Research on the Effectiveness of Virtual Endorsers: A Study Based on the Match-Up Hypothesis and Source Credibility Model

Abstract

In the digital era, businesses are actively integrating advanced technology and innovative marketing strategies to achieve sustained growth. Notably, virtual endorsers play a key role in driving enterprises towards digital transformation in the field of digital marketing. Understanding consumer attitudes towards the use of virtual endorsers in digital marketing is especially important for enterprises employing digital tools to realize a sustainable business model. To this end, this research adopts the match-up hypothesis and source credibility model as its theoretical framework, delving into the impact of virtual endorser–product fit and credibility features (attractiveness, expertise, and trustworthiness) on product attitudes, as well as the interactive effects of these features with product types. We collected feedback data from 376 participants through an online questionnaire and validated our hypotheses using the PLS-SEM model. The results demonstrate that virtual endorser–product fit positively affects credibility, with higher credibility further enhancing consumers’ attitudes toward products. Additionally, the source credibility model partially mediates the relationship between the match-up hypothesis and product attitudes, with attractiveness exerting the most significant impact. Finally, we observed variations in consumer attitudes toward products endorsed by virtual endorsers based on product types. The findings of this study provide a solid theoretical basis for a deeper understanding of consumer attitudes towards the application of virtual endorsers in marketing and offer practical suggestions for businesses to leverage digital tools for sustainable development.

1. Introduction

In recent years, emerging disruptive technologies have spurred rapid societal advancement and altered business practices, furthering the achievement of sustainable development goals. According to the United Nations 2030 Agenda for Sustainable Development, achieving sustainability necessitates consideration of three primary dimensions: environmental, social, and economic [1]. Sustainable development involves ensuring an organization’s financial performance while protecting humanity and the planet. In this context, sustainability can be defined as the capacity of a system to endure over the long term without diminishing or depleting the resources it relies on [2]. Digital technology has created new opportunities for innovation, collaboration, and empowerment, driving sustainable development [3]. Digital channels and media have become key touchpoints in the customer shopping journey, exerting greater influence on consumers compared to traditional media, and are pivotal channels for disseminating sustainable development. According to Henninger et al. [4] (pp. 147–148), digital technology can be a powerful tool for communicating sustainable values and enhancing sustainable retail business models. Thus, digital sustainability refers to the use of digital tools to promote sustainable development goals and organizational activities [4].

Against the backdrop of this paradigm shift, virtual characters created through digital technology, as influencers of purchase decisions, have become particularly important in the study of consumer decision-making. Virtual endorsers created through AI, 3D, and CGI technologies, such as Lil Miquela, Noonoouri, and Imma, have garnered considerable attention on social media and established collaborations with numerous prominent brands, demonstrating their immense potential in marketing [5,6]. These digitally created personas, promoted via social media platforms, transcend the temporal and spatial limitations of traditional endorsers, bringing new trends to the advertising industry [7,8]. Wielki [9] highlighted the growing importance of digital influencers in the functioning of digital promotion ecosystems and their impact on sustainable development. In the fashion industry, companies and brands can strategically utilize digital influence to promote sustainable clothing consumption, particularly in the wake of the COVID-19 pandemic, with an ongoing increase in online sales among younger consumers [4].

Virtual endorsers, as an innovative marketing tool, can enhance brand awareness and consumer engagement, thereby increasing corporate revenue. Due to their relatively low production and maintenance costs, coupled with high customizability and adaptability [10,11], these virtual entities enable businesses to achieve broader market coverage and more precise target audience positioning at lower costs. Moreover, virtual endorsers can also serve as effective vehicles for promoting social responsibility and positive values [12], such as participating in charitable activities or advocating environmental consciousness, thereby enhancing the brand’s sense of social responsibility and image. Due to their digital nature, virtual endorsers play a significant role in reducing the demand for physical resources in traditional advertising campaigns, such as lessening the use of physical materials and alleviating the environmental burden of live events, thus reducing waste generation and carbon emissions, which is positively meaningful for achieving corporate environmental sustainability goals [13]. These digital endorsers enable companies to market in a more responsible and sustainable manner, responding to modern corporate concerns about environmental protection and social responsibility. However, how to properly utilize and manage these virtual endorsers, ensuring their endorsement effectiveness aligns with the long-term strategies of the brand and product, particularly in keeping with the requirements of digital transformation for sustainable development, remains a focal issue of concern for both academia and industry.

In marketing research, the ‘Source Credibility Theory’ and ‘Match-up’ theory are often applied to assess the promotional effect of endorsers [14,15,16,17,18]. The match-up theory elucidates the connection between the fit of endorsers with products and the credibility of the source (including the perceived attractiveness, expertise, and trustworthiness of the information transmitter) [14,15,16,17,18]. For example, the fit between endorsers and the products they promote can influence consumers’ attitudes toward the product, brand, and advertising [17,18]. If this match-up is insufficient, consumers may perceive it as product advertising paid for by promotional fees, leading to negative reviews and a decrease in trust [19].

