FrameworkLessonGrowth and business modelsFounders and companiesTechnology
Network effects start in one small market, never across the whole one
Andrew Chen · The Cold Start Problem
The idea
Chen's book is the operational sequel to the theory of network effects. Networks do not launch nationally. They start in an atomic unit small enough to reach density, a campus, a city, one category of seller, and only once that unit works on its own do they expand to the next. Most networks die before the first unit tips, and most that die did so by spreading too thin. For an investor the implication is to ask where the first dense market is and whether it is self-sustaining, which is a sharper test than the size of the eventual one.
Judge a network by whether its first small market is dense and self-sustaining, not by the size of the market it aims at.
When it works
Early-stage diligence, growth strategy questions, and any marketplace that reports total users rather than density.
When it may fail
Some products have no atomic network and grow by direct value; forcing the framework on them produces false worries. And a dense first market can still fail to generalise.
How to use it
Ask for metrics for the first market alone: liquidity, repeat rate, the share of transactions that would not happen without the platform. If the company only has blended numbers, that is the diligence question.
Where it came from
Book The Cold Start Problem · Andrew Chen
What the source says. How a network is actually started, city by city or campus by campus, and what usually kills it before it tips.
Connected ideas
Worth knowing first
A platform sets rules for a transaction it does not performName the two sides, say which one pays and why, and describe exactly how one side's growth helps the other.
Supports
Competition and profit rarely occupy the same marketAsk how a business escapes competition before asking how large its market is.
Says something similar
Markets now sort winners from losers faster, and profits concentrate among fewer companiesGrowth is arriving faster and concentrating harder, so a company's trajectory in its first year tells you more than it used to.
Concepts
Put it to work on L3VLUP
More like this
- Company insightGrowth and business modelsLetter located
Forgo the revenue you could capture if the platform grows more by letting others earn it
A platform that lets others earn on it grows faster than one that captures every dollar; the forgone revenue is the investment.
From Shopify's 2015 IPO letter from Tobi Lütke · Tobi Lütke
- Decision principleFounders and companiesLetter located
Given a choice between flattering earnings and future cash flows, take the cash flows
Tell investors in advance that you will optimise long-term cash flows over reported earnings, then do it, so the market can price the choice rather than be surprised by it.
From Amazon's 1997 letter to shareholders · Jeff Bezos
- Case studyAccountingBook
Growth consumes cash, and a profitable company can still run out
Profit is not cash, and growth widens the gap; model the cash a business needs to grow before admiring its sales.
From Shoe Dog · Phil Knight
- Company insightArtificial intelligenceLetter located
Inference has overtaken training, and the flywheel runs on tokens
A compute vendor's case is a flywheel from agents to tokens to compute; the buyer's case is the return on that compute, and both have to be read together.
From NVIDIA's fiscal 2026 letter from Jensen Huang · Jensen Huang