Marketing and Consumers

Social Network Marketing: What Works?

Purchase decisions are influenced differently in social networks than in the brick-and-mortar world, says Harvard Business School professor Sunil Gupta. The key: Marketers should tap into the networking aspect of sites such as Facebook.

When marketers want to reach users of social networks such as Facebook, MySpace, or Cyworld, they have two choices: buy advertising or start a viral campaign.

New research by Harvard Business School professor Sunil Gupta suggests that viral may be the way to go in these connected worlds. But first it's important to understand both who influences purchase decisions in online communities and which groups of users can be influenced.

Viral campaigns truly leverage the network aspect of these social networking sites.

"By understanding the social network of users, firms can better understand and influence consumers' behavior," says Gupta, who coauthored the working paper "Do Friends Influence Purchases in a Social Network?" [PDF] with Raghuram Iyengar, of the Wharton School, University of Pennsylvania, and Sangman Han, of Sung Kyun Kwan University in South Korea.

Gupta recently discussed the research with HBS Working Knowledge.

Sarah Jane Gilbert: Your research attempted to answer this question: Do friends within a social network influence online buying behaviors of others in the group? What did you discover?

Sunil Gupta: To answer this question we used data from Cyworld, an online social networking site in South Korea with almost 21 million members. Cyworld users buy virtual items to decorate their home pages. Our research shows that some users are influenced by the purchases of their friends while others are not.

Q: One of your findings was that certain types of group members are more influenced by social pressure than others. Can you discuss the distinctions among user groups and how they influence buying?

A: We found three distinct groups. Low-status members (48 percent of the users in our sample), who are not well connected to others, are generally unaffected by the purchases of other members. Middle-status members (40 percent of the users) are moderately connected and show a strong and positive effect due to friends' purchases. In other words, this group shows strong "keeping up with the Joneses" behavior.

Members of the high-status group (12 percent of the users) are well connected and are very active on the site. However, these users show a negative effect due to friends' purchases because they want to remain distinct. Instead of buying items like the other members, this group tends to pursue non-purchase-related activities (e.g., uploading their own content).

Q: Were you able to quantify social influence in terms of how it increased or decreased the percentage in sales revenue?

A: The impact of the low-status group on revenue is negligible. Social influence increases revenue from the middle-status group by 5 percent. In contrast, social influence leads to almost a 14 percent drop in revenue from the high-status group.

Q: Despite the ability to gather millions of users, the business viability of social networking sites seems uncertain. Is there any good news in your findings for the Facebooks and Twitters of the world?

A: Following the success of Google, social networking sites such as Facebook have been trying an advertising-based business model. However, studies show that the click-through rate of ads on social networking sites is extremely low—simply because people don't go to these sites to seek information about specific products. Therefore, the advertising-based business model has had only limited success on social networking sites.

If the purpose of advertising is to influence consumers' purchases, our research shows that there is another way to influence their behavior. Imagine that Sony wants to promote its new digital camera. Sony can either advertise on Facebook and accept a very low click-through rate, or give away free cameras to several Facebook members (potentially at a lower cost than advertising) and generate a viral campaign. Our research shows that this viral campaign is possible. We further show what type of users are more likely to be influenced by such a campaign.

Interestingly, Sheryl Sandberg [HBS MBA '95], the new COO of Facebook, recently talked about such a campaign. On Valentine's Day, Honda offered 750,000 Facebook members a heart-shaped virtual gift complete with the Honda logo that could then be passed on to other members. It is exactly this type of viral campaign that has the potential to be an enormous source of revenue. Unlike banner ads, these viral campaigns truly leverage the network aspect of these social networking sites.

Q: What are some key takeaways from your work for retail and other online sites in general hoping to influence online purchasing decisions?

A: The fact that people are influenced by their friends is not new. We all know it at some level or another. However, by understanding the social network of users, firms can better understand and influence consumers' behavior.

Consider a cosmetic company such as Mary Kay. Its website currently allows consumers to upload their picture and try different cosmetics virtually to see how they may look before making a purchase. However, if users (especially young consumers) really value the opinion of their friends, Mary Kay may benefit from linking its online site to a social networking site such as Facebook where users may even get instant feedback from their friends before making a purchase.

Q: What are you working on now?

A: I am continuing my research to better understand how information and influence spreads in social networks. I am also working on understanding the relative role of offline and online advertising to determine how firms should optimally allocate their resources across different media.

Latest from HBS faculty experts

Expertly curated insights, precisely tailored to address the challenges you are tackling today.

Strategy and Innovation

Social Responsibility

Diversity and Inclusion