What to Read Next
Already a member?Sign in
Marketers who want to determine the value of a particular online influencer need to look beyond just the size of a person’s network connections.
That’s the finding from research conducted by Zsolt Katona, assistant professor at the Haas School of Business, UC Berkeley.
Conventional wisdom, reflected by influence-ranking sites such as Klout, is primarily to count the number of a person’s connections to assess his or her ability to influence others.
But Katona’s research has discovered that determining the value of a particular influencer is more complex, and that finding the value of an influencer depends on several underlying factors in the network structure of that individual with the target set of consumers.
The full paper, Competing for Influencers in a Social Network, was presented at the Marketing Science conference in Istanbul. Its key findings:
- Influentials who are most valuable to a company for a particular target consumer or set of consumers are those who provide sole or “exclusive” influence. In other words, the most valuable person influences consumers who are not influenced by any other influentials online. A company that wants to invest time or resources in cultivating an influential person online should focus on those where the target consumers are being influenced in that product/service arena only by that person and not by anyone else.
- Often, though, an influencer does not have an exclusive relationship with a set of target consumers. In these “non exclusive” relationships, influencers who are most valuable are those where the sought after consumer/set of consumers has a very small set of additional competing influencers. Interestingly, the next most valuable influentials are those who influence target consumers who have a very high number of additional competing influencers. The least valuable influentials? Those who influence target consumers who have a middle range of additional competing influencers.
As for what defines small, middle and high numbers, Katona says they depended on the specific parameters set up by his model, but for other real-world applications “small, medium and very high” need to be considered as relative, not objective, until further research determines what exactly the numbers are in a specific setting.