Make em laugh, make em cry:

www.marketingscience.info Make em laugh, make em cry: Understanding Which Emotions Drive Video Sharing on Facebook Karen Nelson-Field, Erica Riebe and Kellie Newstead Report 58 for Corporate Sponsors October 2011 Contents Abstract... 2 Report... 3-7 References and Further Reading... 8-10

Make em laugh, make em cry Abstract Marketers are increasingly using viral video; if a video can go viral and be widely shared via social media it provides advertisers with great low-cost reach. To date however, an understanding of what makes video content more or less shareable has been more art than science. We examine 400 pieces of usergenerated video shared via Facebook in terms of emotional content (arousal and valence) and see how this relates to content sharing. Our research shows it is high arousal emotional responses that most stimulate sharing, in particular hilarity and anger. So it is content that makes us laugh or cry, not just smile or frown, that triggers sharing. Positive emotions on average do better than negative emotions at stimulating sharing. However while anger is an exception (in that it is a negative emotion associated with high rates of sharing) it appears very difficult to create content that elicits this emotion. Other negative emotional responses such as shock and disgust appear far less likely to stimulate sharing. By far the most commonly elicited emotional responses were amusement and boredom - but these videos really don t get shared. So the lesson for advertisers would seem to be to aim for content that causes a marked physiological response (i.e. makes people laugh or perhaps cry) at all costs don t simply amuse (or worse, bore) them. The Ehrenberg- Bass Institute provides a fundamental program of University research into buyer behaviour & brand performance. Corporate membership delivers new discoveries & knowledge to marketing departments around the world. EHRENBERG CENTRE Ehrenberg-Bass Institute Report 58 for Corporate Sponsors FOR RESEARCH IN MARKETING PAGE 3 LONDON SOUTH BANK UNIVERSITY

Make em laugh, make em cry: Understanding Which Emotions Drive Video Sharing on Facebook Karen Nelson-Field, Erica Riebe and Kellie Newstead The Ehrenberg- Bass Institute provides a fundamental program of University research into buyer behaviour & brand performance. Corporate membership delivers new discoveries & knowledge to marketing departments around the world. INTRODUCTION In less than three years social media has become the most common Internet based activity and has fundamentally changed the way in which we interact with and consume media (Qualman 2009). It has dawned the era of the connection generation, a generation that is said to crave interaction with, and connection to, vast networks like never before (Pintado 2009). At the core of such connection is sharing by personal referral; sharing of information, photos, opinions, entertainment and news. For marketers, such sharing provides an unparalleled opportunity to disseminate information to vast communities, to now spread a message to literally hundreds of thousands of potential brand buyers in a matter of days, without the traditional mass media costs. SHARING FOR EXTENDING REACH The power of personal referral has been long regarded as a critical influencer of consumer behaviour (Buttle 1998; East, Hammond et al. 2008; Romaniuk and Nguyen 2011). Known today as word-ofmouth (WOM), such interpersonal communication is based on the notion of trust i.e. that a recommendation from an [honest] peer makes for a more credible source than a message with commercial intent (Arndt 1967; Arora 2007; Romaniuk 2007). With the advent of the Internet, we now see the emergence of a new form of online WOM - ewom or viral marketing communications. Likened to the spread of an epidemic, viral marketing (particularly in an unpaid form) still embodies that trust element attached to peer-to-peer communication, however in an online context (Dobele, Lindgreen et al. 2007). The rise of video sharing giants such as YouTube and Google Video, combined with increased broadband connectivity and improved sharing functionality across most social networking sites, has cemented the role of viral video in the marketing mix of many corporates (Cashmore 2009; Tsai 2009; Eckler and Bolls 2011; O Malley 2011). Its stronghold in marketing strategy is evident from the transfer of Ehrenberg-Bass Institute Report 58 for Corporate Sponsors

