The term network effects gets thrown around a lot, particularly when describing tech platforms. However, only a select number of businesses experience genuine network effects, and not all network effects are created equally. A recent video by venture capital firm Andreesen Horowitz sheds some light on this topic, and provides useful information to build out a framework for understanding the economics of businesses, and those which do, and don’t, benefit from network effects.
If we start with the definition of network effects, it refers to a situation where as a network gains users, that network becomes more valuable to all users. A well-referenced example of a network is that of the telephone. The very first telephone had basically no utility until a second telephone came into existence –that way a call could be made. As telephones proliferated, all users had more options of people they were able to contact, and the value of the network increased. In its most simplistic (and perhaps oversimplified) form, network effects have been said to follow Metcalfe’s law, whereby the network effects are proportional to the square of the number of users in that network, n2.
Source: Andreesen Horowitz
Uber (NYSE: UBER) has been described as a network business, and the enormous $82 billion valuation its IPO was priced at suggests that at least some portion of the investor base believes that Uber has strong network effects that will help the business to at some point reach profitability. But this is where following Metcalfe’s law provides a parochial and oversimplified representation of Uber’s network effects. While Uber does benefit from network effects, they are relatively weak for the business as it stands today, and have arguably plateaued.
Let’s consider Uber when the platform was in its infancy. As more drivers were added to Uber’s network this led to greater coverage and reduced wait times for users, which led to more users joining and incentivised more drivers to join. In other words, driver supply reinforced user supply, and vice versa. By adding more drivers to the platform, Uber could reduce wait times from say 20 minutes to 5 minutes. While this is a genuine network effect, it is not one that can continue unrestrained. At a certain point the wait times are short enough such that any additional drivers joining the platform add minimal incremental value to existing Uber users. Uber’s scale after a point has rapidly diminishing marginal returns.
Source: Andreesen Horowitz
The problem for Uber is that once 5 minute wait times are achieved, riders become ambivalent as to what ride-hailing platform they use. If you were to remove any branding from the vehicle, could you really tell the difference between an Uber vehicle, and one from Lyft? The issue is that once multiple ride-hailing platforms reach an adequate level of liquidity –in this sense the ability to pick up riders within 5 minute windows –it becomes more difficult to retain users, and the specific platform becomes less important to users. As D’Arcy Coolican and Li Jin from Andreesen Horowitz note, Uber needs to compete along additional vectors that go beyond merely having a network, such as price, brand strength, user experience, and loyalty programs.
There is significant network overlap between Uber and Lyft in a number of markets, and this fact, combined with the relatively low switching costs for riders, means that it is unlikely that the intense competition will abate going forward. The corollary of this is that the economics for Uber could remain poor, or even degrade from here if retaining riders or drivers requires additional subsidies and incentives.
The Uber example demonstrates how not all networks are created equally, and highlights the perils of gaining a false sense of security around the defensibility of a business on the basis that it experiences network effects.
George Hadjia is a Research Analyst with Montaka Global Investments. To learn more about Montaka, please call +612 7202 0100.