Academic Research
An Algorithmic Marketing Approach to Analyzing Consumer Well-Being: Incorporating Psychological Factors in Customer Loyalty
In recent years, there has been growing interest in consumer well-being in marketing research. This study examines psychological loyalty, which connects corporate profits with consumer well-being, and proposes an algorithmic marketing approach to analyze survey data from the Matsuya Ginza Department Store to identify specific variables that impact consumer well-being. To clarify the structure between each variable and consumer well-being, we considered various gradient boosting machine learning models, which emphasize classification accuracy for qualitative data, and constructed an ensemble learning model. We also conducted clustering on Matsuya Ginza customers, analyzed the variables that significantly contribute to consumer well-being in different clusters, and developed specific measures to improve products and services. Furthermore, using SHAP (Shapley Additive Explanations) and ICE (Individual Conditional Expectation), we conducted instance-level analysis to show to what extent consumer well-being tends to increase or decrease in relation to important variables for each instance.
In recent years, there has been growing interest in consumer well-being in marketing research. This study examines psychological loyalty, which connects corporate profits with consumer well-being, and proposes an algorithmic marketing approach to analyze survey data from the Matsuya Ginza Department Store to identify specific variables that impact consumer well-being. To clarify the structure between each variable and consumer well-being, we considered various gradient boosting machine learning models, which emphasize classification accuracy for qualitative data, and constructed an ensemble learning model. We also conducted clustering on Matsuya Ginza customers, analyzed the variables that significantly contribute to consumer well-being in different clusters, and developed specific measures to improve products and services. Furthermore, using SHAP (Shapley Additive Explanations) and ICE (Individual Conditional Expectation), we conducted instance-level analysis to show to what extent consumer well-being tends to increase or decrease in relation to important variables for each instance.
Infusing Affective Computing Models into Advertising Research on Emotions
This article discusses promising avenues for integrating affective computational approaches into advertising research on emotion. We review affective computing methods for different modalities—text, visual, and audio—and present advertising research examples and computational tools for each modality. We discuss different state-of-the-art multimodal tools for emotion recognition and propose an evaluation framework for advertising researchers to compare and select appropriate affective computing models. Finally, we discuss how affective computing approaches may fill some research gaps to advance emotion-based advertising research and theory building. This paper contributes theoretical insights, ethical considerations, and practical guidelines essential for the methodological advancement of the emerging field of computational advertising research.
Paper Link: https://doi.org/10.1080/00913367.2024.2409254
Authors: Taylor Jing Wen, Ching-Hua Chuan, George Anghelcev, Sela Sar,Joseph T. Yu, Yanzhen Xu
ABSTRACT
This article discusses promising avenues for integrating affective computational approaches into advertising research on emotion. We review affective computing methods for different modalities—text, visual, and audio—and present advertising research examples and computational tools for each modality. We discuss different state-of-the-art multimodal tools for emotion recognition and propose an evaluation framework for advertising researchers to compare and select appropriate affective computing models. Finally, we discuss how affective computing approaches may fill some research gaps to advance emotion-based advertising research and theory building. This paper contributes theoretical insights, ethical considerations, and practical guidelines essential for the methodological advancement of the emerging field of computational advertising research.
How to Mitigate Fashion Subscription Hesitation: Two-Step Exploration Using Theory-Based Causal Modeling and Machine Learning Predictive Modeling
Although a fashion subscription offers significant environmental benefits by transforming physical products into shared services, most customers are reluctant to adopt it. This hesitation, exacerbated by poor communication from brands that primarily emphasize its personal benefits, hinders its sustainable growth. This study aims to examine specifically which concerns increase hesitation, and the role of explicitly informing consumers about the service’s environmental benefits in mitigating the impact of consumer concerns on their hesitation. Data were collected through an online experiment with more than a thousand U.S. adults nationwide and analyzed using a two-step analysis. First, theory-based causal modeling was conducted to examine the effects of consumer concerns on hesitation, accounting for ambivalence as a mediator and informed environmental benefits as a moderator. Second, machine learning was used to cross-validate the findings. Results show that certain types of consumer concerns increase hesitation, significantly mediated by ambivalence, and confirm that informed environmental benefits mitigate the effects of some concerns on hesitation. This study contributes to building on the hierarchy of effects theory by exploring negatively nuanced constructs – concerns, ambivalence and hesitation – beyond the traditional constructs representing the cognitive, affective and conative stages of consumer decision-making. Findings provide strategic guidance to brands on how to communicate the new service to consumers. Leveraging theory-based causal modeling with machine learning-based predictive modeling provides a novel methodological approach to explaining and predicting consumer hesitation toward new services.