advertising budgets from TV, search and direct response into viral video campaigns (O Malley 2011). When executed correctly, a viral video campaign not only offers the marketer the benefit of extended reach (quickly and cost efficiently), but also an ability to extend the life of an offline campaign, curb advertising avoidance, bypass apathetic media consumers, strengthen a brand s distinctive assets, create publicity and enhance memory cues for the brand (Dobele, Tolemen et al. 2005; MindComet 2006; Dobele, Lindgreen et al. 2007; Nelson-Field and Klose 2010; Southgate, Westoby et al. 2010; Eckler and Bolls 2011). WHAT MAKES VIDEO GO VIRAL? For practitioners the potential benefits of viral video campaigns may seem abundant; however success can be hit and miss. While some videos may be shared tens of thousands of times in a few short hours, others fall very short of expectations. So why do certain pieces of content get shared more than others? Given the increasing popularity of viral video in marketing strategy, there is little research in terms of understanding the elements of successful viral transmission (particularly in the social media context). While common industry wisdom suggests videos that are funny, inspiring, original and surprising are more shareable (O Neill 2010; Daly 2011; Weiner 2011), there is little evidence to back up these beliefs. The only studies which have looked at this area involve an infiite array of possible confounding variables, for example offline campaign expenditure (Elberse, Lee et al. 2011), prior exposure to advertising (Peracchio and Meyers-Levy 1997), prior product usage (Maclnnis and Jaworski 1989) and funded seeding tactics (Reichgut 2011). These confounds make it difficult for researchers to draw any meaningful or reliable conclusions about the characteristics of videos that really do make them go viral. EMOTION AND AROUSAL In this research we look at how sharing is affected by the extent to which video content elicits particular emotional responses from its audience. We draw on a range of studies, including research from the psychology literature that consider the concept of social sharing 1 and the role of emotions on content sharing. Most of these studies are either conducted outside of the context of social media, or have other methodological limitations that justify the further examination of this issue 2. We model this research on that of Berger and Milkman (2010) who considered how emotional response was linked to the email sharing of 2,566 articles from the New York Times most shared list. Berger and Milkman took a physiological approach suggesting that arousal, an established construct of emotion, (Baumeister and Bushman 2010) might be the key to driving viral diffusion (also see Berger 2011 (Berger 2011)). For example, while hilarity and amusement are both positive emotions, the degree of sensory alertness associated with the different emotions will likely differ (i.e. hilarity will likely cause laughter, while amusement may not) ((see Cacioppo et al (2000)). Drawing on previous literature, this study addressed the following research questions: 1. Which emotions are most elicited in viral video content, and are videos with these emotions present the most shared? 2. Are videos that elicit high arousal emotions (either of positive or negative valence) shared more than low arousal emotions? 1.Social sharing is a generalisable concept that suggests that most emotional experiences are shared (typically in the course of a conversation) shortly after they occur. Furthermore, that the extent of sharing (in relation to the frequency and number of people with whom the incident was discussed), is directly related to the strength of the emotion felt. 2. On provocative creative appeals see Porter and Golan (2006) and Brown, Bhadury, and Pope (2010). On components of good branding and distinctiveness see Southgate, Westoby and Page (2010). On emotional appeals see Dobele, Lindgreen, Beverland, Vanhamme and van Wijk (2007). PAGE 6 Ehrenberg-Bass Institute Report 58 for Corporate Members

