Customers & Markets
How to Start and Scale Network Effects
How networked products get their first users when an empty network is worthless, and how they keep growing afterward.
The big idea
Products that connect people, such as marketplaces, messaging apps, and social platforms, face a starting problem: nobody wants to join a network that has no one on it. Chen, drawing on his work at Uber and as an investor, lays out a five-stage path from that cold start to a durable moat. The key move is to begin with the smallest network that can sustain itself, win the side of the network that is hardest to attract, and then repeat that pattern market by market.
Key takeaways
An atomic network is the smallest group of users that makes the product valuable on its own and can keep it alive. For a workplace chat tool, that may be one team. For ride-sharing, enough drivers and riders in a single city. Build that first, then copy the pattern, rather than launching everywhere thinly.
Most networks have a side that is harder to attract and does most of the work: drivers, hosts, sellers, or content creators. If they show up and stay, the easier side usually follows. Founders should invest disproportionate time and money in recruiting and keeping this group.
A network does not need to be big everywhere to work. It needs density in one place, community, or campus. Once one network tips, the next ones become easier because of the reputation and playbook from the first. Choosing where to start is a strategic decision.
Chen describes growth at scale as the combination of three effects: an acquisition effect, where users bring in other users; an engagement effect, where more users make the product more useful and sticky; and an economic effect, where scale improves monetization and lowers costs. Strengthening all three keeps growth compounding.
Growth eventually slows. Channels saturate, early users' enthusiasm fades, spam and bad actors appear, and new users may not resemble the first ones. Chen treats this as a normal stage and argues that teams must keep finding new networks and markets, each of which can feel like a fresh cold start.
A strong network makes the product harder to displace, because rivals face their own cold start while you already have density. Competition between networked products often happens market by market, and the side with the stronger local network usually wins those fights. Defending existing networks matters as much as opening new ones.
Early networks often need help. Tactics include launching with an invite-only group, offering a useful single-player tool that later becomes a network, manually doing work behind the scenes, or subsidizing one side until the network can stand on its own. These are expensive, so tie them to a clear tipping-point goal.
The framework
Build the first atomic network when the product has no users and little value.
Reach enough density that new networks form more easily, one market at a time.
Strengthen acquisition, engagement, and economic effects to sustain growth.
Address saturation, fading novelty, and bad actors as growth slows.
Use network strength to defend against competitors facing their own cold start.
Apply it this week
Read next
When to trade efficiency for speed, when not to, and how a company must change as headcount grows.
A structured way to test customer-acquisition channels instead of guessing, built around 19 channels and the Bullseye process.
How to choose the one number that matters at your stage, measure it honestly, and turn it into evidence of traction.
This page is an original summary and commentary prepared by XRAYCLOUD.AI with AI assistance. It is not affiliated with or endorsed by the author or publisher, and it is no substitute for the book — read it for the full argument.