A multi-stage model for online viral marketing

Jeremi Karnell, CMO & Founder | One to One Interactive
August 29, 2003

Is viral marketing an art or a science? Is it possible to build a model that guarantees a successful word-of-mouth campaign, or must we resign ourselves to hoping that our message gets through to that elusive group of social influencers who will turn our product into a best seller? These questions are worth giving some serious thought, since a successful viral campaign can exponentially increase your brand's reach at a relatively low incremental cost.

Word-of-mouth communications have received extensive attention from both academics and business practitioners for the last several decades. As early as the 1950s, researchers demonstrated that personal conversations and information exchange among acquaintances could not only influence consumer purchase decisions, but could also shape their expectations, pre-usage attitudes, and even post-usage perceptions of products and services. Zoom forward to the present, where the Internet has increased the ease and velocity of communications within and between social networks. Online marketers now seek to embrace this phenomenon by adopting viral marketing approaches that leverage the power of interpersonal networks to promote their products and services. The concept assumes that electronic peer-to-peer communications are an effective means to transform communication networks into influence networks, capturing recipients' attention, triggering interest, and eventually leading to adoption or sales. Yet it has been difficult to identify substantial evidence supporting these assumptions or explaining how the mechanisms work, which is perhaps why viral marketing is currently viewed as more of an art than a science.

De Bruyn/Lilien Study

In June of 2003, two researchers from the University of Pennsylvania State University SMEAL College of Business (Arnaud De Bruyn and Gary L. Lilien) observed the reactions of 1,200 recipients after they received an unsolicited e-mail referral from one of their acquaintances. Their objective was to identify the influence of word-of-mouth communications at each stage of a recipient's purchase decision-making process:

The De Bruyn/Lilien research leveraged Stanley Milgram's famous 1967 "small world experiment" that scientifically established the concept of "Six Degrees of Separation" (i.e., that through just six or fewer intermediaries, you could be socially linked to any other individual). De Bruyn and Lilien applied this concept to the Internet, and used a pool of 634 individuals to conduct their experiment. Each participant was assigned a target individual, personally unknown to them but living in the United States. Participants were then asked to send an e-mail to one of their acquaintances most likely to know the target person, who would in turn contact another person, and so on, until the target person was reached. Each e-mail gave a brief description of the study and a link to a website urging them to continue the chain of e-mails and complete a survey for the chance to win $1,000.

De Bruyn/Lilien observed whether or not the recipients (1) opened the e-mail, (2) clicked to the link to visit the website, (3) began to answer the survey, and (4) completed the survey. The above steps were used as proxies of the recipient's awareness, interest, evaluation, and final decision stages of their decision process.

Individuals who participated in the survey were asked to answer questions about the acquaintance they had chosen to be their next "link". Twenty questions in the survey were related to the relationship the sender had with his or her next link: the nature of the relationship, demographic similarities, closeness, overlap of social networks and similarities in likes and dislikes. Another twenty questions tapped into various aspects of the sender's self-reported opinion leadership (i.e., how influential his/her word is among acquaintances).

After data cleansing and exclusion of incomplete surveys, De Bruyn/Lilien were able to retain 1,174 survey responses. They analyzed the data across the following relationship related constructs:

• Closeness: The combination of the emotional intensity,intimacy, and reciprocal services that characterizes a relationship.
• Perceptual Homophily: Defined as a person's similarity of values and experience with another person.
• Demographic Homophily: Demographic similarity between the sender and the recipient across 4 dimensions: sex, level of education, age, and occupation.
• Overlap of Social Network: The measure of whether or not the sender and his or her recipient shared the same network of acquaintances.

Four additional items measured the geographic proximity between the sender and the receiver, the frequency of physical encounters and the frequency of e-mail communications, as well as the nature of their relationship.

• Technology Expertise: Internet related
• Personality Strengths: One's social activity and social leadership
• Opinion Voice: Based on ones's likelihood to voice his opinion, possess self-confidence, consider himself an innovator and/or early adopter, seek out information and media exposure.

Results

The De Bruyn/Lilien study revealed several interesting facts about how word-of-mouth communications influence the different stages of an individual's decision-making process. First, they measured that Closeness significantly influenced the decision of the recipient to open the e-mail they received, hence facilitating awareness. Once the recipient opened the e-mail, Perceptual Homophily increased the chance that the recipient would click on the link and visit the website. That is, referrals from sources with similar tastes and preferences are more likely to generate interest. Neither closeness nor perceptual homophily influenced the other stages of decision making, suggesting that although strong personal ties are needed to generate a positive response from a viral marketing campaign, they are not sufficient to influence whether the recipient would actually buy the product or service.

Demographic Homophily was the only dimension to have a significant impact across each stage of the decision making process. The study found that the closer two individuals are to each other in sex, level of education, age, and occupation, the lower the likelihood they have influence over awareness, interest, evaluation and final decision (i.e. female college students are found more likely to be influenced by older, more educated, male sources when considering the purchase of a DVD player). However, the study also found that "like" individuals are more likely to have similar product needs and wants. Once similarities in age, sex, and occupation are combined with similarities in likes, dislikes, and values (or once Demographic Homophily is combined with Perceptual Homophily), there is greater likelihood in these individuals influencing one's interest in a product or service. For example, in their search for a physician, couples with children are found to be more influenced by word-of-mouth referrals from other couples with similar demographic and value characteristics. De Bruyn/Lilien found that the above two effects may systemicatically coexist, but the net effect depends entirely on the product category in question.

Social Overlap increased the likelihood that the e-mail would be opened and read (i.e. increasing awareness). De Bruyn/Lilien viewed this dimension's impact as small since two individuals that have a high social overlap are likely to interact often, and therefore it is expected that a facilitated flow of communications and information exchange would occur regardless.

Opinion Leadership of the sender did not bear any direct influence on the recipient's actions. However, in conjunction with both closeness and demographic homophily, opinion leadership did positively influence the final decision, but not as a main effect.

Conclusion

Taking the above observations into consideration, De Bruyn/Lilien suggests that marketers designing a viral campaign should consider using different levers in tandem to best facilitate the process. Since closeness is the most important factor influencing the chance that an e-mail is opened, viral e-mails should be highly personalized with the sender's name in the "From" line, "Subject" line, and body of the message.

Once the e-mail is opened, similarities in likes and dislikes between the source and the target of the message become the most important drivers to trigger the recipient's interest (or click). Therefore, the content of the e-mail message should refer to the experience of the sender with the product or service in question.

Once on the website, the results of the De Bruyn/Lilien study suggest that the relationship between the sender and the receiver has no influence to trigger further action. In fact, friends driven from a website because they are friends with the source are less likely to take actions. Therefore, other marketing tactics should be used (e.g. free-trials, discounts, testimonials, etc.). These levers will serve as necessary incentives to keep the prospect motivated while the opinion leadership of the sender will begin to have more influence.

In light of the De Bruyn/Lilien study, it is no surprise that some online marketers who have tried to implement "send-to-a-friend" and "viral marketing" campaigns have faced considerable challenges. By understanding the influence word-of mouth communications has on a multi-stage model of decision-making, marketers are better able to deploy the right mix of viral tactics depending on the life stage of their product/service and the goal of their marketing program.

back to White Papers