Does your user table look like this? (1)

Networks have been around for a long time. Think about the telephone which was invented in the late 1800s; the value of the telephone network increased exponentially as more and more users bought and started using them. Fast forward, and today you find several multi-billion dollar business that rely on these effects to drive their massive advertising engines.

Platforms that generate network effects are massively powerful ecosystems which increase in value almost magically. At tryb, we’ve been pitched by dozens of founders and CEOs who claim that their products generate network effects. Unfortunately, most the time this is simply not true.

Network effects matter for businesses because they are simultaneously a barrier to exit for existing users of the platform and a barrier to entry for the competition. As a founder or an investor, it’s important to appreciate the mechanics and nuances of network effects that are particular to any business, in order to understand what questions to ask and what the corresponding business strategy should be.

What are network effects?

In short, network effects are considered present when the value of a product or service is dependent on the number of others using it 2. In our example of the classic telephone, the value of a telephone increases the more users join the telephone network. At first, you can call yourself which is a relatively pointless activity (for most of us). However, as your friends and family members join the network, you can call them and they can call you – thus, the value of the network increases because more people have joined.

The best examples of networks in modern business are the likes of Facebook, LinkedIn and Google, which have become valuable as more users join the respective services. As more of your friends and colleagues join these services the more valuable the platform becomes to you.

Properties of a Network

For a company’s platform or ecosystem to have any sort of network at all, nodes (i.e. users) need to be connected in some way. These connections can look very different depending on the product.

Node type

Nodes within the network can be homogenous (all the same) or heterogenous (different). Homogenous nodes exist within the telephone network. Everyone using the network is a single class of user all interested in doing the same thing: making a phone call. Uber is a heterogenous network with two different classes of users being brought together on the platform. One group of users are people looking for a ride, the other group providing a ride.

Hub and spoke, decentralised and distributed networks (3)

Connections and clustering

Properties to consider when thinking about node connections are the number of connections to a single node and how concentrated the cluster is around a particular node. In a hub and spoke shaped network, the central node is powerful with most the connections centralised around the hub. Microsoft Windows is a good example of a hub and spoke network with Windows as the hub having most of the market share. LinkedIn is a good example of a decentralised network where nodes and connections cluster around central influential figures. Distributed networks are popular today with the rise of distributed ledger technology and crypto currencies. Each node participates in a near equal way.

Connection direction

Connections can be unidirectional or bidirectional. Modern examples of unidirectional connections are within the Twitter ecosystem where you may follow many people but none of them follow you back. Facebook on the other hand is more likely to have bidirectional connections where you are either friends with someone or you are not. It’s more likely that two friends will follow each other than on Twitter where influential figures will attract a lot of unilateral connections.

Complements

The last property we’ll discuss borrows from the economic idea of complementary goods. Complementary relationships are evidenced by the observation of the increase in demand of one good when the demand increases in another. For example, as the demand for printers increases, there is an observable increase in the demand for ink cartridges. Networks can be thought of in the same way only instead of demand, it’s the value of the network that’s important. A classic example of complements are Microsoft Windows and Microsoft Office.

Strategies for building networks

Some companies will inherently be more likely to generate networks by the nature of their business. Despite this head start, it’s still important to understand strategies to build and foster these networks. Network effects are more rare in B2B businesses than B2C because the more defined product scope and reduced user base.

Entry strategy

Start with a small, niche group and expand outward to include others through time. Facebook is a good example. Mark Zuckerberg rolled their service to Harvard first, then other schools and finally to everyone. This helped them overcome the “cold start problem” where the first few users find no value in the platform.

Growth drivers to adoption

Facebook was successful early at Harvard because they used “growth hacks” to overcome the cold start problem and drive adoption early. They were able to augment the natural growth of the platform through. The startup space is full of stories about how founders bootstrapped early adoption of their platforms in many clever ways.

Critical mass attainment

Understand what KPIs drive growth of the platform and optimise the company operations around those KPIs. A peer to peer lending platform may find that the more transparent default rates are for potential lenders, the more they lend. That company should optimise the experience for transparency in default rates.

Sustained engagement

The best way to ensure sustained engagement is to create a product users love. Roll out new features constantly, engage with the user community and make sure the platform is easy to use. There are many engagement techniques used by large networks to drive critical mass which is the point at which network effects take hold. These generally include offers to invite friends, teasers to show you what others are doing and to connect and showing people close to your network but not yet in it.

Subsidise

Bars are a great example of subsidising one side of the market to entice the engagement of the other side. Offer a ladies night with cheap drinks to attract women to the bar. As the theory goes, the more women, the more men, growing the overall audience. The subsidy is cheap drinks for women.

Does my business have network effects?

There are many ways to measure network effects and it depends on the business. Some of the basic ways to measure the effect is to understand how user engagement (however it’s measured) changes with the quantity of users on the platform. For example, if user growth is increasing while user engagement is increasing, there is evidence of network effects. If the business is operating a two-sided market place, then does the transaction volume increase as the number of buyers and sellers increase? If you can measure an increase in user base as you introduce more features, you may have evidence of network effects. As a16z says, “come for the tools, stay for the network”.

The key takeaway in asking yourself if your product has network effects, is recognising that there is no “one-size-fits-all” approach. It’s critically important to have a sound understanding of the properties of these networks before claiming a business or product has these effects. Investors will closely scrutinise founders who claim that they are able to create and capitalise on network effects.

Some observations

First, because of the power of these effects, a company that is able to demonstrate true network effects is likely to end up taking the lion’s share of any specific market.

Second, there is a tremendous amount of jargon used in the media and founders often mention network effects without a firm understanding of the implications or the ability to qualify or quantify what they are representing.

Finally, network effects are hard fought and fairly rare. It takes experience and near perfect execution to fully exploit these effects.