Yale School of Management

Center for Customer Insights

Advancing the frontiers of consumer understanding

The Value of Friends

Word of mouth marketing has received much attention in recent years, with particular focus on how to choose the appropriate “seeds”—those who spread information. New research demonstrates the surprisingly simple and elegant way in which friends can serve in this role

The value and power of a message rests with the messenger. Marketers are intimately familiar with this idea. They launch word of mouth campaigns around it. They construct entire economies (e.g. the social media influencer) around it.

And yet, which method for spreading information and increasing the adoption of some new product or technology is most efficient remains uncertain.

A new working paper by Vineet Kumar and K. Sudhir of the Yale Center for Customer Insights helps lend clarity to the issue. Entitled, “Can Friends Seed More Buzz and Adoption,” the study finds that without detailed information about the structure of a network—information that is often expensive or difficult to obtain—the traditional approach of “seeding” leaders and influencers with information is not the most effective way to spread word of mouth. Instead, just look to people’s friends.

The intuition for this finding lies in what’s known as the “friendship paradox,” which the paper describes in straightforward terms: “On average, your friends have more friends than you do.” Kumar and Sudhir write, “The strategies [used here] leverage the Friendship Paradox to identify more connected individuals for seeding irrespective of the underlying network structure.”

The researchers borrowed data from a 2013 study that looked at the effect of word of mouth on adoption of micro finance in 43 villages across Southern India. This prior research sought to understand the sway of local leaders and included comprehensive mapping of the social networks within villages. Kumar and Sudhir used this foundation to build a theoretical model that investigated other possibilities: without using knowledge of the network structure, what happens when you seed information randomly instead of targeting leaders? What happens when you seed people’s friends? What about the friends of leaders?

The researchers found that seeding somebody’s friend with information on micro finance was about 21% more effective than seeding at random—and, critically, about 8% more effective than seeding a leader. The greatest gains in word-of-mouth adoption—23% over random seeding—resulted from seeding the friends of recognized leaders. The results were especially pronounced when there were few seeds, something that occurs when the process of seeding an individual is expensive (e.g. adoption of a new technology that requires extensive training).

Kumar and Sudhir acknowledge that the structure of a network can interact in important ways with seeding strategies. Very dense networks, for example, boost the effectiveness of seeding leaders. But the benefits of trying to customize seeding strategies to network structure must be balanced against the cost of learning that network structure, the researchers write, especially since “data on the right or relevant network for a particular purpose is often unavailable.” Given this, looking to friends makes simple, inexpensive sense.