The demand for ethical fashion products has grown in recent years, with consumers caring more about how the product is made. Based on the deontological underpinnings of ethical fashion, the present research provides recommendations to ethical fashion brands by leveraging the emotion of awe through visual representations. This research examines 5362 Instagram posts of five ethical fashion brands and shows that the presence of awe-inspiring visual elements increases consumer engagement on social media for ethical fashion brands. Further, this research identifies the moderating roles of the color green and the size of the visual elements in this context. Two follow-up experiments provide causal evidence for these relationships. Experimental evidence for the role of deontological beliefs as a mediator of the relationship between feelings of awe and social media engagement with ethical fashion brands is also provided. Overall, the findings offer a more nuanced understanding of how ethical fashion brands can develop effective visual advertising campaigns on social media and the underlying process driving the effect, which is consistent with deontological ethics.

Artificial Intelligence has outperformed human experts in functional tasks such as chess and baduk. How about creative tasks? This paper evaluates AI's capability in the creative domain compared to human experts, which little research has been conducted so far. We propose a novel Prompt-for-Prompt to generate social media creatives via prompt augmentation by Large Language Models. We take the most popular Instagram posts (with the biggest number of like clicks) in top brands' Instagram accounts to create social media creatives. We give GPT 4 several prompt instructions with text descriptions to generate the most effective prompts for cutting-edge text-to-image generators: Midjourney, DALL E 3, and Stable Diffusion. LLM-augmented prompts can boost AI's abilities by adding objectives, engagement strategy, lighting and brand consistency for social media image creation. We conduct an extensive human evaluation experiment, and find that AI excels human experts, and Midjourney is better than the other text-to-image generators. Surprisingly, unlike conventional wisdom in the social media industry, prompt instruction including eye-catching shows much poorer performance than those including natural. Regarding the type of creatives, AI improves creatives with animals or products but less with real people. Also, AI improves creatives with short text descriptions more than with long text descriptions, because there is more room for AI to augment prompts with shorter descriptions.

Identifying significant features (SFs) is important because they are driving factors of a target outcome. However, it is difficult when they have much more features than observations. The problem becomes more challenging when there are multicollinearity and infrequent common features. In such case, standard explainable methods such as OLS and Lasso often fail to identify many SF. To tackle these problems, we propose a stable model to maximize the number of SFs using selective inference called SFLasso-SI. First, in each point in the regularization path, SFLasso-SI conducts selective inference for conservative significance test. Then, it chooses the optimum value of regularization that maximizes the number of SFs. Our extensive experiments across different types of data - text, image, and video show that our SFLasso-SI can find the biggest number of SFs while maintaining similar prediction accuracy as the benchmarking methods.

PurposeThis paper aims to theorize and investigate the use of effective color features in artificial intelligence (AI) influencers, an emerging marketing trend in the social media context. Design/methodology/approachBy analyzing 6,132 pictures posted by ten AI influencers on Instagram, this paper examines the effect of warm colors in AI influencers' social media posts on consumer responses, and how other color features may moderate the effect of warm color. In addition, two experimental studies reveal the underlying process driving the effect of warm color. FindingsWarmer color generated more favorable consumer responses, with brightness significantly moderating the relationship between warm color and favorable consumer responses. Moreover, the results of the experiments establish that perceived warmth and emotional trust mediate the causal effect of warm colors on consumer responses. Research limitations/implicationsThere is still little understanding about consumer perceptions of AI influencers and their acceptance of AI influencers' product recommendations. As such, this research offers theoretical understanding of the color features influencing the effectiveness of recommendations by AI influencers. Practical implicationsBrands have started deploying AI influencers as their brand ambassadors to make product recommendations, representing a new wave of advertising on social media. The findings will thus benefit marketers in developing effective product recommendations using AI influencers. Originality/valueThe present research provides a novel understanding of how visual features, such as color can influence the effectiveness of AI influencers.

While social media marketing plays an increasingly significant role in the B2B context, there is still limited research examining how to increase B2B social media engagement. The need for more research is becoming more pressing given the recent shift in social media content from text-centric to image-oriented experience. Building on the elaboration likelihood model and employing computerized visual and textual analysis tools, the present research investigates how informational and emotional cues in B2B social media content influence viewer engagement. By analyzing 5362 single-image posts from the Instagram accounts of 10 B2B companies across five industries, this research demonstrates that informational cues are more effective than emotional cues in driving viewer engagement, and that informational cues are even more effective when presented in visual content, as compared to textual content. The findings of this research contribute to the literature on B2B social media literature and offer managerial implications for B2B social media marketing.