3. Are videos that elicit high arousal positive emotions shared more than high arousal negative emotions? THE DATA AND CONSTRUCT In this research we sought to examine the relationship between video sharing and the ability of the content to elicit particular emotional responses. To do this, first we gathered a total of 400 videos from an aggregator site (Facediggs). The site provided us with the cumulative numbers of shares that occured in real time for each piece of content being shared via the share function in Facebook. This gave us an aggregate behavioral and continuous measure of the virality of each piece of content. We used only user-generated (non-commercial) video, not commercial and/or branded content, thereby controlling for numerous potentially confounding variables (as mentioned) that may have affected findings in previous studies (the ratio of user-generated to commercial content at the time of data collection was approximately 1:1). To account for differences in the availability of the content (i.e. video that is available online for 30 days has a greater opportunity to be shared than a video that has just become available), we compared the average number of shares per day for each video. Our use of actual sharing (as opposed to simple views or forwarding probability) constitutes a considerable contribution to the literature as sharing is a behavioral metric which captures an element of buy-in (i.e. someone liked it enough that they actually did share it), and eliminates any error that might result from using a prediction from just a small sample of coders. The second step in this research was to subject this video data to the coding of emotions. 14 independent coders were asked to watch a random sub-set of videos and indicate the emotions they personally felt from a list of 16 potential emotional responses. Drawn from the literature, the list included positive and negative emotional pairs (e.g. happy and sad) for 4 emotional descriptors (e.g. temperament). High arousal extremes for these low arousal positive and negative emotions were then also included (e.g. exhilaration is the high arousal version of happiness and anger is the high arousal version of sadness). Table 1 shows the 16 emotions included on the list and their allocation to the positive/ high and negative/low extremes of the valence and arousal constructs being included in our study. To lessen the impact of subjectivity, all videos were double coded by two separate individuals. The level of inter-coder agreement was 89%. Our analysis involves a simple, yet descriptive, method that enables us to easily observe overall patterns across the range of variables in the data being used. We begin by describing which emotions the 400 highly shared videos are most likely to elicit. We then consider the average number of shares per day that occurs for videos that elicit these emotional responses. RESULTS 1. While most videos are either amusing or boring, they are not the most shared. Amusement and boredom are by far the most commonly elicited emotions when audiences are exposed to user-generated video content. See Table 2 and 3. Surprisingly, shock, anger and exhileration are very rare emotional reactions, suggesting Table 1: Emotion construct; valence and arousal Emotional Descriptor Positive Negative High Arousal Low Arousal High Arousal Low Arousal Awe Astonishment Surprise Shock Discomfort Temperament Exhilaration Happiness Anger Sadness Motivation Inspiration Calmness Frustration Boredom Humour Hilarity Amusement Disgust Irritation Ehrenberg-Bass Institute Report 58 for Corporate Members PAGE 7

Table 2: Propensity for emotions to be elicited (%) Emotional Descriptor Positive Negative High Arousal % Low Arousal % High Arousal % Low Arousal % Average % Humour Hilarity 9 Amusement 40 Disgust 3 Irritation 4 14 Motivation Inspiration 9 Calmness 7 Frustration 1 Boredom 31 12 Temperament Exhilaration 3 Happiness 23 Anger 2 Sadness 4 8 Awe Astonishment 9 Surprise 9 Shock 2 Discomfort 7 7 Average % 8 20 2 12 that it is very difficult to create content that elicits these emotions. 2 Videos that evoke high arousal emotions are most likely to be shared. The content that is most likely to be shared is that which elicits high arousal emotions. These are emotions that are likely to cause a marked physiological response (i.e. sensory alertness, attention or energy). The two high arousal emotions of hilarity (videos eliciting this emotion averaged 6,392 shares per day) and anger (averaging 5,293 shares) are key contributors to this finding (as shown in Table 3). So in essence while emotions such as amusement, boredom, happiness and even surprise are the most prevalent in video content today, overall they fall short in terms of their capacity to be shared based on their low arousal nature. 3 Videos that evoke high arousal, positive emotions are shared more than videos that evoke high arousal, negative emotions. While overall we find that in most cases videos that elicit high arousal positive emotions are shared more than those that evoke other emotional responses, there are a couple of exceptions. Shock and anger have a slightly stronger relationship with sharing than their high arousal positive pair. For example videos that evoke shock are shared more (2865) than videos that evoke its high arousal positive pair (astonishment, 2184) and videos that evoke anger are shared slightly more (5293) than videos that evoke its high arousal positive pair (exhilaration, 4187). 4 Overall videos that evoke positive emotions are shared more than videos that evoke negative emotions. Table 3: Average shares per day for videos eliciting particular emotions (#) 3 Emotional Descriptor Temperament (n = 257) Humour (n = 452) Awe (n = 220) Motivation (n = 390) Average shares per day High Arousal n = 247 Positive Low Arousal n = 637 High Arousal n = 65 Negative Low Arousal n = 370 Average shares per day per video Exhilaration 4187 Happiness 3142 Anger 5293 Sadness 4642 4316 Hilarity 6392 Amusement 2038 Disgust 2862 Irritation 2800 3523 Astonishment 2184 Surprise 3168 Shock 2865 Discomfort 1625 2460 Inspiration 3557 Calmness 1976 Frustration 1588 Boredom 622 1936 4080 2581 3152 2422 3. The n s in this table show the number of videos that elicited an emotional response of this type (from one or other of the 2 coders), while the numbers in the body of the table show the average number of times each video that elicited that particular emotion was shared in a day. PAGE 8 Ehrenberg-Bass Institute Report 58 for Corporate Members

