Deep-Dive Research Spotlight: Professor Grant Packard
Language is more central to business than ever before. Social media and online review text are mined for consumer sentiment and behavioral prediction. Customer service chatbots and digital advertisers test and learn how subtle language variations impact satisfaction, clicks, and purchases: often in near real-time. The advent of large language models and accessible ways to use them (e.g., ChatGPT) are transforming a huge range of organizational efforts. One of Schulich’s marketing professors is a leader in this burgeoning field of research, revealing how language reflects consumer attitudes, impacts brand and employee perceptions, and shapes behavior.
Grant Packard, Associate Professor of Marketing and Director of the Master of Marketing degree program at Schulich, has been engaged in research on language in the marketplace since 2012. Since that time, he has published over 20 articles on language-driven insight in elite academic and managerial outlets such as the Journal of Marketing Research, Journal of Consumer Research, Journal of Consumer Psychology, Harvard Business Review, MIT Sloan Management Review and others. His research has been covered by media including The Globe and Mail, Harvard Business Review, The Wall Street Journal, CBC, NBC and many more. Packard has been invited to present more than 70 times at conferences, symposia, and corporate events. Most recently, he’s published a series of articles and run workshops to help scholars and organizations learn how to apply natural language processing and text analysis methods to their own efforts.
The increasingly clear value of language for managerial and behavioral insight has led to a massive boom in research on the subject, with an over 300% increase in such research published in the last decade. Over two-thirds of consumer research scholars and a majority of organizations now use language analysis and modeling to inform efforts in marketing and beyond. Due in part to this boom, Packard is one of just a handful of scholars worldwide who currently serves as an Associate Editor at three of the top five academic marketing journals, as ranked by the Financial Times.
What are some of the key insights Packard and his colleagues have discovered in their exploration of marketplace language? We sat down with him to learn more.
Online Reviews and Social Media
For decades, research interested in language in the market was focused on marketing communication, asking whether firms used the right words in advertising. But starting around 15 years ago, the scale of digital media and e-commerce activity and ‘big data’ capabilities made massive amounts of consumer-generated text accessible to organizations and scholars.
Packard’s dissertation research at the University of Michigan was part of the first wave of research diving deep into consumer language in online reviews. Consumers turn to reviews in part because they seem more trustworthy than advertising. But Packard’s research revealed several ways this trust may be misguided. One article revealed that people who feel like they don’t know enough about a product category are significantly more likely to write reviews than those who feel confident in their knowledge. They do this to compensate for insecurity in their category knowledge. Someone who feels as if they don’t know a lot about cars, for example, is more likely to go out of their way to signal their expertise about cars, and why the one they chose is the best option.
Another of Packard’s dissertation papers showed how the effect of boasting or bragging about one’s own category knowledge in online settings depends heavily on other trust cues. If received from someone you know (or feel like you know), boasting enhances persuasion. But if that person is (or feels like) a stranger, the same behavior makes you trust them less, and makes you less interested in the product they’re talking about.
Packard has also looked at particular parts of speech (e.g., verbs) in digital media with a co-author from The Wharton School at the University of Pennsylvania. Unsurprisingly, verbs that convey explicit endorsement (“I recommend this wine”) are more persuasive than those that just signals one’s own attitudes (“I enjoyed this wine”). But experts tend to avoid explicit endorsements because they know others might not have the same preferences. Packard and his co-author showed this pattern of language use can cause people to choose objectively inferior products because they end up following category novices’ recommendations. Other work they’ve published on verb tense revealed that using present, rather than past, tense to share the same attitude (“That movie is/was great!”) is more persuasive.
Cultural Products
A second area of Packard’s research focuses on language in the marketplace of ideas. Songs, movies, books, articles and other cultural products seem likely to succeed or fail, in part, on the words contained within them. Packard and his colleagues have investigated this phenomena in a variety of ways.
In the arts, one project revealed the importance of lyrical novelty when it comes to a song’s commercial success (e.g., Billboard chart rankings). Using a natural language processing method known as latent Dirichlet allocation, Packard and his co-author revealed that artists who lyrically push the bounds of their genre were more successful. Drake’s success, for example, is linked in part to his willingness to signal personal and emotional uncertainty in his lyrics, which is atypical in rap and hip hop.
Another article revealed the importance of second person pronouns in this setting. “You” pronouns in song lyrics, for example, engage listeners because they helps the listener imagine a “you” in their own life. So, when Whitney Houston sings “I will always love you,” audiences imagine someone they (not Whitney) will always love, which makes them like the song more.
Other cultural products research Packard has been involved with in this domain has revealed how narrative dynamics (e.g., the pace or circuitousness of a movie script) are linked to consumer reception of cultural products, and how linguistic style has a significant impact on the adoption of new ideas even in a domain where content should be paramount: academic articles.
Sales and Customer Service
While the language used by marketers, consumers, and in cultural products has seen a lot of attention, sales and service conversations have seen relatively little attention. This is surprising given firms spend, on average, more on these efforts than any other aspect of marketing, and more than any other human resource investment. The dearth of research in this area seems due in part to the difficulty of observing and capturing this more private form of marketplace language, whether it happens in person, by email, or in online chats.
But Packard and colleagues from the University of Alberta, University of Michigan, and Wharton have led a concerted effort to shed light on understanding customer-firm conversations. His first work in this area returned to the personal pronouns he studied in song lyrics. Where he found employees avoid referring to themselves as an individual, instead talking about themselves as a member of the firm ‘team’ (“We can help with that” rather than “I can help with that”), this was exactly the opposite of what they should be doing to satisfy customers and increase sales. In another article, Packard’s research team studied the importance of concrete language. Using specific, concrete words signals that the employee is paying attention to the customer’s needs (e.g., “Did you find the blue jeans you were looking for?”) rather than generically going through the motions with more abstract words (e.g., pants, items).
Most recently, Packard and colleagues from Cheung Kong GSB (Shanghai) and Wharton introduced a new approach that combines dynamic modeling and machine learning to identify when language matters. Rather than merely saying employees should speak concretely, whether this matters could be contingent on when in a conversation it is used. The approach’s potential was demonstrated by challenging the ‘warmth competence paradox’, a longstanding recommendation from psychology and management research that people should present themselves as warm or competent, but never both. Packard and colleague’s research updated this belief by revealing that using warm language is more important at the start and end of an interaction, while competent language matters more only in the middle ‘business’ part. In essence, firms can maximize satisfaction and purchases by being both warm and competent in one interaction, but at appropriate times.
What’s Next
Packard’s latest research is turning to receptiveness to political messages from ‘across the aisle.’ His research team is using large language models, corpus dictionaries, and human judgment to gain insight on language cues that might encourage people to process and consider message content from an ‘opposition’ politician rather than quickly rejecting it due only to the source. He hopes this effort will contribute to efforts to reduce political polarization, a widely recognized challenge to individual, interpersonal, and societal well-being in Canada and beyond.
Packard noted the importance of Schulich’s community of world-leading researchers in encouraging and supporting his efforts. Beyond the research culture and events at Schulich, he’s benefitted greatly from an internal research grant and the Schulich Research Excellence Fellowship. In addition, Packard noted the importance of the Schulich Research Office’s support in winning external grant applications (e.g., SSHRC, Marketing Science Institute). He also enjoys engaging in the development of future professors by supervising student research within the Schulich PhD program and with students from schools in the U.S., Asia, and Europe with whom he is currently working.
This article previously appeared in the annual “Spotlight on Research”. It has been reformatted and edited for digital distribution.