The science of tuning brands into social media communities

Understanding the complex nature of social media connections defines the ability for social media marketers to achieve consistent results. The principles of online authority are based on Eigenvector Centrality, a core part of the Google search algorithm and Betweenness Centrality, which is part of Facebook’s EdgeRank algorithm.

But how are principles applied to help ISM to do a better job? Establishing authority is a crucial part but only half the social media story. To influence different social users we have to understand how people operate within a network. Exploring the principles of social networks is providing a framework for ISM to move beyond the discovery of authorities to the mapping of and prediction of behaviour.

This simple diagram shows four types of behavioural node in two fictitious networks:

Using Eigenvector Centrality the authorities are node type 3, as they have the largest number of connections that are closely related through mutual connections. Using Betweenness Centrality node type 1 holds the authority as it links with two separate networks, closely followed by node type 2.

To predict behaviour, ISM combine the principles of Eigenvector and Betweenness Centrality and then imagine content being placed into the network. At this point, each type of node exhibits different behaviour and gives the overall network a specific dynamic. We gave them the following names corresponding to the numbers on the diagram.

1) Bridge Node

2) Gateway Node

3) Centrality Node

4) Consumption Node

Bridge Nodes connect two different networks. These users can be spotted by the breadth of subjects they interact with compared to a relatively low number of connections. Bridge users are significant contributors to viral distribution and achieving exposure.

Gateway nodes connect bridges to communities. These nodes are crucial to onward distribution but typically hold strong opinions about what is relevant to their connections. Gateway users tend to be either very active and vocal or complete removed from the conversation but regular sharers.

Centrality nodes, as the name suggests, sit at the centre of communities and will have the highest number of connections within a single network. The authority of a community node can be misleading using Eigenvector centrality. Celebrities are the classic example of a centrality node, whereby they broadcast to many users but rarely interact with third-party conversations.

Consuming nodes are listeners, typically on the outer reaches of their community. They will tend to be passive and likely to include new users either to social media or the individual community.

Describing these relationships when there are only five nodes in each network is one thing. However, ISM is able to scale these principles to much larger networks including 10s or 100s of thousands of users, albeit with a lot more data analysis.

This example explores the subject of motorsport on twitter, combining the networks of Audi, Mercedes-Benz and Jaguar.

By combining a number of networks the different relationships start to appear whether Bridge, Gateway, Centrality or Consumption nodes.

Tools do exist to define behaviour, including Klout and Kred, but these tools look at individuals and not whole networks.

Our approach is different because rather than looking at a star footballer, we look at the whole team and in the context of the league. By looking at individuals in the context of the surroundings, ISM is consistently demonstrating that we are able to produce content that resonates with target audiences and increases organic reach. With the IPO of Facebook likely to spark a greater emphasis of advertising to social media users we are happy to be achieving results without bribery or advertising.

Written by Dan Naylor