Skip to main content

Network Diffusion with Social Media Influencers in Marketing


As social media becomes a growing presence in the lives of consumers everywhere, marketing companies are eager to optimize their efforts in convincing people to use their products. Recently, marketing agencies have dedicated a lot of their efforts toward identifying influencers in various communities – individual accounts that influence a large number of accounts in some group of people. Commonly thought of examples of influencer accounts can be seen as “FitTea Detox Tea ambassadors” on instagram (like celebrity endorsements) or campus representatives for specific clothing brands. These accounts are not limited to product endorsements. A good influencer is someone who is perceived as a good connector of people. Not only do they have expertise in some area and a lot of public acclaim, they are perceived as someone who wants to empower their followers and grow their network for public good (rather than personal gain through compensation or growing fame). The recent use of social media influencers follows a 1/9/90 rule in marketing – markers want to cater to the 1% of the public (the influencers) who will spread their message to 9% of the public (the amplifiers who spread their message) to the remaining 90% of the public. This is designed to get the most bang for the buck.

It is challenging to identify influencers, particularly if you are not particularly familiar with the group they have influence in. Influencers cannot simply be identified by number of followers because accounts can be made quickly and there are huge networks of spam accounts that follow each other. For a marketing firm to invest in a spam influencer would be a huge waste because they would not gain many actual users in the desired 9/90% of the population because those are also spam accounts instead of real users. With evaluation of how different clusters relate an connect with authentic clusters (not spam ones), markers have employed data science to create tools to identify real influencers. It is easy for a spam network to connect to large authentic networks by following a few influencers. By connecting to real clusters, these spam accounts may seem authentic, so researchers wanted to figure out how to spot fishy connections. Regarding authentic clusters, various communities are typically linked by 4 key people. If a spam cluster is connected by more than that, then it is likely spam. It is also important not only to identify the spam networks, but real users who interact with them. These users may have an artificially inflated number of followers and may not be as good of an influencer as perceived. If they are connected to a spam cluster that has infiltrated real networks, they are not a sound investment to employ as an influencer supporting your company because they do not have real followers.

This article gave me a stronger understanding of diffusion in networks and what it means in a sense of marketing. Marketing firms are very interested in identifying clusters within their target audience so they can use their resources most effectively and invest in influencers who will convert as many users as possible quickly. It is interesting to think that they must not only consider the clusters on a surface layer but the accounts that comprise them – realizing that those connected to spam accounts are an unsound investment and must make them reconsider their perceptions of connectivity within their user base. This makes their calculations for cluster density more challenging to consider because they need to look at the number of authentic connections within a cluster and and also consider the threshold for conversion from all accounts versus personal accounts. By investing in influencers in a spam account, a company could not only lack sales but harm their image by being associated with spam. It is really important for them to understand the networks of their users to market effectively and influence with success.




Leave a Reply

Blogging Calendar

November 2018
« Oct   Dec »