Network Effects are magical. They are the pixie dust that makes certain Information Technology businesses, especially on the Internet, into juggernauts. They can be found in both consumer and enterprise companies. Network Effects are special because they:
- Provide logarithmic growth and value creation potential
- Erect barriers to entry to thwart would-be competitors
- Can create “Winner Take All” market opportunities
Network Effects are like a flywheel–the faster you spin it the more momentum you generate and enjoy. But not all markets lend themselves to Network Effects. They are not the same as Economies of Scale where “bigger is better.” To be certain, Economies of Scale can give strong competitive advantage and defensibility to the first to get really big (or Minimum Efficient Scale as the economists call it.) For example, SAP and Oracle benefit from having massive revenue bases which enable them to employ armies of engineers who develop rich feature sets and also to hire huge sales forces. However large these companies are today, though, their growth rates, especially in their early years, were far more modest compared to those Network Effect companies whose growth resembled a curved ramp off of which they launched into the stratosphere.
There are four main types of Network Effects:
- Classic Networks, in which the value of a product or service increases exponentially with the number of others using it. Communications networks like telephones, fax, Instant Messaging, texting, email, and Skype are all examples. Metcalfe’s Law captured this as a simple equation where the Value of a network = N², where N is the number of nodes. Typically, each node in a classic network is similar to each other and possesses both send and receive capabilities. This will become clear juxtaposed against the other network effects below where there are different types of nodes. Other examples of classic Networks are social networks (eg Facebook) and payments (eg PayPal).
- Marketplaces, where aggregations of buyers and sellers attract each other. Lots of sellers means variety, competition, and price pressure, which all serve to attract more customers. And because the customers flock, more sellers are enticed to participate in the marketplace. eBay, stock exchanges, and advertising networks are all examples. One nuance of marketplaces, however, is they differ in terms of the scale required for acceptable liquidity. For example, ad networks can achieve sufficient reach and liquidity at relatively low levels which is why you see thousands of online ad networks, where they each exhibit network effects but not in a winner take all fashion. Stock exchanges and payment networks require far greater scale for network effects to operate, which is why you see much greater concentration in these industries.
- Big Data Learning Loops. “Big Data” is all the rage in techland, but just having gobs of data is not necessarily a Network Effect, nor any sort of competitive advantage per se. What you really need is unique data and algorithms that process that data into insights which then lead to decisions and actions. A flywheel effect comes when you get a critical mass of data that you mine for insights; pump that value back in to your product or service; which attracts more users which get you more data. And so on. Venrock portfolio company Inrix is a good example, where they mine GPS data points to derive automotive traffic flow data. The more commercial fleets, mobile app users, and car companies they can get data from, the better their traffic analysis becomes, which gets them more users and hence more data. They turn data into an accuracy advantage that earns them the right to get even more data.
- Platforms are a very special and powerful form of network effects. In Information Technology, a true “platform” is where other developers build technology and businesses on top of your technology and business because you offer them one or more of the following:
- Lots of users/customers, and you represent a distribution opportunity for them
- Compelling development tools, technology, and (sometimes) advantageous pricing
- Monetization opportunities
Example include Operating Systems like Microsoft Windows, Apple App Store, and Amazon Web Services.
Each of these four types of network effects can be extremely powerful on their own. Yet, even more power is derived when a business can harness multiple types of network effects in synergistic ways. Google, Apple and Facebook do this for sure, but a less well known example is Venrock portfolio company AppNexus that operates a real-time online advertising exchange and technology platform. The exchange aggregates advertisers, agencies, publishers and ad networks for marketplace liquidity, but also offers a hosting and technology platform for other AdTech companies and ad networks to augment their own businesses. And the vast troves of data AppNexus processes every millisecond flows back into the system as optimized and targeted ad serving.
Network Effects are what you want fueling your business. Sometimes you just need to get clever about discovering and harnessing them.
Great article and exactly what I am trying to understand better to determine what kind of network effect I should project for our user growth. N^2 makes sense in early days but once an organisation reaches certain size, I am not so sure that user growth will continue to explode as N^2.
We are about to launch a browser based freemium presentation software app…..
I would appreciate if you can give me any guidance or point me to any public data that can be used as a sanity check for user growth in YR1, 2, 3…
thank you again for the post!
Good stuff, BDA!!
Instead of thinking of these as “network” effects, I’ve been thinking of them as “auto-catalytic” effects. Anytime the output of the interaction feeds the input of the interaction, you’ll experience accelerating growth. The mistake I’ve seen many people make (and I know, because I’ve made it) is to overestimate the homogeneity of the users of their product. For example, social networks explode when the users invite users just like them (Facebook). Where it breaks down is when the person-who-invites invites people-who-don’t-invite. This is common in many of the second generation social networks, built around maven personality types, who invite audiences, rather than other mavens.
Also, one nerdy comment. I dislike the flywheel analogy, as flywheels are more capacitors–smoothing energy delivery, and storing it for sudden delivery. Chemical and biological analogs are obvious, electrical and mechanical analogs perhaps not.
Thank you! It is definitely an remarkable online site!
Very nice post. I just stumbled upon your weblog and wished to say that I have really loved browsing your blog posts.
In any case I’ll be subscribing on your feed
and I’m hoping you write again soon!
Excellent post. I’m going through many of these issues as well..