Network Effects, The Modern Flywheel Propeller
Exploring what a network effect really is, and the implications associated with it
Looking back on 2020, I’m thrilled that I’ve been able to keep myself writing as long as I have. Next year I’d like to continue to improve my research and writing skills, in addition to minimizing my procrastination. A lot of these don’t actually get written until a few days before (or even day of). Thank you all for reading some of these, and bearing with me as I feel them out. I’ll be taking the rest of the year off after this, but plan to explore this week’s topic even further once 2021 rolls around in a recurring series. Thank you all for reading with me this year!!
---
Network effects is an extremely complex phenomenon that is also fairly intuitive, on the surface. Per Investopedia, a “network effect is a phenomenon whereby increased numbers of people or participants improve the value of a good or service“. A pretty straightforward example of the power of network effects is Etsy. Etsy is powered by a very wide array of different artists and vendors, which supply the website with an even wider array of goods, ultimately leading to a wider array of customers who have a simpler time finding what they are looking for. The growth driven by the network effect makes Etsy more powerful and useful.
The phenomenon has really only taken off over the last 20 years as a result of the internet and distributed users who find value from a site/product. However, it stems from Metcalfe’s Law, a theory about the effect of a telecom network being proportionally related to the amount of people who use that network, which was postulated in 1993 (though first presented in 1980).
The intuitiveness of the effect can be seen pretty easily in the diagram here. The top part of the image shows that when only two telephones exist, there is only one network available. When five exist, there are 10 connections available. On the bottom, a diagram showing 12 telephones illustrates the 66 connections that are possible. So on and so forth. While telephones illustrate the ability to talk to a wider network of phone owners, other products and their network effects can add value to the product /in addition/ to the user. Of course it relies on phone owners also being phone users.
But not all networks are the same. A blog on a16z’s website covers the distinctions pretty well. For example, network strength can be evaluated based on whether or not “users/inventory are commoditized or differentiated”. In a nutshell, two great examples are used by the writers. First, companies like Uber and Lyft offer a supply of “interchangeable”, and therefore commoditized, drivers. The strength of their networks asymptote, or essentially plateau, over time until they have found diversification in offerings (e.g. UberEats or Lyft Health). Alternatively, the second option of a more diverse supply/inventory would be more along the lines of Airbnb, where the diversity of offerings “suits unique preferences of the customer”. With Airbnb, you can find every single available stay for $150-$350 in NYC, but also stays for $250-$500 in NYC that also have a water-facing view. Both options show more than just a standardized set of rooms, akin to the dilemma of Uber and Lyft. Similarly, it can be distinguished by the contribution levels of the user. E.g. a vendor on Etsy who is adding new art/work/etc. daily is more valuable to the network than a vendor who adds something new every month.
Modern network effects are so scalable and fast-paced because the cost of onboarding and serving an additional user is close to zero. When the software or algorithm is in place, much of the actual bumps in the road to growth actually fall upon the users/inventory (e.g. the Uber or Lyft drivers). Hence the power of the network effect.
Ben Thompson wrote an excellent article that explores a different view, which focuses on the type of impact that network effects have. As you can see in the image below, network effects of some of the largest companies in the world can vary pretty heavily on the internalized vs. externalized spectrum. Facebook, being on the internalized polar end can be described as follows:
For Facebook the network effect that matters is users — a social network’s most important feature is whether your friends and family are using it. This network — given it is the product! — is completely internal to Facebook
Makes sense. The users of Facebook make it more powerful because you go to Facebook to see what others are doing. The more people that do that , the more Facebook grows and gets better as a product. On the opposite side, there is Microsoft, which goes with the quote below, indicating that as Microsoft’s suite of products become “must-haves” (think of the business or education world without the Microsoft Office suite) and expand externally by the product.
Microsoft, befitting the point I made above about the expansiveness of its ecosystem, has the most “externalized” network effect of all: there is very little about Windows, for example, that produces a network effect (Office is another story), but the ecosystem on top of Windows produced one of the greatest network effects ever. Developers had to develop for windows, and the stickiness of using windows software grew exponentially with each user
At the end of the day, network effects is the phenomenon of an experience getting incrementally better with each new user/joiner. Many of the biggest and best companies today have exemplified the concept, and can attribute their success to it. Facebook, Snapchat, TikTok, Github, YouTube, and so many more. To explain network effects is a difficult task because they vary widely by company and effect, however the commonality lies in the flywheel of making an underlying product better as people continue to use it.