There has been a long debate about effective emotional appeals on charity advertisements. While many charity organizations recently shifted from negative emotions to happy emotions on their social media, it is not clearly proven whether this strategy is more effective. The objective of this study is to find more detailed unknown information to optimally use emotional charity advertisements on social media. We investigate the effect of 1) emotional valence, 2) their match between images and textual descriptions, 3) their length, and 4) their post timing on social media engagement. By automatically extracting emotions expressed both in facial images and textual descriptions from 3,066 charity posts from Save the Children's official Instagram account using the computerized emotional content analysis, we provide findings on what, how much, when, and how charity managers can come up with a clear configuration for their social media advertisements.

Purpose For the case of many content features, This paper aims to investigate which content features in video and text ads more contribute to accurately predicting the success of crowdfunding by comparing prediction models. Design/methodology/approach With 1,368 features extracted from 15,195 Kickstarter campaigns in the USA, the authors compare base models such as logistic regression (LR) with tree-based homogeneous ensembles such as eXtreme gradient boosting (XGBoost) and heterogeneous ensembles such as XGBoost + LR. Findings XGBoost shows higher prediction accuracy than LR (82% vs 69%), in contrast to the findings of a previous relevant study. Regarding important content features, humans (e.g. founders) are more important than visual objects (e.g. products). In both spoken and written language, words related to experience (e.g. eat) or perception (e.g. hear) are more important than cognitive (e.g. causation) words. In addition, a focus on the future is more important than a present or past time orientation. Speech aids (see and compare) to complement visual content are also effective and positive tone matters in speech. Research limitations/implications This research makes theoretical contributions by finding more important visuals (human) and language features (experience, perception and future time). Also, in a multimodal context, complementary cues (e.g. speech aids) across different modalities help. Furthermore, the noncontent parts of speech such as positive “tone” or pace of speech are important. Practical implications Founders are encouraged to assess and revise the content of their video or text ads as well as their basic campaign features (e.g. goal, duration and reward) before they launch their campaigns. Next, overly complex ensembles may suffer from overfitting problems. In practice, model validation using unseen data is recommended. Originality/value Rather than reducing the number of content feature dimensions (Kaminski and Hopp, 2020), by enabling advanced prediction models to accommodate many contents features, prediction accuracy rises substantially.

While the importance of electronic Word-of-Mouth (eWOM) for Business-to-Business (B2B) firms is increasing, the use of B2B firm-generated content for driving positive eWOM is less understood. Given the emergence of image-oriented social media platforms, this study investigates how color features increase positive eWOM in the B2B versus B2C context by analyzing 13,356 images on Instagram. The results reveal key differences in color features in the contexts of B2B and B2C. Specifically, cool colors are more appealing in B2B content, while warm colors work better in B2C content. Further, darkness, saturation, and colorfulness moderate the cool effect in B2B content, such that darker, less saturated, and more varied colors increase the effect of cool color. In the B2C context, only colorfulness increases the effect of warm color. The findings of this research contribute to the literature examining the different drivers of eWOM between B2B and B2C social media and offer managerial implications for B2B and B2C firms on ways to encourage positive eWOM.

Research on brand transgressions has suggested that when a brand is involved in a transgression and perceived as harmful, consumers will punish the brand. The present research seeks to extend this literature by investigating how having a cute brand logo may reduce consumer punishment of a transgressing brand. Across five experimental studies, this research shows that a brand logo with high (vs. low) levels of cuteness associated with a transgressing brand can motivate consumers to protect the brand from harm, thus reducing consumer punishment of that brand. Notably, such motivations to protect the brand are driven by an incremental belief about the brand’s development. Moreover, the cuteness effect is attenuated in the case of repeated transgressions. Theoretical contributions to the literature on cuteness, brand logo, and brand transgression are discussed, as well as practical implications.

