When consumers speak positively about a product or brand on social media it can have a very large influence on other consumers’ purchase decisions. In fact, recent research has found that user-generated content has a greater overall influence on purchase activity than marketer-generated content, i.e., marketers’ attempts to advocate on social media for themselves, or coverage in the popular press. Cognisant of this, many marketers now attempt to benefit from consumers’ influence over peers, engaging in organised, directed efforts to ‘engineer’ social media content and marketing campaigns that purposefully foster consumers’ creation of user-generated content.

When engineering content with an eye toward ‘virality’, two main factors need to be considered. First, the features of the content. A variety of content characteristics have been found to influence the likelihood that it will be propagated on social media, e.g., the emotions it invokes. Second, the social network structure of the consumer audience that a marketer hopes to permeate with its message. Prior research documents that an individual is more likely to be influenced into adopting a product by peers with whom they hold strong social ties (friends), and with whom they share more mutual third-party connections (onlookers).

Maintaining a shared audience, known in the academic literature as structural embeddedness (see Figure 1), amplifies an individual’s influence over a peer because the presence of a shared audience simultaneously increases the potential upside from the peer behaving ‘well’, e.g., acceptance by the social group, and the downside of behaving ‘badly’, e.g., ostracism. In a recent study, we sought to explore the combined influence of content characteristics and network structure, examining the possible value of gaining synergies between the two.

Figure 1. Low vs. high structural embeddedness

Our study was motivated by the observation that the very same mechanisms that cause a shared audience to amplify peer influence also have long been acknowledged as playing a particularly key role in pro-social behaviour, e.g., whether and when individuals make donations to a charity. For example, ample research has found that donations to charity are heavily influenced by the presence of onlookers. Based on this observation, we hypothesised a synergistic, super-additive effect, between structural embeddedness and the influence of social media content advocating contributions toward pro-social, public goods. That is, we hypothesised that structural embeddedness would amplify peer influence for all kinds of advocacy content, but that this influence would be amplified to a greater degree when the advocated behaviour involved a prosocial element.

A recent example that illustrates the point is the “ALS Ice Bucket Challenge,”a viral phenomenon that spread across social media in the summer of 2014, wherein individuals were influenced by their friends to contribute to a social cause, i.e., advancing awareness of ALS by engaging in the “ice bucket challenge,” donating money to researching a cure for the illness or both. A key reason the challenge spread so quickly and so broadly is that the solicitation to contribute to this public good was frequently made in front of an audience of the target’s socially proximate peers. As a result, the potential benefit (damage) to the target’s reputation and image from (not) responding was significant. Thus, the virality of the ice bucket challenge appears to have resulted from a uniquely synergistic combination of context (the message was propagated through embedded networks on social media) and content (targeted individuals were asked to contribute toward a pro-social cause, where the socially desirable response was to conform).

If our hypothesis were true, this would be potentially quite important, given that many brands and products today engage in cause-marketing and corporate social responsibility efforts. Our hypothesis, if confirmed, would suggest that marketers might best take advantage of such initiatives by considering consumer social media network structure or, conversely, by tailoring content messages in light of observed network structures.

We tested our hypothesis using data on over 1,000 crowdfunding campaigns from Kickstarter, the world’s largest reward-based crowdfunding platform, where individuals can raise money to support new ventures. We paired daily fundraising data from Kickstarter with daily mentions of each campaign on Twitter, examining the relationship between the two. Based on prior work, we expected that when campaigns were advocated on Twitter, the impact on fundraising would be stronger when the social network structure among the Twitter users involved exhibited greater structural embeddedness, i.e., when the Twitter users had more mutual third-party connections. Further, our hypothesis about the synergies between embeddedness and pro-sociality caused us to expect that structural embeddedness would matter most when a crowdfunding campaign was pursuing a venture involving some prosocial, public good benefit (a prime example of this would be the Kickstarter campaign for Syrian Aid Relief).

We confirmed both of our hypotheses in the data. Our findings suggested that the returns to structural embeddedness among Twitter users for pro-social campaigns were roughly double the returns to campaigns that lacked any pro-social element. In dollar terms, the average benefit of structural embeddedness for pro-social campaigns relative to campaigns with no pro-social element translated to approximately $200 in additional fundraising, per day, or nearly $6,000 over the life of a typical 30-day campaign.



Yili Hong KevinYili Hong is an associate professor (with tenure) of information systems at the W. P. Carey School of Business, Arizona State University. His research focuses on online markets and consumer uncertainty. His works appeared or are forthcoming in top academic journals. He is the winner of the 2014 ACM SIGMIS Best Dissertation Award, runner up of the INFORMS ISS Nunamaker-Chen Dissertation Award and 2012 ICIS Best Paper Award. He received his Ph.D. in information systems from the Fox School of Business at Temple University.

Yuheng Hu is is an assistant professor of information and decision sciences at the College of Business Administration, University of Illinois at Chicago. His research interests are in the area of social computing and machine learning. His current and past projects include characterising the effect of social media sharing for crowdfunding campaigns, and understanding and predicting people’s behaviours, sentiments and engagement with public events and life events. Prior to
joining academia, he was a researcher at IBM Research.

Gordon Burtch is an associate professor (with tenure) of information and decision sciences and a McKnight Presidential Fellow at the University of Minnesota’s Carlson School of Management. His research, which focuses on the economic evaluation of information systems, employs empirical analyses rooted in econometrics and field experimentation to identify and quantify the drivers of individual participation in online social contexts. Prior to entering academia, he was employed as an information systems auditor, a hardware design engineer, and most recently as a technology consultant with Accenture Canada in Toronto.