“Consumer content sharing is key for the success of any online marketing campaign,” says Francisco Villarroel Ordenes, who studies the intersection of social media marketing, services, and machine learning for textual data. “But the high competition for consumer attention on social media sites is making content managers’ jobs extremely difficult. Brand content is much more likely to end up buried in big data than getting shared.”
To address that challenge, his research focuses on developing strategies that use text-mining to harness brand-related consumer activity. For a recent study, “Cutting Through Content Clutter: How Speech and Image Acts Drive Consumer Sharing of Social Media Brand Messages," which was published last year in the Journal of Consumer Research, Villarroel Ordenes and colleagues looked at 41,000 social media brand posts (on Facebook and Twitter) shared over two years by eight high-profile consumer brands, including Nike and Amazon. They used machine learning to analyze the data, breaking down the factors that made posts more or less likely to be passed along by consumers.
The authors, a group that also included Dhruv Grewal (Babson College), Stephan Ludwig (University of Melbourne), Ko De Ruyter (Kings College), Dominik Mahr (Maastricht University), and Martin Wetzels (Maastricht University), drew on speech act theory, the idea that using language is a way of taking action, rather than simply a way to share information. They assessed how brands’ use of language in posts (to make an assertion, express an emotion, or direct an action) affected consumers’ reactions, and particularly their likelihood of sharing a message.
Inform, Don't Direct
The study’s first finding notes that consumers tune out content that tells them what to do. Directive messages (“Come on Friday for the final sale!”) get less traction than informative (“On Friday we launch our new product”) and emotional brand messages (“We love Fridays”).
“Our findings, across both Facebook and Twitter, resonate with previous research that indicates that messages with socioemotional intentions are more likely to be exchanged,” the authors write. Interestingly, factual (or assertive) posts are more likely to be shared on Twitter than on Facebook, showing that each social media platform has a slightly different audience with particular motivations.
The study also looked at how message-sharing was influenced by the ways brands used rhetorical standbys like alliteration (“Functional, fashionable, formidable”) and repetition (“New year, new car”). The research found that, in general, such messages made more of an impact, especially on Facebook. On Twitter, rhetorical devices generally made users more likely to share posts, but explicit advertising cues—such as a directive message telling them to do something concrete that also has alliteration—really seem to turn them away.
Images matter too. The research analyzed 9,215 brand posts containing images and found that a promotional tweet featuring a directive image (such as a photo of a person pointing toward a Toy Story character) accompanied by a directive message (“Check out the deal of the day”) is less likely to be shared than one with an emotional or informative message (“To infinity and beyond”). Too much direction appears to overburden consumers.
“People don’t like being told to do something on social media,” Villarroel Ordenes says. “They’re used to that in advertising. In social media, brands have to figure out how to join the conversation instead.”
Watch Villarroel Ordenes describe his research in a video taken during the Isenberg Faculty Speakers Series, 2019.