Purpose Given the socialisation of men and women to their gender roles and expression of emotion, this study aims to investigate whether there are gender differences in the use of emotive language in electronic word-of-mouth (eWOM), specifically in online reviews. The authors propose that female reviewers will use strong emotive terms, such as love, more frequently in online reviews than do male reviewers. The authors further propose that the gender of the reviewer influences audience responses to the reviewer’s use of emotive terms in online reviews. Design/methodology/approach The authors conducted secondary data analysis of restaurant reviews (Study 1) to provide evidence on whether the gender of the reviewer affects the frequency of use of emotive terms in an online review. In addition, three separate experiments (Studies 2–4) were conducted to test the theoretical arguments. Findings The results of the secondary data analysis indicated that female online reviewers used the term “love” much more frequently in their reviews than male reviewers, whereas there was no usage difference for the term “like”. The experimental studies further showed that an emotive review by a male reviewer containing the word “love” resulted in a higher evaluation of the restaurant being reviewed than a non-emotive review containing the word “like”. This difference was stronger when the overall rating was less salient and for consumers who believe (vs do not believe) that men and women use emotional language differently. Research limitations/implications First, the paper extends our understanding of gender differences in emotional expression within the domain of eWOM and online reviews as well as our understanding of consumer responses to these gender differences. Second, the authors identify a boundary condition for these gender effects, namely, the overall rating score. Third, the authors find that consumer beliefs regarding gender stereotypes in emotional expression provide an explanation for these effects. Practical implications The results of the research indicate that the electronic algorithms operating on review sites might be modified in terms of their criteria for selecting the reviews to display to consumers, as consumer decision-makers may find greater utility in reviews written by male reviewers that contain strongly positive emotive terms. Originality/value The research extends the knowledge on gender differences in emotional expression in online reviews by demonstrating the actual usage patterns and differing responses to the emotional expressions of each gender.

There is a growing demand for meat substitutes among consumers, given that excessive meat consumption is associated with negative consequences for personal health and the environment. However, the market shares of such meat substitutes remain low, thus highlighting the need to further investigate how to increase consumer acceptance of meat substitutes. The present research investigates social media data of plant-based meat brands and explores how visual features could lead to a high number of likes, which is a numerical representation of social acceptance. The findings of this research show that social media posts with warm color, vertical symmetry, and horizontal symmetry receive a higher number of likes. Further, there is a joint effect between warm color and vertical symmetry, such that vertical symmetry would strengthen the positive effect of warm color on the number of likes. These findings offer a more nuanced understanding of how to increase consumer acceptance of meat substitutes and how to promote plant-based meat brands in social media.

There is a growing interest in explainable machine learning methods. In our investigation, we have collected heterogeneous features from two series of YouTube video ads and seven series of Instagram picture ads to form our datasets. There are two main challenges that we found in analysing such data: i) multicollinearity and ii) infrequent common features. Due to these issues, standard estimation methods, such as OLS, Lasso, and Elastic-net, are only able to find a small number of significant features. This paper proposes a method called Significant Feature Lasso (SF-Lasso) to maximize model explainability by identifying most of the significant features that affect a target outcome (such as online video and picture ad popularity). Experiments show that SF-Lasso is able to identify much more significant features while maintaining similar prediction accuracy as what Lasso and Elastic-net can obtain. The human evaluation shows that SF-Lasso is better at identifying true features that appeal to ad viewers. We also find that the number of significant features is mainly affected by the model size (i.e., the number of active variables) and the correlations among explanatory variables.

An upcycled product-a new product created from reused waste materials-has two identities: a past identity derived from the source materials and a present identity associated with the product's current value. This research investigates how the lateral display of an upcycled product's past identity relative to its present identity changes consumers' evaluations of upcycled product advertisements and their word-of-mouth (WOM) intentions. Across four studies, including an eye-tracking experiment, we demonstrate that consumers evaluate an upcycled product advertisement more (vs. less) favorably when the product's past identity is presented to the left (vs. the right) of its present identity, which, in turn, affects consumers' WOM on social media. We suggest that this lateral display effect is driven by consumers' spatial representation of time. Our research carries managerial implications for marketing professionals who want to effectively advertise their upcycled products and contributes to the literature on upcycling by demonstrating the role of past-identity location in upcycled product advertisements.

A desirability function approach to a multiresponse problem is proposed considering process parameter fluctuation which may amplify the variance of response. It is called POE (propagation of error), which is defined as the standard deviation of the transmitted variability in the response as a function of process parameters. In order to obtain more robust process parameter setting, a new desirability function is proposed by considering POE as well as distance-to-target of response and response variance. The proposed method is illustrated using a rubber product case in Ribeiro et al. (2000).

A loss function approach to a multiresponse problem is considered, when process parameters are regarded as random variables. The variation of each response may be amplified through so called propagation of error (POE), which is defined as the standard deviation of the transmitted variability in the response as a function of process parameters. The forms of POE for each response and for a pair of responses are proposed and they are reflected in our loss function approach to determine the optimal condition. The proposed method is illustrated using a polymer case. The result is compared with the case where parameter fluctuation is not considered.