Authors: Jiyun Kang, Catherine Johnson, Wookjae Heo, Jisu Jang
Although a fashion subscription offers significant environmental benefits by transforming physical products into shared services, most customers are reluctant to adopt it. This hesitation, exacerbated by poor communication from brands that primarily emphasize its personal benefits, hinders its sustainable growth. This study aims to examine specifically which concerns increase hesitation, and the role of explicitly informing consumers about the service’s environmental benefits in mitigating the impact of consumer concerns on their hesitation.
Data were collected through an online experiment with more than a thousand U.S. adults nationwide and analyzed using a two-step analysis. First, theory-based causal modeling was conducted to examine the effects of consumer concerns on hesitation, accounting for ambivalence as a mediator and informed environmental benefits as a moderator. Second, machine learning was used to cross-validate the findings.
Results show that certain types of consumer concerns increase hesitation, significantly mediated by ambivalence, and confirm that informed environmental benefits mitigate the effects of some concerns on hesitation.
This study contributes to building on the hierarchy of effects theory by exploring negatively nuanced constructs – concerns, ambivalence and hesitation – beyond the traditional constructs representing the cognitive, affective and conative stages of consumer decision-making. Findings provide strategic guidance to brands on how to communicate the new service to consumers. Leveraging theory-based causal modeling with machine learning-based predictive modeling provides a novel methodological approach to explaining and predicting consumer hesitation toward new services.
The Service Robot Customer Experience (SR-CX): A Matter of AI Intelligences and Customer Service Goals
This research aims to generate a nuanced understanding of service robot customer experiences (SR-CX). Specifically, this work (1) empirically explores the impact of different AI intelligences (mechanical, thinking, and feeling AI) on SR-CX (i.e., SR-CX strength and SR-CX dimensionality) and its downstream impact on important service outcomes (i.e., overall service experience and service usage intentions), and (2) considers the moderating role of consumer service goals (hedonic versus utilitarian). Drawing on insights from two field studies and two online experiments, this research demonstrates that SR-CX—which is impacted differently by varying AI intelligences—affects service outcomes. Specifically, more sophisticated AI intelligences lead to enhanced service outcomes for customers with hedonic service goals across settings by strengthening SR-CX and triggering a more extensive set of SR-CX dimensions. This pattern, however, is less clear for customers driven by utilitarian service goals. For these customers, the role of SR-CX strength and SR-CX dimensionality varies across settings. These findings, which empirically support the importance of SR-CX, may help organizations to strategically leverage robots with different intelligence levels along service journeys of customers with different service goals.
Paper Link: https://doi.org/10.1177/10946705241296051
Authors: Bart Larivière, Katrien Verleye, Arne De Keyser, Klaas Koerten, Alexander L. Schmidt
ABSTRACT
This research aims to generate a nuanced understanding of service robot customer experiences (SR-CX). Specifically, this work (1) empirically explores the impact of different AI intelligences (mechanical, thinking, and feeling AI) on SR-CX (i.e., SR-CX strength and SR-CX dimensionality) and its downstream impact on important service outcomes (i.e., overall service experience and service usage intentions), and (2) considers the moderating role of consumer service goals (hedonic versus utilitarian). Drawing on insights from two field studies and two online experiments, this research demonstrates that SR-CX—which is impacted differently by varying AI intelligences—affects service outcomes. Specifically, more sophisticated AI intelligences lead to enhanced service outcomes for customers with hedonic service goals across settings by strengthening SR-CX and triggering a more extensive set of SR-CX dimensions. This pattern, however, is less clear for customers driven by utilitarian service goals. For these customers, the role of SR-CX strength and SR-CX dimensionality varies across settings. These findings, which empirically support the importance of SR-CX, may help organizations to strategically leverage robots with different intelligence levels along service journeys of customers with different service goals.
Computer Vision Models for Image Analysis in Advertising Research
This study introduces computer vision models for image analysis in advertising research. It reviews the literature in social science and computer science and identifies three categories and nine types of image analysis. The study uses these categories and types as a framework to select 12 computer vision models and compare them on their capability, accuracy, availability, and usability. Nine models are single-functional models, and three are multi-functional models; all 12 have been used in advertising research. The study also demonstrates how two models are used to classify a sample of image ads and assess the aesthetic scores of these ads to answer the research question about the relationship between content categories and aesthetic scores in image ads. It outlines several key steps for the use of computer vision models in advertising research and proposes future research directions. The study can serve as a guide to advertising researchers.
