A New Way For Marketers to Target the Best Influencers on Social Networks
When Kim Kardashian West posts about using a new product on Instagram, even if only a small percentage of her more than 190 million followers buy it, it’s worth it for the product’s marketers who struck the deal with one of the internet’s top influencers. This kind of influence is happening at all levels across social networks online – from Instagram and Facebook to video gaming platforms – and marketers are scrambling to figure out the best people to target to spread their messages. Research from Maryland Smith has a better way to find them.
Michael Trusov, a Maryland Smith marketing professor, worked with Xi Chen of Rotterdam School of Management in the Netherlands and Ralf van der Lans of Hong Kong University of Science and Technology on the new research, forthcoming in the Management Science. They created a model that can find who to target and when to target them to have the most influence over other users on a social network platform.
“This paper is about estimating the social influence in large online social networks, environments like Facebook, Instagram or some gaming platforms where a lot of people participate and social interaction is a significant component,” says Trusov. “In this type of environment, the action of one person may have impacts on the actions of other people.”
The researchers looked at a large online video game platform to study how the actions of some individuals impacted the actions of others within the platform. In the gaming environment, people get together to play against each other or in teams against other teams, says Trusov. Sometimes they might be friends outside the games, but in other cases, people may just team up with others that they meet within the game environment.
“They’ll likely shape when you want to play,” he says. “You end up choosing your game play time based on when you think others will be on the platform. But in turn, others might be thinking the same things about you.”
Trusov and his co-authors developed a new tool that can efficiently estimate the influence one person may have over others in video game platforms and, more broadly, other large social network environments, based on game theory and community detection.
“We’re saying, according to economic game theory models, we can do better in identifying the influence one person may have on other people within a social network environment,” he says. “The algorithm we developed is very scalable, so it can be applied to massive networks.”
For marketers, Trusov’s algorithm is very valuable for identifying the right influencers to target. He explains how it works in a gaming environment:
“Say you want more people to come and play more in your network. You may incentivize certain individuals within this platform so they will change behaviors of others on the platform – maybe by giving them extra bonus points or a free virtual goods to motivate them to participate. And as they participate, the others who are depending on them to play, will adjust their own behavior accordingly, bringing additional value to the platform. This is the practical application of this type of research.”
But it’s not just about pinpointing the influencers; success also depends on when you want to stimulate that person to exact their influence. Trusov and his co-authors introduce a time dimension when determining optimal influence – an element that hasn’t been explored by research in the past.
“In other words, there are certain periods of time where the influence might have a stronger versus a weaker impact on the network,” says Trusov. For example, the algorithm may reveal a certain influencer has the biggest impact on a Tuesday morning, so if you want to reach their followers, you’ll get them to do something for you then, he says.
With the rise of social networks, influencer marketing has become a very prominent marketing tactic, says Trusov, and it’s getting more and more popular. “What is still challenging is to come up with a good metric that allows companies to assess the effectiveness of their campaigns that involve influencers, but also to measure the impact of specific individuals. What is the expected return if we go after this influencer versus another influencer?”
Right now, says Trusov, marketers use various strategies for finding influencers to work with, based on numbers of followers, reach and engagement, for example.
“Here, we are getting into actually observing the changes in behavior and this is one of the most important markers for measuring influence,” he says. “If I have a ton of network connections but nobody cares about what I do, there is no impact from my behavior on their behavior, so it’s less important to marketers. But if I know that behaviors are actually changing according to the actions of this person, then that would be a much stronger indication that this person is influential.”
“It really boils down to saying who is actually influential and who is not,” says Trusov.
Read the full research, “Efficient Estimation of Network Games of Incomplete Information: Application to Large Online Social Networks,” in Management Science.