Book Review: The Cold Start Problem by Andrew Chen

Along with the mythical "product-market fit," "network effects" must be the most commonly-used phrase in the pitch decks of SaaS startups. These two concepts help bridge the gap between the unremarkable today, with little to show for in terms of financials, and a glorious future, where the product in question will undoubtedly take off. Theoretically, every SaaS product is just a product-market fit and network effects away from becoming a unicorn.

Nir Eyal's Hooked was the first book we reviewed that laid out what network effects can achieve for products. Eyal's book proposed a model to understand the habit-forming dynamics that got users hooked on certain products. The hooks he talked about, such as "like," "share," and "retweet," made people come back to the product and kept the network effects humming.

But Hooked was different from The Cold Start Problem: The former studied network effects at the product level, explaining how they took over once the users developed an itch to use the product.

On the other hand, The Cold Start Problem looks at how the network effects come to be and elevate not only individual products but whole industries. In addition to its macro-level focus, the book also takes a scholarly approach and breaks network effects down into acquisition, engagement, and economic effects.

As insider as it gets

The author Andrew Chen is an entrepreneur and investor who helped popularize terms such as "viral loop," "growth hacking," and "network effects," and thus, played no small part in shaping the Silicon Valley culture as we know it today. Currently a general partner at Andreessen Horowitz, Chen led Uber's Rider Growth product teams from 2015 to 2018, during a period when the company was rewriting the book on growth before its IPO. Chen's experience at Uber provides the basis for his discussion of network effects in the book. However, he goes beyond that and conducts a series of interviews with decision-makers from other tech giants. Airbnb, Uber, Twitch, PayPal, LinkedIn, Clubhouse, and the like all have their dedicated chapters in the book, which Chen uses to dwell upon a particular aspect of network effects.

A network effect refers to a phenomenon whereby some products become more valuable as more people use them. The usual example is the telephone, which would be useless if you were the only one in the world using it. A second telephone owner that you know about would at least make your device usable. However, for this invention to realize its full potential, a large group of phone owners, out of whom you would have a few that you would like to talk to every day, would be necessary.

Without securing a certain number of users, networked products like marketplaces, ride-sharing apps, or social media platforms will find it difficult to convey their value proposition to existing users. Your Facebook feed would be pretty dull if you had only one contact. Would Twitter be the same menacing and addictive platform that it is now if there were only a handful of people on it?

I'll see your Metcalfe and raise you… a meerkat!

The book's title is a clear allusion to the trouble of getting your car started on a cold winter morning. It cleverly captures the challenge of getting your user base to a size where it can support itself before your existing users get fed up with the lack of interaction and disperse. This challenge of getting your network over the hump is the problem that Chen tries to solve in his book.

In the startup world, you are advised not to question how a startup got its first hundred or thousand customers because you wouldn't want to know as anything goes in that process. Chen's book is about developing a framework that startups can employ to build that initial network that can sustain itself without collapsing. His source of inspiration, however, is quite uncommon.

Usually, any discussion of network effects centers around a discussion of Metcalfe's Law. Chen does away with convention, arguing that Metcalfe's law doesn't say anything about the quality of the engagement and that more nodes are not necessarily better. Instead, he turns to Warder Clyde Allee, a professor of ecology, and the study of population dynamics for inspiration. What Chen formulates as Meerkat's Law tells us that there is a tipping point—called an Allee threshold—in animal populations beyond which a group becomes more defensible against predators, until the population hits the carrying capacity of the ecosystem in question.

This analogy helps readers understand that a critical mass is needed to make a network self-sustaining. Before that tipping point, there is not enough interaction, and hence value, to justify staying in the network for people. Once that tipping point is reached, everything starts to click, and the organic interaction within the network creates enough value to keep people engaged and make them stick around. The Cold Start Problem, in a sense, deals with the problem of "crossing the chasm," "overcoming the trough of sorrow," or "going from zero to one," in startup talk.

Key concepts from the book

Andrew Chen makes use of some key concepts to get his point across throughout the book. Here are three of them:

Atomic network: "The smallest possible network that is stable and can grow on its own." For example, for Slack, this was a team of three. Members of an atomic network get to enjoy the product's value proposition, so they will stick around. Their engagement rate will increase, which, in turn, will start driving the network effects.

Once a healthy atomic network is built, the next thing is to copy and paste the formula until the tipping point, where the network begins to thrive. This is where "growth hacks" like viral videos or referral programs involving sign-up bonuses work wonders. What happens when you don't have atomic networks? Your product fizzles out as Google+ did, despite all the initial fanfare, the media blitz, and the 90 million users announced at the beginning.

The hard side of the network: "The small percentage of people that typically end up doing most of the work within the community." These are the busiest drivers in the case of Uber, people who pull in millions of views with every video they post on Youtube, or power gamers watched by millions of other gamers on Twitch.

People making up the hard side of a network do tasks that require more work in their respective networked products. Therefore, satisfying their needs and improving their user experience is critical for the network effects to kick in. When these stars are unhappy and leave the network, they can take a big chunk of the user base with them, which can set off "anti-network effects" and trigger the unraveling of the whole network at some point.

Moat: This term refers to the levers a network can use to defend itself. However, contrary to popular belief, Chen makes an important point that networks do not have an inherent defense capability resulting from merely being a network. In the market, the network that more effectively puts into action acquisition, engagement, and economic effects tends to win. While the smaller network wants to capitalize on niche segments, the larger network takes advantage of higher monetization to fend off newcomers.

Features cannot serve as moats as they are easy to copy. It's nearly impossible to copy a network with its intrinsic value and engagement. That is your moat.

Final thoughts

The Cold Start Problem has all the characteristics of a great book: A cleverly chosen title, a clear theoretical framework, and real-life examples to illustrate how that theory is to be implemented. Andrew Chen takes the time to explain how businesses succeed when correct levers are pulled and drives his point home by discussing the instances where even the giants falter when they don't pull the right lever.

Coming from a Silicon Valley veteran, The Cold Start Problem is a breath of fresh air, offering explanations instead of slogans as we are used to seeing from other Silicon Valley greats (*cough*Peter*cough*Thiel). It is a must-read for anyone planning to build or market a networked product.