Imagined Network Effects
Sunday, March 17th 2024
Because network effects are the holy grail of internet businesses (stat here on how the greatest percentage of value is created by companies with network effects), they can be an object of great desire by those without such effects.
However, I believe most successful network effects are part of the original DNA, designed into the product core; and on rare cases, can be bolted on later through product discovery (the most common being come for the tool stay for the community).
But in general these true network effects are either there structurally from the outset or not.
The common network effects are protocols (TCP/IP), communications (Skype, telephone), and marketplaces.
Marketplaces, I would argue, aren’t true network effects. If the definition is the utility increases with n+1 user, an additional buyer doesn’t give every other buyer higher utility. In fact it could make it worse (higher prices, slower responsivensss).
As NFX argues, asymptotic network effects may look like true network effects but incur weaknesses missing from those with uncapped advantages.
The second hidden weakness in network effects that have imagined strengths are those with multi-tenancy.
In both cases, the core value of a network effect, defensibility, can be undermined, especially when the network effect is localized as in the case of Uber.
But the greatest imagined network effect stems from a more critical flaw: unproven utility.
Network effect design, as part of the core product, still needs to adhere to the physics of successful products: customers find utility in the effect.
Often this can be disproved by simply asking customers whether they currently do not use the service without the additional utility from more users; and also ensuring in the case of n-sided networks each party truly will contribute.
Systems designed for network effects don't always lead to real network effects.
Let's use an example the use case which inspired Metcalfe's law: Ethernet.
In this case, Ethernet, and other forms of telecommunications, increases value as more people use it. The utility is directly proportional and associated with the job to be done: send packets of information to others. Being able to send more means more utility.
2-sided marketplaces, such as eBay, depend upon increasing of the cross-side actor: buyers benefit when there are more suppliers; suppliers benefit when there are more buyers.
Here, it is critical that utility remains high and it is better to have in a network.
So in the case of eBay, which was founded on the idea of rare items, had direct utility the more suppliers there were of, say, Beanie Babies or Pez collectibles. Similarly, these unique items needed a way to aggregate users from across the world; and the more buyers, the higher the prices (increasing utility to the seller).
But the existence of network effects as a mechanism does not mean that the actors will find utility in the increase number of users.
For example, Uber has asymptotic network effects: as the wait time based on the supply of drivers crosses around 5-minutes, each additional driver likely decreases utility (perhaps increases it marginally by lowering prices). There's also a negative correlation: the more Riders, the more competition, which could attract Drivers but could also result in longer wait time because of competition.
The second contributor to effective network effects