In addition to arousal, we found that valence alone also has an impact on whether video content will be shared. This is consistent with previous research (See Eckler and Bolls (2011)), that provided the most established evidence to date of emotional tone having an impact over the potential for that content to be shared. In comparison to the role that arousal plays in determining whether a video will be shared, valence plays a lesser role. However, it is still important for marketers to create content that is positive, rather than negative if they wish for that content to be shared with a wider network. For instance, while videos that evoke shock may be shared less than its high arousal positive pair (astonishment), videos that evoke nearly any of the other positive emotions (irrespective of arousal) are still shared more. So while shock is also touted by industry as a key driver of viral video, this research shows otherwise. CONCLUSION This research finds that the things that make us laugh (not just smile), make us feel exhilarated (not just happy), make us truly inspired (not just calm) and make us angry (not just unhappy) are more likely to trigger forwarding behaviour. Therefore to improve the sharing opportunity, when designing creative content, advertisers need to upscale the degree of emotional arousal. However, some emotional responses look difficult to achieve based on their rarity in the user-generated content. For an advertiser this suggests that such creative ideas are going to cost more, in that they will take agencies longer and greater effort to come up with. It s also very possible that such content may overwhelm branding, or that it may be less related to the brand than it would be if gaining such an emotional response from viewers were not a consideration. And similarly, aiming for such an emotional response may mean that it is simply more difficult to also include branding in their content. So while it is much better for a video to achieve hilarity (high arousal) than happiness or mere amusement (low arousal), the latter appears much easier to achieve. An advertiser might be on safe ground to aim for hilarity but be willing to accept happiness or amusement, particularly if it results in a better branded video. This raises concerns about the use of content designed to engender anger. If an ad can make a viewer angry our data suggest they are likely to share it, but for many brands this will be extraordinarily difficult to do in a brand relevant way, and so runs the dual risk of the content being merely irritating and it being poorly branded. We suspect that few advertisers probably want to go into the angry space, and there maybe legal barriers that prevent it, so it is perhaps best to aim for videos that elicit positive emotions (hilarious, exhilarating or inspiring). Certainly it is very important to avoid making boring videos, unsurprisingly few people want to share these with their friends. This study looked at videos that were shared. The fact is, this would be a small proportion of videos produced. Most videos would not go viral. Ehrenberg-Bass Institute Report 58 for Corporate Members PAGE 9

References and Further Readings Arndt, J. (1967). Word of Mouth Advertising and Informal Communication. Risk Taking and Information Handling in Consumer Behavior. Boston, Harvard University: 188-239. Arora, H. (2007). Word of Mouth in the World of Marketing. The IcFai Journal of Marketing Management VI(4): 51-65. Baumeister, R. and B. Bushman (2010). Social Psychology and Human Nature. Belmont CA, Wadsworth, Cengage Learning. Berger, J. (2011). Arousal Increases Social Transmission of Information. Psychological Science Online 22(7): 891-893. Brown, M., R. Bhadury, et al. (2010). The Impact of Comedic Violence on Viral Advertising Effectiveness. Journal of Advertising 39(1): 49-65. Buttle, F. (1998). Word of mouth: understanding and managing referral marketing. Journal of Strategic Marketing 6: 241-254. Cacioppo, J. T., G. G. Berntson, et al. (2000). The psychophysiology of emotion. Hanbook of Emotions. M. L. a. J. M. Haviland-Jones. New York, The Guilford Press. Cashmore, P. (2009). YouTube: Why do we watch? CNN Tech. Retrieved January, 2011, from URL: http://articles.cnn.com/2009-12-17/tech/cashmore. youtube_1_youtube-viral-video-most-viewed?_s=pm:tech. Daly, J. (2011). The 10 Laws Of Viral Video Success. Defuse. Retrieved January, 2011, from URL:http://www.defuse.com/blog/the-10-laws-of-viral-videosuccess/. Dobele, A., A. Lindgreen, et al. (2007). Why pass on viral messages? Because they connect emotionally. Business Horizons 50(4): 291-304. Dobele, A., D. Tolemen, et al. (2005). Controlled infection! Spreading the brand message through viral marketing. Business Horizons 48: 143-149. Ehrenberg-Bass Institute Report 58 for Corporate Members PAGE 11

