Member Only Access
Summary
Research analyzing how a brand’s Facebook posts can generate e-word of mouth (eWOM) has often relied on content analysis and users’ motivations. A new study offers a big-data approach for predicting the impact of a post’s unique features on eWOM activity, with some unexpected outcomes.
Using software-based regression analysis—a form of statistical modeling—and data from 46 companies listed on the Fortune 100 Index—researchers in South Korea and the U.S. dissected the impact of a post’s message strategies. Results showed that eWOM activity increases when brands use a combination of informational- and emotional-based messaging, including:
- the use of multimedia content, such as photos and videos;
- brand names;
- hashtags;
- subjective words evoking emotional, evaluative and judgmental expression (vs. objective and factual information);
- social-processing words evoking cognitive and emotional processes in the text. Plural pronouns, for example, can make consumers “feel stronger ties and positive emotions…for the brand.”
Contradicting earlier research, the study also showed that using a link in a brand post as an informational message strategy would decrease the number of “likes” from consumers. And in terms of emotional messaging: “The more positively intense a brand post was, the less likely it would receive a positive comment from consumers.”
Taemin Kim (taemin.kim@inu.ac.kr) is an assistant professor in the Department of Mass Communication at Incheon National University, South Korea. His research interests include brand communication in social media, branding and advertising effects.
Hyejin Kim (hkim123@depaul.edu) is as an assistant professor at the College of Communication, DePaul University. Her research interests include social-media advertising and eWOM with computational research approaches.
Yunhwan Kim (yunhwankim2@hufs.ac.kr) received his doctorate in arts in communication from Hankuk University Foreign Studies, Seoul, South Korea. His research interests include analyzing photo and video data with computer vision techniques, and investigating public opinion with agent-based modeling.