomic network egg is only half the battle. You need to make it take-off. Make it flock its wings together with its siblings. Keeping predators at bay.
For this you can helm the trio of forces that make the network effects. To perpetuate and reach escape velocity.
1. The Acquisition Effect.
Good stable networks beget more networks. You can leverage the power of your own network to spread the wings further.
Gmail did this with artificial scarcity (invite-only). Paypal did this with double-sided rewards. Christianity did this with a strong salesforce called Apostles. Followed by murderous inquisitions.
If your product isn’t working well for your network, you might have to pivot. Otherwise you can consider sharpening the product wedge that brings true value. This means better and more grandiose myths. Better legends. Or more focus on a better tool for video editing.
Consider this especially for hard sides. For many singles night clubs, the hard side to get on board is ladies. They offer discounts to this side. You can choose to follow this ladies night strategy.
Viral Factor. The ratio of new users invited by the existing users gives you the viral factor. A viral factor of 0.5 means if 500 people joined the network they will invite another 250. You want to stay above a factor of 1 to be viral.
First Movers. Acquisition is easier when there is a salient value proposition. First mover advantages are valuable, as networks tend to compound. Like Amazon Web Services having a 10 year head start. But if the first movers don’t keep the users engaged, they are certain to collapse. This brings us to the second effect.
2. The Engagement Effect.
Religions without their rituals will wither away.
Networks need reinforcement. The nodes need to be sticky. Or they loose their adhesive. It helps to make rituals fun, engaging, reproducible. The best network apps have much better retention curves.
This is the core part of the success for the best retention curves. You can drive engagement by new use-cases. Thinking in game design.
You can write a whole book about engaging and hooking network participants. Lucky for us, it is already written.
3. The Economic Effect.
Economies of scale become evident as a network grows. Letting you have a a better business model. This lets you subsidise and focus more on conversion. To strengthen your network further. You can also charge a premium for a better network.
There is often a Pareto principle at play. 80% of the network derives its monetizable value from 20% of participants. This is prevalent on the hard side of networks. 20% of drivers for Uber, 20% Youtubers. You need to know your 20% well. They are likely to make you money. Or defer you from doing so.
Interlude 2. The Grief of Google+.
When Google launched Google+, it did not make sure that the atomic network was stable enough. Before working on acquisition throughout its other channels. This lead to Google pumping traffic towards a product that had not proved its market and network fit yet. We all know how the story ended.
There was no user base of an atomic network that invited others. It had no reinforcing loops around an engagement effect. It could not leverage any economic effects. Having been a barren land with a deserted excuse for a retention curve.
Part III. The Death & Elixir (or Moat of Networks).
With inertia, also comes friction. Networks are never invincible. They tend to drown in their own moats. Skype, Yahoo, and many others. Leaders in a generation, bleeding in another.
Hard side of networks might revolt, like the drivers did for Uber. Or creators for Vine. Marketing efforts might degrade (with the Law of Shitty Clickthroughs). The network might reach market saturation.
At this stage, the competition is fought mostly on the hard side of the network. Their leverage levee makes every kilo matter. The major fight between Ola and Uber is on the driver side.
When your moat is too big, you stop noticing the leaks. An unbundling is triggered. Craigslist has been a grandfather for such rebirths. With Airbnb, Uber, Zillow and others. Within their niche they offered “sufficient departure” over Craiglist. With stronger acquisition and engagement loops.
A network crash can occur if these issues are not navigated well. Like with Pyramid schemes. Positioned as perpetual motion machines. Mythical beasts. Powered by new entrants. Crashing the party when the fuel ends.
The bigger players can make their moats stronger by bundling values together. Creating an ecosystem tying networks together. To grow further, spin off new atomic networks in different directions. Become a super-app. If the newer networks stick, welcome them into the network family. With open arms.
Takeaway.
Whether you are building a religion, a product or a language. Networks will shape your value and you need to learn how to shape the networks.
You do this by building atomic networks that are defensible with key product features. Then focus on the acquisition, and engagement effects with better economics.
Finally, defend your atomic network eggs. A growth drought might be inevitable. So look outward.
If you think this sounds interesting, bookmark these other great reads:
The Cold Start Problem by Andrew Chan (2022)
Thinking in Systems by Donella Meadows (2008)
Why Content is King by Every (2020)
More DALL.E:
Disclaimer: All the art used in this piece was created with help of DALL·E by OpenAI. With prompts provided by Harsha. They are used here non-commercially.
A network is any collection of objects in which some pairs of these objects are connected by links.
A cold-start is an attempt to start a vehicle's engine when it is cold.
Sidenote: In systems thinking, reinforcing feedback loops amplify effects. Balancing feedback loops stabilize them. These two processes often exist together in any system, including network systems. More on this another time.
Note: Interoperability already existed with bank to bank transfer methods like NEFT, RTGS, etc. But UPI was to bring interoperability at the level of convenience of wallets. Having to share 4 large numbers and codes to make a payment clearly wouldn’t have helped grow payments. UPI’s magic is doing this with a single identifier.