Network effects

Network effects change what winning looks like. In a normal business a rival with a better product takes customers. In a network business the rival must also assemble the people, and an empty network with a better interface loses to a crowded network with a worse one. Uber’s 2019 filing said it sought the largest network in each market because it expected liquidity to support a margin advantage.

They also change how much a company can rationally spend early. Uber’s 2019 filing said incentives to attract users on both sides could depress margins until a marketplace reached enough scale to reduce those incentives. The subsidy pays for density, and density is what later makes the product cheaper to run and better to use. A company that funds growth this way is betting that the effect exists. Investors who accept the bet without checking for it fund a business that only looks like a network.

They make reach the thing an acquirer buys. Facebook’s release offered WhatsApp’s shareholders and employees $4 billion in cash, $12 billion in stock and $3 billion in restricted stock units, with the shares and units together representing 7.9% of Facebook’s shares. The case for paying that was a user base that, in Facebook’s words, was on a path to connect 1 billion people.

Finally, the effect explains why the same products are hard to displace and easy to lose. When people leave, the product gets worse for the people who stayed, which pushes more of them out.

The effect runs backward. Myspace lost more than half of its U.S. visitors in thirty months, and each departure made the site less useful to the friends who remained. A network at its peak looks unassailable because everyone in it is a reason for everyone else to stay, and the same fact makes the exit self-reinforcing. Size alone is no protection once a better place for the same people exists.

The effect can also be weaker than the strategy assumes. Uber’s 2019 filing identified low entry barriers, low switching costs, and well-capitalized competitors in major regions as risks; it also said the company lowered fares and paid driver incentives to compete. The filing reported significant losses since inception. With low switching costs, riders and drivers can try rival apps, and a rider already waiting an average of five minutes gains little from the five-hundredth driver in the city. Uber believed its network could yield a margin advantage; the filing described that as a belief, not a lock.

Finally, the arithmetic behind the effect can mislead. A square-of-users estimate can make rapid growth look attractive even while current profits are weak. Briscoe, Odlyzko, and Tilly argued that value rises more slowly and warned that overestimating the effect can support excessive growth spending. The studies disagree, so a company should show in its own retention and engagement data that existing users get more out of the product when the network grows.