The Cold Start Problem

Updated 2026-04-10

The Cold Start Problem, published in January 2022, is Andrew Chen’s book about the mechanics of network effects. Chen wrote it because he felt his own understanding of network dynamics had been “unforgivably shallow” and because the field lacked a precise shared vocabulary.

Core Ideas

Five Phases of Network Growth

Chen structures the book around five phases:

  1. Cold Start - the zero state: one user joins, but none of their contacts are already in the network
  2. Tipping Point - the threshold at which the network starts to sustain itself
  3. Escape Velocity - growth becomes self-reinforcing
  4. Ceiling - natural growth begins to flatten
  5. Moat - the network itself becomes the defensive barrier against competitors

Three Forces

Instead of treating “network effects” as a single concept, Chen distinguishes three separate forces:

  • Acquisition Effect - the network enables viral, low-cost user growth
  • Engagement Effect - interaction increases as the network gets denser
  • Economic Effect - monetization and conversion improve as the network matures

Atomic Network

The Atomic Network is the smallest self-sustaining unit of a network, the minimum cluster that works without external support:

  • Slack: 2,000 exchanged messages inside one team led to 93% retention afterward
  • Uber: one single city with enough drivers and riders
  • Facebook: one single campus

The term lets you break a global-sounding problem down into the first real task: how do I make this one cluster work?

Hard Side

Every network has a “Hard Side,” the minority, often around 5%, that does most of the work and is hardest to acquire:

  • Wikipedia: a small core of prolific contributors
  • Uber: drivers
  • YouTube and Instagram: top creators in a power-law distribution

The strategic question becomes: who is the hard side of my network, and how do I win them?

Come for the Tool, Stay for the Network

Many successful networks begin as single-user tools and only layer in network effects later, because attacking the cold-start problem head-on is so brutal.

Early Version of Enshittification

As networks mature, they often shift from growth to extraction. Chen does not use the term “enshittification”, Cory Doctorow coined it in January 2023, one year after the book, but he describes the same pattern: switching costs take over the retention work that the product used to do, and the interests of the platform and its users diverge.

Growth Hacks and Scrappy Plays

Chen documents many concrete early-stage tactics:

  • PayPal: a $5 referral bonus, one of Silicon Valley’s most cited growth hacks
  • Substack: paid prominent writers such as Scott Alexander to bring their audience, “buy the chicken”
  • Airbnb: automatic Craigslist cross-posting through bots
  • Reddit: the founders seeded the front page themselves with fake accounts
  • Tinder: campus-party seeding, with women as the hard side in dating apps

Sources