SnapChat lost $1.3 billion earlier this year in market value after Kylie Jenner’s tweet criticized the app’s new layout for being too confusing to navigate. Jenner’s tweet certainly proves that influencers play an important role when it comes to swaying consumer buying habits. Influencer marketing is increasingly becoming ubiquitous – among both established brands and newcomers. For marketers designing influencer marketing frameworks its crucial to understand how influence works.

The influencer marketing spends globally is expected to reach 10 billion dollars by 2020. In India, the searches for the term influencer marketing has increased nearly four-fold in the last two years.

Nearly 89% of marketers in India consider influencer marketing as an effective medium as per a recent survey conducted by Zefmo. Marketers consider Influencer marketing for product launches, content promotion and event promotion.

How Influence Works?

Information cascades are phenomena in which an action or idea becomes widely adopted due to influence by others. Cascades are also known as “fads” or “resonance.”

Diffusion is a process by which information, ideas and new behaviour spread over the network. For example, the adoption of a new product begins on a small scale with a few “early adopters”. As more people adopt the product, their friends and neighbours start using it. As the information spread across the network, it creates a cascade.

Sociologists have studied cascades for many years to understand how diseases spread. More recently, researchers have investigated cascades to understand how influence works in viral marketing.

Diffusion models can be divided into two groups:

Threshold Model

In the threshold model, each user in the network has a threshold. A user adopts an idea or a product if other users in the group have already adopted the idea or the product. Every user has a threshold that dictates the activities they perform.

The activities that a user performs has a strong correlation to peer pressure. Users with higher threshold will not tend to do anything that no one else is doing. Users with a lower threshold will tend to do whatever they think is right or tend to follow one or two strays who are doing something that no one else is doing.

Independent Cascade Model

In the independent cascade model, every time a user purchases a product or adopts an idea the other users in the network will also do the same in all probability.

While the above models explain how influence works they are based on assumed rather than measured influence effects. Recent studies on actual diffusion also measure the importance of factors like the presence of highly connected individuals, or the effect of receiving recommendations from multiple contacts.

Examples of two product recommendation networks. Left: First aid study guide. Notice many small disconnected cascades. Right: Japanese graphic novel (manga). Notice a large, tight community.

The above image shows two typical product recommendation networks. Most product recommendation networks are made up of a large number of small disconnected components where we do not observe cascades. Then there is usually a small number of relatively small components where we find recommendations propagating.

Also, notice bursts of recommendations and collisions. Some individuals send recommendations to many friends which results in star-like patterns in the graph.

Designing Influencer Marketing Model

Now that you know how influence works how do you use this information to develop an influencer marketing model? We don’t have a framework or a step-by-step guide for developing an influencer marketing model but certain principles that you can deploy to make sure that you derive the intended results.

Choosing a Diffusion Model

Let’s say you are a brand trying to influence a network to change buying behaviour or adopt a new product. To change behaviour or improve adoption of the product you will have target the users with a lower threshold. Users with a lower threshold are your early adopters.

If you are a brand that serves an existing market, then you can leverage the independent cascade model to study how your information or idea gets consumed by the network.

A recent study by MIT Media Lab found that fake news has very distinct diffusion characteristics which allow it to spread faster than real news. Surprisingly the study has found that bots don’t have a critical role in the diffusion process.

Choosing an Influencer

Influence is ultimately a matter of context and trust. Besides deep expertise and a broad network, for a user to be perceived as an influencer he/she should have the following characteristics:

  • Has deep empathy for the users in a certain domain and spends a considerable amount of time to listen, learn, understand and solve problems in that domain.
  • Focuses on a single domain or multiple domains related to the same topic.
  • Constantly updates knowledge and is an early adopter of new trends.

Most influencer marketing strategies fail because they choose influencers based on follower counts. The actual value lies in the relationships that the influencers have with their followers which requires the investment of time, intellect and even emotion.

Topology

Topology helps you to understand how fast can information travel across a network. You can measure someone’s influence by understanding the nature and the strength of their bi-directional links with other users in the network. A network with a centralized distribution can allow for rapid diffusion of information.

Complex Contagion

Discussion models are binary in nature. Either a node (user) is affected or not. Within the model, all it matters is whether one of the nodes (users) affects it or not.

Complex contagion is a model a user requires multiple exposures before an individual adopts a product or an idea. For example, adoption of new technology is expensive especially for early adopters but less so for those who wait. We can then label the network as a complex contagion to understand how many users should adopt the product for a given node (user) will do likewise.

There might also be two competing events propagating across the network. For example, how will an individual vote for two different candidates based on the social networks they are part of. We could define how many of the nodes (users) neighbours need to vote for a particular before the node also decides to vote for the party.

This means that the node will only act after a certain defined threshold value is reached. Its more complex form of contagion as their will be feedback when the node changes its state it will affect the value of nodes around it.

Diffusion Models and the Way Forward

Recent research from the University of British Columbia found that the propagation of information to a user does not always lead to the adoption of a product or an idea. There may be many factors that affect a user’s adoption decision, including the user’s interests, budget and time. Research also found that opinions can propagate from a non-adopter who is active and can promote adoption from others.

There is still a long way to go before a truly realistic model is developed to understand the adoption of a product or idea. It is crucial for marketers to understand how influence works using network theory because it allows you to design an effective influencer marketing model and tells you why something worked.

LEAVE A REPLY

Please enter your comment!
Please enter your name here