Book review: The Cold Start Problem

I wanted to read this book to understand more about network effects. In the Judge Business School course I did, there was a heavy focus on strategy and how we’ve moved from Porter’s Five Forces (where you analyse the competitive forces within a particular industry to try and achieve dominance) to a more nuanced view that considers your strategy in a much wider ecosystem. As a quick example of that wider ecosystem, Apple may have started as a high-quality manufacturer of computers, but now that network effect extends out to phones, music and TV. Apple’s moat is its strong network effect (”come for the tool, stay for the network” rings true as a family with an Apple subscription).

The network effect is often described as “the more people use your product the better it becomes” or by Metcalfe’s law “the financial value or influence of a … network is proportional to the square of the number of connected users of a system”. Both these statements might well be true, but they are difficult to turn into something concrete. The Cold Start Problem breaks that down into something more actionable.

So, what is the Cold Start problem? Well, it’s exactly what it says on the tin. It’s that starting phase when you’ve not got a network. It’s that moment before your social network has users. Before your photo site has photos. Before anyone has used your code sharing platform. Once you’ve fixed that cold start problem, you reach a tipping point. The Allee threshold where your network population is strong enough to be self-sustaining. If you’ve hit upon the right idea, you might then hit escape velocity with super rapid scaling. And then finally, you hit the ceiling. You’ve captured the market and growth has stalled (for now?). Finally, you’ve got a moat. You have a network that provides a long-lived source of competitive advantage (perhaps this is where enshittification begins!).

I found a lot of value in the book through the stories Andrew shared. The insights into Uber’s growth were fascinating. For example, the analysis of how many drivers you need to reach a critical mass in a city. Too many and drivers don’t have any fares. Too few and passengers wait too long. The balancing out of incentives (give/get incentives to recruit or $500 for new drivers) against demand was fascinating. Being on the front-line as Uber expanded world-wide must have been an experience and a half (not all good!).

Another concept that clicked was an “atomic network”. A self-fuelled network. With something like DropBox it’s file sharing between two people, with Slack it’s a team of three folks. You don’t have to have grand ideas - you might not be trying to connect every single user on the planet, perhaps it’s just comic book sellers, or some other niche (see Paypal). Once you’ve established that network you’ve reached a self-sustaining state and you can look at ways in which that can start to scale. Perhaps it’s connecting other groups of atomic networks together? For example, Slack made a PLG motion and was often in multiple places within an organization.

The Law of Shitty Clickthroughs was also enlightening, particularly for someone like me that has never worked in the marketing space. “Over time, all marketing strategies result in shitty clickthrough rates”. This is kind of obvious in retrospect, as humans we learn that ads are (mostly) useless and ignore them. And then we install ad-blockers. I have a “Promotions” tab in Gmail. I can’t remember the last time I read something in that tab, I just ignore it. A network might provide that but equally overusing it will lead to its inevitable demise. (perhaps this is another enshittification example).

Finally, we come to the ceiling / moat part of network growth. There comes a certain size when supporting the network is really hard. For example, you might be plagued by spam and find it almost impossible to deal with. The folks on the network might not reflect the original aims of the group etc. I’d guess Twitter/X is currently in this state and it’ll almost certainly be a case-study in the future.

I’d recommend this book as I think it gives a great set of stories that expand your range of thinking.