My main takeaways from The cold start problem

Rarely I have been waiting for a business book with the same excitement that I had when waiting for ”The cold start problem”. I have been a fan of Andrew Chen and his retention and growth theory since long ago so I couldn’t wait to see how he would unpack the dynamics of networked products.

And I must say that he really delivered. The cold start problem is an excellent playbook for everyone who is working with or is interested to understand more about networked products. It mixes concrete examples from companies that everyone recognizes (Uber, Google, Air B&B, Slack) with theories about the different phases of networked products. And most importantly he puts words and frameworks on phenomena that I have seen happening in front of my eyes when working with these products and that I couldn’t really fully articulate myself.

What I particularly appreciated from the book is the fact that it goes deep into details of networked products, showing how there is not one networked effect that, once is achieved, is the silver bullet that will make a company invincible. There are instead different phases that a networked product goes through, from conception to stabilization, growth, and in some cases decline. And all these phases are important to recognize and work accordingly with.

”Networked products are products that become more and more valuable as more people use them. Those generate network effects: the value of the network is to bring everyone together, and the effect is given by the fact that the network self-grows itself by having more value as the network grows and more people use it. Examples of networked products are Slack, eBay, Facebook, Air B&B, Uber…"

The core of the book is ”The cold start theory” which is a framework that Andrew developed to synthesize the different stages that a networked product goes through.

If I had to summarize it in one sentence I would say that it is about finding your atomic network, the first stable network in your product that delivers value (the cold start problem) focus on virtuous viral loops to increase the value you deliver to the network (the tipping point), help the network grow by focusing on viral growth functionality (escape velocity), nurture it by always focus on the hard side of the network and on its quality. And repeat (The ceiling and The moat).

Andrew framework from the book, with my own notes

When reflecting on this framework on my own experience working with networked products, I realized that this book really gave some concrete ways of reframing common themes. From the myth that networked products are invincible, to a way of overcoming the hen or egg problem typical of two-sided marketplaces. These are my main takeaways:

Beware of the myth: there is not such a thing as ”reaching network effects” and creating a bulletproof product

One of my most important takeaways from Andrew’s framework is that it makes it really clear that there is no ”one network effect” but different phases and nuances that go into creating products that become more valuable as more people use them. Many times I have heard ”product X has network effects. it is a competitive advantage that is impossible to beat”

And as much as once networked products are undoubtedly hard to beat once they are established in the market, they are not invincible. Examples are Skype and Google + who lost their competitive advantage for other networked products such as Slack and Facebook.

What Andrew puts a spot on is that creating, growing, and maintaining a networked product requires discipline and understanding of the phase that you are in. And every single phase needs to be treated differently, with specific frameworks to make sure that you can create, grow and maintain the network effect.

If you are lucky enough to have created a networked product, it will give you a true advantage, but it will not last forever. Be mindful of the phase your product is in and build for it, always taking an extra eye on both the quality and the hard side of the market.

Finding your atomic network is more complex than finding your product-market fit

The beauty of networked products is that once you find your first atomic network, the smallest possible network that is stable and can grow on its own, then the growth really fulfills itself. You’ll have to product-develop to help the network grow through its different phases, but you will not have to product-develop in order to deliver core value to the network itself.

But for all this magic to happen, you need to find an atomic network.

I used to think about this phase as finding product-market-fit.

But while reading this book I realized that finding an atomic network is more complex than finding a product-market fit for a product, as you will have to deliver value for someone and at the same time make sure that you have the right incentives in the network.

Let’s put this into a concrete example Andrew gives: Tinder. At its beginning, Tinder found its product-market fit by proving that there were students on campus interested in the idea of online dating. But in order to find the atomic network, a good balance between the two sides of the romantic match needed to be found as well. So from the beginning, the right incentives to get in the hard side of the network (in their case women) needed to be in place. Otherwise, the network would not grow, and the product-market fit would be useless.

This way of framing networked products (a product AND a network) opened a new way of thinking to me. When working on a networked product make sure to both solve a problem and have the right balance in the network from the beginning.

Forget the hen or the egg problem, there is always a hard side of the market

Many times when thinking about a networked product, for example a marketplace, I got stuck in the ”hen or the egg problem”: is it the demand side or the supply side that is most important? There is no demand without supply, but the opposite is also true. Instant headache. Until this book came for the rescue.

One of the most interesting concepts that Andrew introduces is ”The hard side of the network”. The hard side of the network is that part of the users who creates the majority of the value and that therefore have a disproportionate amount of power and you need to win over.

A concrete example: for YouTube, the hard side is creators. They have to a lot of effort to create content, and at the same time for every piece of content created, the network can attract an exponential amount of watchers.

There is always one side of the network that has to put in more work for the network to function and that you need to nurture even more. And the beauty of it is that the hard side of the network is not always the same as the network evolves. One example is a B2C marketplace where the hard side of the network shift from supply to demand. In the beginning, the hardest part is to attract demand, but as the network grows and establishes itself and it becomes hard to shift demand because of convenience, supply gains power and becomes the hard side of the network. Until something else happens, and the balance might shift again.

Thinking about the hard side of the network really opens up another perspective: there will always be one side to nurture more than the other: but which one is it? And how and when should you re-evaluate your hard side? How do you align incentives with it to grow the network? Before this book, I had not thought in those terms, but I really had a eureka moment when framing problems in this way.

There is no way around it, you need a killer product to make the network spin

Even if networked products are a special type of product, there is no way around it: to make a networked product work, you need a killer product. As for every other product category.

What I found particularly interesting reading through the book though, is that the pitfall of overestimating the value of your product proposition or diluting it is even more common in networked products. This is because adding new features could be both a way to launch a new network, or to respond to a competitor entering the market.

Keeping in mind what the core of the network is and delivering value to it is vital in all the phases of a networked product.

• When you start it and need to make sure users understand what the network stands for,

• When you grow it and need to make sure all new nodes of the network increase the overall value and do not dilute it,

• When you defend it and might want to add functionalities to respond to competition.

A couple of real-life examples. When Google launched Google + it did not really make sure that the atomic network was stable enough (aka having a solid product proposition and alignment of the network) before working on acquisition throughout its other channels. This resulted in Google pumping traffic towards a product that had not proved its market and network fit yet. We all know how the story ended. Lesson learned: you can’t use a network to power a new product if that product is not good enough.

Another more subtle way of diluting the product value is when an established networked product in the market develops features as an answer to a competitor that is trying to win over one part of the network. One example here is TikTok creating an experience centered around video and trying to win over a part of Instagram and Facebook network. As a response, Instagram and Facebook created stories, a less developed proposition to try to address some users’ needs. This tactic can work to create some early traffic to the feature, but many times it mainly dilutes the core proposition and adds clutter, leading in the long run to a worst design. Lesson learned: choose carefully what you decide to add or not to your product, your main proposition might suffer.

Anyone else who has read the book and wants to share other lessons or takeaways? I am really happy to chat!