Two Essays on Unsolicited Sponsored Content: In-stream Sponsored Content and Virtual Influencer Endorsement

關於非應邀贊助內容的兩篇論文:資訊流內贊助內容和虛擬偶像代言

Student thesis: Doctoral Thesis

View graph of relations

Author(s)

Related Research Unit(s)

Detail(s)

Awarding Institution
Supervisors/Advisors
Award date8 Jan 2024

Abstract

Nowadays, apart from e-commerce websites, consumers increasingly rely on other types of platforms (e.g., news websites and social media) for information on products or services. This trend enables platforms to generate revenue by pushing unsolicited sponsored content to consumers. However, consumers may ignore or resist unsolicited sponsored content initiated by brands and focus on solicited organic content.

To solve this problem, on the one hand, brands are increasingly relying on the attractiveness of platform-generated content (i.e., organic content) and inserting sponsored content in the stream of organic content (i.e., in-stream sponsored content). For example, news websites such as USA Today and Yahoo News insert ads into the stream of news articles. On the other hand, with the development of the metaverse, brands are embracing the content generated by virtual users such as virtual influencers (i.e., “user”-generated content) by sponsoring virtual influencers who are computer-generated imagery as digital recommenders to promote products. For example, major fashion brands such as Prada sponsor Lil Miquela, a famous virtual influencer, to endorse clothing on social media.

An important research field of Information Systems (IS) is how to recommend products to consumers to guide consumer choices (e.g., Benbasat & Wang, 2005; Li & Karahanna, 2015; Qiu & Benbasat, 2009; Xiao & Benbasat, 2007). Unsolicited sponsored content is one type of product recommendations that favorably recommends sponsors’ products to consumers without prior consent from consumers. The format of presenting unsolicited sponsored product recommendations is likely to affect consumers’ responses because unsolicited sponsored product recommendations are disruptive to consumers’ original activity. Nevertheless, previous IS literature has mainly focused on how to improve consumers’ responses to sponsored product recommendations by adding external cues (e.g., Deng et al., 2022; Wang & Wang, 2019; Xiao & Benbasat, 2015), for example, adding ad disclosures to sponsored product recommendations increases users’ perceived integrity of sponsored product recommendation agents (e.g., Wang & Wang, 2019), thus ignoring how the format of presenting unsolicited sponsored product recommendations affects consumers’ responses. Therefore, this thesis focuses on the format of presenting unsolicited sponsored product recommendations in the contexts of in-stream sponsored content on news websites (i.e., the format of presenting recommendations) and virtual influencer endorsement on social media (i.e., the format of presenting digital recommenders).

Study 1 focuses on the format of presenting unsolicited sponsored product recommendations in the context of in-stream sponsored content on news websites. Currently, platforms tend to present in-stream sponsored content in a native way, that is, presenting in-stream sponsored content similar to organic content, probably to avoid disturbing consumers’ platform experience. However, this common practice may mislead consumers even with ad disclosures. It is unknown whether the congruence between in-stream sponsored content and organic content always benefits consumers as expected by platforms. Based on ad-context congruence literature, I defined the congruence between in-stream sponsored content and organic content as nativeness with three dimensions: topic, design, and style nativeness (i.e., the congruence in the topic, visual design, and communication style). Drawing from the information presentation literature and persuasion knowledge model (PKM), I then investigated how nativeness affects persuasion knowledge for both skeptical and non-skeptical consumers and how nativeness affects consumers’ behavioral responses (i.e., attention and clicks on in-stream sponsored content) on multiple-layer web pages.

Study 1 conducted a series of laboratory experiments to validate the hypotheses. The results of Study 1 show that nativeness reduces ad recognition on the front pages while increases inferences of manipulative intent (IMI) on the landing pages. Moreover, ad skepticism moderates the effects of nativeness on IMI. However, different dimensions of nativeness affect ad recognition, IMI, attention, and clicks of in-stream sponsored content differently. For example, style nativeness is a stronger predictor of ad recognition and IMI than design nativeness.

