Customer Service
An AI Method to Score Celebrity Visual Potential
Xiaohang Feng et al.
Journal of Marketing Research, forthcoming
Abstract:
It has long been a mantra of marketing practice that, particularly in low-involvement situations, spokespeople should be physically attractive. This paper suggests there is a higher probability of gaining fame and influence (i.e., celebrity potential) than is captured by attractiveness or typicality. The authors identify 11 facial features that may predict celebrity potential by virtue of their purported relationship with charisma and resulting personality trait inferences. Using machine learning methods and a sample of 22,000 faces, the authors calculate the direction and strength of the correlation of each feature with celebrity potential. The model is 95.92% accurate in predicting whether a given face belongs to a celebrity or noncelebrity, and it allows calculating a celebrity visual potential (CVP) metric for any face. Two controlled experiments and two studies using photographs of faces of Instagram and LinkedIn users further validate that the model-generated CVP is consistent with human-rated CVP, showing predictive power above and beyond facial typicality and averageness. This paper challenges prior assumptions about the importance of attractiveness in spokesperson choice, offers a useful additional metric for marketers, and provides novel insights about the relative importance of various inferred personality traits for celebrity potential.
Toxic Content and User Engagement on Social Media: Evidence from a Field Experiment
George Beknazar-Yuzbashev et al.
Columbia University Working Paper, January 2025
Abstract:
Most social media users have encountered harassment online, but there is scarce evidence of how this type of toxic content impacts engagement. In a pre-registered browser extension field experiment, we randomly hid toxic content for six weeks on Facebook, Twitter, and YouTube. Lowering exposure to toxicity reduced advertising impressions, time spent, and other measures of engagement, and reduced the toxicity of user-generated content. A survey experiment provides evidence that toxicity triggers curiosity and that engagement and welfare are not necessarily aligned. Taken together, our results suggest that platforms face a trade-off between curbing toxicity and increasing engagement.
The Slang Paradox: Connecting or Disconnecting with Consumers?
Bryce Pyrah et al.
Journal of Marketing Research, forthcoming
Abstract:
Many brands use slang in their marketing communications to connect with consumers. However, through analyses of two Twitter datasets and four experiments, we find that the use of slang in marketing communications could have unintended negative consequences for brands. We theorize and demonstrate that slang use in marketing communications can be perceived as inauthentic because such messages violate consumer expectations, leading to less favorable brand perceptions. To refine our understanding, we test conditions under which slang use aligns more with consumer expectations and is therefore perceived as less inauthentic. Specifically, we find that source characteristics play an important role. First, we show that brand personality impacts this process: Consumers have less favorable attitudes when sincere brands use slang, but not when exciting brands use slang. Second, when influencers (instead of brands themselves) publish brand messages, the negative effect of slang disappears. In addition to contributing new theoretical insights, this research provides practical guidance on effective social media engagement strategies by identifying conditions when slang use is perceived as inappropriate or not in marketing communications.
The memorability of voices is predictable and consistent across listeners
Cambria Revsine, Esther Goldberg & Wilma Bainbridge
Nature Human Behaviour, forthcoming
Abstract:
Memorability, the likelihood that a stimulus is remembered, is an intrinsic stimulus property that is highly consistent across people -- participants tend to remember or forget the same faces, objects and more. However, these consistencies in memory have thus far only been observed for visual stimuli. Here we investigated memorability in the auditory domain, collecting recognition memory scores from over 3,000 participants listening to a sequence of speakers saying the same sentence. We found significant consistency across participants in their memory for voice clips and for speakers across different utterances. Regression models incorporating both low-level (for example, fundamental frequency) and high-level (for example, dialect) voice properties were significantly predictive of memorability and generalized out of sample, supporting an inherent memorability of speakers' voices. These results provide strong evidence that listeners are similar in the voices they remember, which can be reliably predicted by quantifiable low-level acoustic features.
Quality Certifications Influence User-Generated Ratings
Matt Meister & Nicholas Reinholtz
Journal of Consumer Research, forthcoming
Abstract:
Platforms present various certifications to signal the quality of their offerings to prospective consumers. For example, Airbnb.com designates some hosts as "Superhosts" to distinguish properties that provide superior experiences. Platforms also present user-generated ratings -- typically elicited and presented as "star ratings" -- from their customers for the same purpose. This research investigates the interaction of these signals of quality and suggests a potential downside to platform-provided certifications: They decrease subsequent ratings. In an analysis of over 1,500,000 ratings from Airbnb.com and three follow-up studies, we find that properties with the superhost designation receive lower ratings. We assess the robustness of this result in several ways, including comparing ratings on Airbnb with those for the same property of Vrbo. In three follow-up experiments, we find that the net effect of certifications can lead to reduced choice share: The positive effect of signaling quality is more than offset by the negative effect of reduced ratings. This suggests that consumers are not sufficiently aware of this effect of quality certifications on ratings when choosing.
Overestimating Stars, Underestimating Numbers: The Hidden Impact of Rating Formats
Deepak Sirwani, Srishti Kumar & Manoj Thomas
Journal of Marketing Research, forthcoming
Abstract:
Some retailers use stars, while others use Arabic numerals to present product ratings. Do consumers evaluate product ratings differently depending on the format? Which format more accurately represents the true magnitude of ratings? Across 12 experiments, we find that neither format is veridical. Consumers overestimate fractional star ratings and underestimate fractional Arabic numerals (e.g., 3.5). The overestimation of graphical ratings arises from the visual-completion effect: when the visual system perceives an incomplete image of a star, it instinctively activates the complete image, causing consumers to anchor their magnitude judgments on rounded-up numbers. Importantly, our results show that this overestimation of star ratings can be mitigated by using visually complete stars. Conversely, the underestimation of Arabic numeral ratings stems from the left-digit effect, which leads consumers to anchor magnitude judgments on rounded-down numbers (i.e., evaluation of 3.5 is anchored on digit 3). Thus, both star and Arabic numeral ratings are systematically misestimated by consumers, with the extent of misestimation varying based on the fractional value and the star-filling technique employed. These findings demonstrate that prevalent rating formats are misleading, highlighting the need for new industry standards.