Paper Link: https://doi.org/10.1080/00913367.2024.2407644
Authors: Hairong Li, Nan Zhang
ABSTRACT
This study introduces computer vision models for image analysis in advertising research. It reviews the literature in social science and computer science and identifies three categories and nine types of image analysis. The study uses these categories and types as a framework to select 12 computer vision models and compare them on their capability, accuracy, availability, and usability. Nine models are single-functional models, and three are multi-functional models; all 12 have been used in advertising research. The study also demonstrates how two models are used to classify a sample of image ads and assess the aesthetic scores of these ads to answer the research question about the relationship between content categories and aesthetic scores in image ads. It outlines several key steps for the use of computer vision models in advertising research and proposes future research directions. The study can serve as a guide to advertising researchers.
The Role of Psychological Distance in Enhancing Identity-Relevant Brand Awareness
Advertisers are increasingly prioritizing brand awareness, as it has become a top marketing objective, yet market trends suggest that much of consumers’ brand knowledge is not easily retrieved from memory. To inform advertising strategies designed to address this issue, we examine how self-identity–driven psychological distance affects the type of brand information that consumers are more readily able to retrieve from memory. Study 1 demonstrates that the psychological distance associated with a self-identity influences consumers’ ability to recall concrete (vs. abstract) identity-relevant brand knowledge. For identity-congruent brands related to a current (future) self-identity, consumers elaborate on and process more concrete (abstract) brand information, resulting in stronger brand associations that are more easily recalled from memory. Study 2 leverages this effect to enhance the recall of brand knowledge by matching the psychological distance of a self-identity with a construal mindset. Study 3 demonstrates that retrieval of newly learned brand information can be enhanced by matching the psychological distance of a self-identity with the concreteness of advertising messaging and identifies consumers for whom this effect may not occur (i.e., those with high product involvement). The research makes important theoretical contributions and suggests actionable advertising strategies for enhancing the retrieval of brand knowledge.
Paper Link: https://doi.org/10.1080/00913367.2024.2343287
Authors: Scott Connors, Katie Spangenberg
ABSTRACT
Advertisers are increasingly prioritizing brand awareness, as it has become a top marketing objective, yet market trends suggest that much of consumers’ brand knowledge is not easily retrieved from memory. To inform advertising strategies designed to address this issue, we examine how self-identity–driven psychological distance affects the type of brand information that consumers are more readily able to retrieve from memory. Study 1 demonstrates that the psychological distance associated with a self-identity influences consumers’ ability to recall concrete (vs. abstract) identity-relevant brand knowledge. For identity-congruent brands related to a current (future) self-identity, consumers elaborate on and process more concrete (abstract) brand information, resulting in stronger brand associations that are more easily recalled from memory. Study 2 leverages this effect to enhance the recall of brand knowledge by matching the psychological distance of a self-identity with a construal mindset. Study 3 demonstrates that retrieval of newly learned brand information can be enhanced by matching the psychological distance of a self-identity with the concreteness of advertising messaging and identifies consumers for whom this effect may not occur (i.e., those with high product involvement). The research makes important theoretical contributions and suggests actionable advertising strategies for enhancing the retrieval of brand knowledge.
When Social Media Gets Political: How Message–Platform Match Affects Consumer Responses to Brand Activism Advertising
Social media has sparked a surge in online activism and sociopolitical movements. Numerous companies have also launched brand activism advertising campaigns to voice their stances on sociopolitical causes. However, it remains unclear how and why consumer reactions to brand activism ads vary across different social media platforms. To bridge this gap, we introduced a novel characteristic of social media—level of politicization—and examined how message–platform congruence in terms of level of politicization affects consumer responses to brand activism ads. In Study 1, we found that on a less politicized platform, a less politicized message (vs. a more politicized message) reduced ad intrusiveness, which in turn positively affected consumers’ ad attitudes and purchase intentions toward the brand. On a more politicized platform, consumers exhibited equivalent levels of responses regardless of the message type. In Study 2, we further identified consumers’ level of issue support as a boundary condition of the message–platform congruence effect, such that the congruence effect was larger when issue support was low.
Paper Link: https://doi.org/10.1080/00913367.2024.2347271
Authors: Xuan Zhou, Chen Lou, Xun (Irene) Huang
ABSTRACT
Social media has sparked a surge in online activism and sociopolitical movements. Numerous companies have also launched brand activism advertising campaigns to voice their stances on sociopolitical causes. However, it remains unclear how and why consumer reactions to brand activism ads vary across different social media platforms. To bridge this gap, we introduced a novel characteristic of social media—level of politicization—and examined how message–platform congruence in terms of level of politicization affects consumer responses to brand activism ads. In Study 1, we found that on a less politicized platform, a less politicized message (vs. a more politicized message) reduced ad intrusiveness, which in turn positively affected consumers’ ad attitudes and purchase intentions toward the brand. On a more politicized platform, consumers exhibited equivalent levels of responses regardless of the message type. In Study 2, we further identified consumers’ level of issue support as a boundary condition of the message–platform congruence effect, such that the congruence effect was larger when issue support was low.