Uber’s Network Effects
https://www.nfx.com/post/the-network-effects-map-nfx-case-study-uber/
Network effects play a role in many modern business models, especially those that rely heavily on technology or media. The attached article discusses how Uber, the leader in mobile ride-hailing, relies on network effects in multiple ways. These network effects are not only applicable to the buyers (riders), but also to the sellers (drivers). Users are motivated to use Uber because they know it is a well-established company that has a large service of drivers, especially if you are in a metropolitan area. The drivers feel motivated to remain or become Uber drivers because they know Uber has a large user base, and drivers know they can get business, especially if they too are in metropolitan areas. One network effect is that adding a new user gives potential drivers more motivation to join, as they have a larger client base. Adding another driver gives users more motivation to sign up, as they become more likely to receive good service (shorter wait times come with more drivers). This is a more advanced type of network effect than we have discussed in class, as it relies on two z-values, z1, and z2, where z1 represents the proportion of the population who are drivers, and z2 is the proportion of the population that is a user. We would need corresponding x1 and x2 values as well, to stay consistent with our original model.
The article also discusses how the benefits of the network effects of Uber are approaching a limiting value. Even as the population of drivers continues to grow, the expected wait time for users can only get so low–an average wait time of 0 is unrealistic. Sticking with our model, this would limit the growth of users, as the network effects come to a halt, which would in turn limit the growth of drivers. Typically, in class, we used f(z)=Cz where C was some positive constant and z was the proportion of the population using the item, to represent the item’s network effect. However, this was a naive model, and from the article, it seems like reality tells a different story. f(z) is closer to some logistic function of z. For small values of z, it makes little difference how many people use it, but eventually, there will be exponential growth once it meets a certain point, and then the growth will level off once most of the people in the target audience has made up their mind about whether they adopt the technology. This is just a proposed adjustment to the model of one variable z. We need an even more advanced model to represent how the two different z’s account for the network effects of both users and drivers.