Have I been using the “Network Effects” term incorrectly all this while?
Over the past one year of hearing entrepreneur’s pitches day in and out, there is one buzz phrase that every pitch definitely contained, “over time network effects will kick in” or “ at scale network effects will make us defensible”. Although I kind of understood what they meant, this weekend I kind of dug a little deeper into what “network effects” actually meant. As it turns out my confusion arose because people kept on using it wrong all along.
Along with network effects, there were other words like “platform”, which I used on a daily basis but hardly knew what exactly they meant.
Sharing a few of my findings on these terms, what they really mean and how to achieve them.
What is a network effect?
In simple terms, network effects occur when value of the product/service increases as and when more number of people use it. In not so simpler terms, it is the surplus that an agent derives from the product when then number of agents consuming the same kind of product changes.
Network effects were first studied in 1970s in the context of long distance telephony, but today have become the critical reason behind success of many software based companies.
Network effects and virality are not the same thing
Even though used many a times interchangeably (I am also guilty of that occasionally), network effect is not the same thing as virality. Virality happens when the product’s use grows through direct user to user transmission.
One of the best ways to understand the difference is by looking through examples, as one explained by Sangeet of Platform Strategy here:
Direct network effects could be one of the ways to achieve viral growth, but is not the only one. Others might include incentives (for example, the referral programs run by Ola/Uber where every new referral gives us a free ride), word-of-mouth (when I tell my mother that Ola/Uber is in fact cheaper than normal taxis for all distances >5kms) or casual contact (for example, the signature “Sent from my iPhone” on iPhone mails).
Companies can use either of these viral mechanism to grow rapidly but that doesn’t mean that the product/service has network effects built-in. While incentives, word-of-mouth and casual contacts are traditional virality mechanisms, network effects lead to inherent product virality.
A product with viral growth means that the product has lower CAC (Customer Acquisition Cost) while one with network effects has high engagement and repeat rates. Both are definitely not the same thing.
Scale and network effects are also not the same thing
Just because a product has scale doesn’t mean it has network effects. When a product has scale, it means the cost of producing one unit is decreased (as fixed costs are distributed over a large number of products), so the product is cheaper. But if product has network effects, product is more valuable to its users.
Types of networks and network effects
Networks and their types
Network is basically a collection of nodes, and a lot of research has been done to study them already. One of the easy resources I found on them were videos by Albert Laszlo Barabasi (like these), which also generated my interest in his book called Network Science.
Based on these nodes and their properties, networks can be:
- Homogenous or heterogenous: Homogenous when users are of the same kind (for example messaging or telephone) or homogenous when users are of different kind (for example, Flipkart with sellers being one of nodes and buyers being another)
- Degree and clustering co-efficient: Based on degree (number of connections to one node) or clustering co-efficient (measure of closeness of nodes), networks can have different properties. For example, Facebook user on average has ~980 connections, while Whatsapp user has ~20 connections. Clustering of networks hugely changes the way users behave/engage on these networks.
- Unidirectional or bidirectional connections: Bidirectional connections (like the friend connections on Facebook) or unidirectional connections (like Twitter/Instagram followers) also deeply affect the network properties.
Direct and Indirect network effects
Direct network effects occur in increase in usage leads to direct increase in value (example, telephone or Skype); while indirect network effects occur when increase in usage leads to production/usage of a complimentary product which in turn leads increase in usage of original product (for example, a VR headset becomes much more valuable if variety of VR content sites increase, and this variety increases as number of VR headset users increase)
Local network effects occur when instead of the entire use base of the network, behaviour of the user depends on a small subset of nodes. The extent of clustering in the network as well as the extent of information each customer possesses may become relevant in this context.
One very important classification I found on the slides by Anu Hariharan of Andreessen Horowitz were between Network, Marketplace and Platform. (I always used product and platform interchangeably before this, but this definition makes much more sense)
Network: General term for interconnected group of people/nodes
Marketplace: network where transaction happens between two sides with distinct set of users on each side
Platform: Network of users and developers, a multi-sided feedback-loop between users, developers and the platform itself (for example, Facebook in its current form)
Strategies for building network effects
Entry strategies:
- Creating Autarky value: Value received by users can be divided into two types: Autarky value (value derived even if there are no other users) and Synchronisation value (value derived from being able to interact with other users). One of the very efficient ways to get early users, is to enable what Chris Dixon calls Single Player Mode. Medium and Instagram did very efficiently by first acting as a tool to write articles and add photo filters respectively, transitioning from a tool to a social network eventually. Users come for these tools but stay for the network.
- Bowling Pin Strategy: One of the strategies pioneered by Facebook, by starting with Harvard before opening to other colleges and then to everyone. Conceptualised by Geoffrey Moore in his book, Crossing the Chasm. He talks about finding a niche where the chicken-and-egg problem is more easily overcome and then find ways to hop from that niche to other niches and eventually to the broader market. This also leads to of what Chris calls exploring irregular network topologies.
- Seeding early content/users: Reddit co-founder Steve Huffman admitted that it was initially seeded with fake profiles posting links to simulate activity. Seeding initially was a non-scalable strategy used by many scalable products today. They key is to seed exactly the type of content you would like to see on the product eventually. Also helps if your initial set of seeders are influencers, who attract a lot of other users. Dating app Aisle used it very efficiently by starting off with offline mixers and inviting celebrities and models on the app as well.
Growth strategies:
- Setting goals for critical mass : Facebook realized early on that it was important to connect each user to atleast 10 friends for them to stay engaged.
- Use triggers and rewards to create habit forming product: Nir Eyal has come up with a great framework to build habit-forming products.
3. Subsidising the harder side of marketplace: More often than not, one of the sides of the marketplace is more harder to acquire than the other. One of the ways to acquire them is use the strategy Olx (advertising only the seller use case) and TrulyMadly (advertising only for girls) used. other could be to subsidise the harder side, the way bars frequently have ladies night.
Bibliography:
http://wwwpub.utdallas.edu/~liebowit/palgrave/network.html
http://platformed.info/virality-viral-growth-network-effects/
http://platformsandnetworks.blogspot.in/2011/07/business-model-analysis-part-5-virality.html
http://platformsandnetworks.blogspot.in/2011/07/business-model-analysis-part-2.html
http://cdixon.org/2009/08/25/six-strategies-for-overcoming-chicken-and-egg-problems/
http://a16z.com/2016/03/07/all-about-network-effects/
http://oz.stern.nyu.edu/io/network.html
https://www.youtube.com/watch?v=NCq1kpuCUq4