Study 2 focuses on the format of presenting digital recommenders of unsolicited sponsored product recommendations in the context of virtual influencer endorsement on social media. Different from human influencers with intrinsic desires to share content, virtual influencers are virtual digital humans who cannot consciously try any product and are controlled by companies for profit. In the beginning, it is difficult for consumers to understand how virtual influencers can have a passionate desire for products since virtual influencers do not exist physically (Andersson & Sobek, 2020). Due to the novelty of virtual influencers, consumers tend to perceive virtual influencers as lacking authenticity when first encountering virtual influencers (Andersson & Sobek, 2020). However, how to remedy virtual influencers’ lack of authenticity is an under-investigated topic. It is unknown whether virtual influencers should be humanoid in appearances (i.e., form realism) or behaviors (i.e., behavioral realism) for the sake of authenticity. Consumers tend to form initial impressions of others based on the visual appearance and nonverbal behavior of others, and initial impressions set the tone of subsequent interaction (Bergmann et al., 2012). Therefore, Study 2 focuses on the impact of virtual influencers’ form and behavioral realism on virtual influencer authenticity during consumers’ initial encounters with virtual influencers. Drawing upon the literature on the perceived authenticity of humans and agents (i.e., software agents such as chatbots), I defined virtual influencer authenticity (i.e., consumers’ perceived authenticity of virtual influencers) with three dimensions: autonomy, congruence, and genuineness. I then examined how virtual influencers’ behavioral realism with animation affects virtual influencer authenticity, how virtual influencers’ behavioral realism with product usage affects virtual influencer authenticity under different product appeals (i.e., desirability or feasibility appeals) from the perspective of conceptual fluency, and how virtual influencers’ form realism affects virtual influencer authenticity under different product forms (i.e., anthropomorphized or normal display of products) from the perspective of perceptual fluency in the initial stages of human-virtual influencer interaction.

Study 2 conducted a series of online experiments in the context of consumers’ initial encounters with virtual influencers on social media combined with a data set from Sina Weibo to validate the hypotheses. The results of Study 2 indicate that virtual influencers’ behavioral realism with animation increases virtual influencer authenticity. Virtual influencers’ behavioral realism with product usage increases virtual influencer authenticity only when the product appeal is feasibility instead of desirability. Moreover, virtual influencers’ form realism increases virtual influencer authenticity only when the product form is normal instead of anthropomorphized. In addition, three dimensions of virtual influencer authenticity affect various consumers’ responses differently.

This thesis has significant contributions to research and practice. First, this thesis extends the research scope of previous IS literature on sponsored product recommendations by exploring how the format of presenting unsolicited sponsored product recommendations affects consumers’ responses. Specifically, as for the format of presenting unsolicited sponsored product recommendations, Study 1 extends the scope of ad-context congruence literature to the nativeness presentation of in-stream sponsored content by adding style nativeness as a new dimension of nativeness. Study 1 also contributes to the PKM and information presentation literature by exploring how nativeness affects persuasion knowledge on multiple-layer web pages and proposing one contingent factor (i.e., ad skepticism). As for the format of presenting digital recommenders of unsolicited sponsored product recommendations, Study 2 extends the scope of the literature on the perceived authenticity of humans or agents to virtual influencer authenticity. Study 2 also contributes to the literature on agents’ realism by examining the effects of virtual influencers’ form and behavioral realism (including animation and product usage) on virtual influencer authenticity in the initial stages of human-virtual influencer interaction and proposing two contingent factors (i.e., product forms and appeals). Second, this thesis also provides guidelines for practitioners on how to effectively present unsolicited sponsored product recommendations (e.g., in-stream sponsored content) and digital recommenders of unsolicited sponsored product recommendations (e.g., virtual influencers) on platforms to improve consumers’ responses.

    Research areas

  • In-stream Sponsored Content, Nativeness, Persuasion Knowledge, Virtual Influencer, Realism, Authenticity