East, R., K. Hammond, et al. (2008). Measuring the impact of positive and negative word of mouth on brand purchase probability. International Journal of Research in Marketing 25(3): 215-224. Eckler, P. and P. Bolls (2011). Spreading the Virus: Emotional Tone of Viral Advertising and Its Effect on Forwarding Intentions and Attitudes. Journal of Interactive Advertising 11(2): 1-11. Elberse, A., C. Lee, et al. (2011). Viral Videos: The Dynamics of Online Video Advertising Campaigns, Harvard Business School. Maclnnis, D. and B. Jaworski (1989). Information Processing from Advertisements: Toward an Integrative Framework. Journal of Marketing 53(4): 1-23. MindComet (2006). Viral Marketing: Understanding the Concepts and Benefits of Viral Marketing. The Relationship Agency. T. R. Agency. White Paper. Nelson-Field, K. and G. Klose (2010). The social media leap:integrating social media into marketing strategy. WM3 Your audience = media consumer + generator, Berlin, Germany. O Malley, G. (2011). Media Buyers Predict Upswing In Online Video Ad Spend. MediaPost. Retrieved May, 2011, from URL:http://www.mediapost.com/ publications/?fa=articles.showarticle&art_aid=149695&nid=126334. O Neill, M. (2010). 7 Factors that drive people to share content with their social networks. SocialTimes. Retrieved November 2010, from URL:http://www. socialtimes.com/2010/11/share-content-social-networks/. Peracchio, L. and J. Meyers-Levy (1997). Evaluating Persuasion-Enhancing Techniques from a Resource-Matching Perspective. Journal of Consumer Research 24(2): 178-191. Pintado, I. (2009). Connection Generation; How connection determines our place in society and business. Charleston, SC, BookSurge Publishing. Porter, L. and G. Golan (2006). From subservient chickens to brawny men: a comparison of viral advertising to television advertising. Journal of Interactive Advertising 6(2). PAGE 12 Ehrenberg-Bass Institute Report 58 for Corporate Members

Qualman, E. (2009). Socialnomics: How Social Media Transforms the Way We Live and Do Business. Hoboken, NJ, Wiley, John & Sons, Inc. Reichgut, M. (2011). Social Video And Sharing: What To Expect MediaPost. Retrieved February, 2011, from URL:http://www.mediapost.com/ publications/?fa=articles.showarticle&art_aid=144200. Rimé, B., C. Finkenauer, et al. (1998). Social Sharing of Emotion: New Evidence and New Questions. European Review of Social Psychology 9: 145-189. Rimé, B., P. Philippot, et al. (1992). Long-lasting cognitive and social consequences of emotion: Social sharing and rumination. European Review of Social Psychology 3: 225-258. Romaniuk, J. (2007). Word of mouth and the viewing of television programs. Journal of Advertising Research 47(3). Romaniuk, J. and C. Nguyen (2011). Online versus Offline Word or Mouth; Examining why givers give and how receivers respond. Working Paper, Ehrenberg-Bass Institute for Marketing Science. Southgate, D., N. Westoby, et al. (2010). Creative determinants of viral video viewing. International Journal of Advertising 29(3): 2-14. Tsai, J. (2009). Video is More Than Viral. CRM Magazine 13: 36-42. Weiner, C. (2011). Getting the most from your Social Video. MediaPost. Retrieved March 2011, from URL:http://www.mediapost.com/ publications/?fa=articles.showarticle&art_aid=145175. Ehrenberg-Bass Institute Report 58 for Corporate Members PAGE 13