The Cold Start Problem by Andrew Chen, published by Harper Business in 2021, is a book about network effects: why products like marketplaces, messaging apps and collaboration tools become more useful as more people join, and how to get one started from nothing. Chen is a general partner at the venture firm Andreessen Horowitz and was previously a growth executive at Uber. The book draws on close to 200 interviews with founders and teams from companies such as Zoom, Slack, Airbnb, Dropbox and Tinder.
What is The Cold Start Problem about?
Every network product faces the same trap at launch. A chat app with no friends on it, a marketplace with no sellers or a dating app with nobody nearby is useless, so new users leave before the network can form. Chen calls this the cold start problem and argues that most network products die here, long before competition matters.
He organises the book around a five-stage framework. The first stage is the cold start itself, where the goal is an atomic network: the smallest group of people that makes the product worth using, such as one office team on Slack or one campus on Tinder. Next comes the tipping point, where you repeat that atomic network until the product spreads on its own. Escape velocity follows, when network effects push growth through three channels: acquisition (users invite users), engagement (more people make the product more useful) and economics (bigger networks earn more per user). Then the ceiling, where growth stalls because of saturation, spam or rising costs. The last stage is the moat, where network effects protect you against competitors.
Each stage mixes stories and tactics. Chen describes paying early drivers to guarantee supply, inviting users in tight groups, building single-player tools that work before the network exists and fighting rivals who target your best users.
What are the main lessons for founders?
- Find your atomic network. Define the smallest useful group, such as one team, one neighbourhood or one school, and make it work there before you spread.
- Win the hard side first. In most networks a minority creates the value: sellers, drivers, hosts, creators. Court them with money, tools or status.
- Give people a reason to come before the network exists. A useful single-player tool, like a photo filter or a notes app, attracts users who later form the network.
- Launch in tight clusters. Invite-only launches and campus-by-campus rollouts make each new user land in a lively room instead of an empty one.
- Expect ceilings and plan for them. Saturation, fraud and falling engagement arrive as networks grow. Treat each one as a new problem needing a fresh tactic.
Who should read The Cold Start Problem?
Read it if you are building a two-sided marketplace, a social or community product, a collaboration tool or any product that gets better with more users. It is especially useful for founders before launch and in the first year, when the empty-room problem is most acute. Growth teams at larger companies will also find the later chapters on ceilings and moats useful.
Founders of single-user software, services or physical products without network effects can skip most of it.
How do you put The Cold Start Problem to work this week?
- Write down your atomic network in one sentence: who, how many and where. Make it small enough to fill by hand.
- Identify the hard side of your network and list five things that would make the first 20 of them show up.
- Pick one launch cluster, a single office, course or neighbourhood, and invite people there in one coordinated push.
If you want to test the idea behind your network before building it, the free Idea Validator on Fe/male Switch checks problem, market and willingness to pay.
Where does The Cold Start Problem fall short?
The book leans heavily on Silicon Valley success stories with large venture budgets, and many tactics, such as paying the hard side, assume deep pockets. Failed networks get less attention than winners, so it is hard to tell how reliably the tactics work. It also says little about regulation, trust and safety or the social costs of networks at scale. Read Platform Revolution for the business model side of platforms and Blitzscaling for the funding logic behind racing for scale.