Therefore, a deeper exploration of how virtual endorsers influence consumer attitudes and behaviors is of significant importance for businesses seeking to leverage the power of digital technology to promote sustainable development. Indeed, scholars such as Franke et al. [6] and Thomas and Fowler [20] have raised the following question: is the efficacy of virtual endorsers related to the category of the promoted product? This research gap holds not only theoretical significance for further exploration but also offers practical guidance for businesses in their digital strategies at the managerial level.

Building upon this, the present study aims to integrate the match-up hypothesis and the source credibility model to investigate whether virtual endorsers can generate positive effects like human endorsers in advertising or vouching for products, and whether these effects vary by product type. The findings of this study are expected to provide insights for scholars and businesses, allowing them to better understand and utilize virtual digital technologies for sustainability in the context of digital sustainable development.

2. Theoretical Foundation and Hypotheses

2.1. Virtual Endorsers

Employing virtual characters for commercial objectives is by no means a recent innovation. Take, for instance, the music sector, where virtual avatars such as Hatsune Miku, who represents synthesized vocal sound created with ‘VOCALOID,’ have garnered substantial success [6]. With advancements in AI and CGI technology, companies can digitally create ‘virtual endorsers’ with unique humanoid visual representations based on the images they desire, and they are seen as ‘credible brand builders in one or more domains’ [21]. As noted by Thomas and Fowler [20], this phenomenon of ‘non-human substitutes for traditional endorsers’ is finding its place in brand communication. Due to these factors, virtual endorsers are increasingly gaining favor among brands. According to Territory Influence, the virtual endorser market is currently valued at USD 4.6 billion, with an anticipated growth of 26% by 2025, offering substantial business opportunities. In 2022, 35% of American consumers made purchases after hearing virtual endorsers promote products or services, with millennials and Gen Z members comprising 40% of these customers [22].

While virtual endorsers are computer-generated, their roles on social media are like those of humans, i.e., content creators and personality performers [23]. On the other hand, due to their digital nature, virtual endorsers are exempt from diseases, fatigue, aging, or other human physiological limitations, and are not constrained by time or geography, enabling brands to use them at any time and place [5,7]. Nowadays, marketers can digitally create virtual endorsers in accordance with the desired image and employ them for advertisements and social media promotion. The efficacy of virtual endorsers lies in their ability to guide users to desired outcomes to varying degrees, which may be influenced by their appearance and visual performance, thereby enhancing the transmission of information and persuasiveness [20]. With their appeal, human-like functionality, and audiovisual characteristics, virtual endorsers, through unique attributes and frequent interactions, influence consumer attitudes and behaviors and are considered to be effective marketing tools as a new form of substitute for human endorsers [24,25].

However, research has pointed out that virtual endorsers lack authenticity and transparency, which could diminish consumers’ positive response to sponsored messages from such endorsers, thereby affecting their credibility and trustworthiness [7,26]. Therefore, for artificial constructs like virtual endorsers, understanding their marketing effectiveness is crucial; however, comprehensive research on the marketing impact of virtual endorsers is currently scarce in the literature. To fill these research gaps, we employ the match-up hypothesis and source credibility theory to explore how the perceived credibility of virtual endorsers affects product attitudes, which helps us to gain a more comprehensive understanding of the value of virtual endorsers in advertising effectiveness.

2.2. Match-Up Hypothesis

The match-up hypothesis proposes that if the image or personality of an endorser fits well with the product they are endorsing, the advertising effect will be stronger [16,17,18,27]. This fit (consistency) has been thoroughly confirmed in the traditional advertising field and is considered one of the most important aspects affecting consumer attitudes and behavioral intentions [14,27]. Specifically, fit refers to the harmony between certain characteristics of the endorser and the characteristics of the endorsed product [16,27]. Empirical studies show that the match between the endorser and the product has a positive and direct impact on advertising effects, with the characteristics of the endorser possibly being transferred to the endorsed product [16]. For example, in actual market applications, athletes often endorse sports-related products [28], and celebrities with outstanding physiques may endorse products related to attractiveness [15]. Till and Busler’s [16] research also supports this viewpoint, as they found that athletes are more suitable than actors for endorsing energy bars, but not for candy bars. However, it is worth noting that if consumers fail to perceive consistency between the endorser and the product, the product may garner negative evaluations from consumers, potentially resulting in adverse advertising effects [27,29].

In recent years, we have observed an increasing prevalence of virtual endorsers. The reasons behind this trend are not solely driven by the sustainability of the digital economy but also stem from their potential value in long-term brand building. Compared to traditional human endorsers, virtual endorsers, as computer-generated digital figures, possess highly customizable characteristics. This allows brands to mold their image flexibly, ensuring a perfect alignment with the promoted product or service, thereby achieving a high level of brand consistency [7,25]. Moreover, the unique novelty of virtual endorsers holds considerable allure for younger consumer demographics who seek novel experiences [6]. Against the backdrop of corporate digital transformation, understanding the long-term implications of this new endorsement paradigm and its contributions to sustainability is crucial. Given this, the current study endeavors to explore how the perceived fit between virtual endorsers and products shapes advertising effectiveness and further assess whether this collaborative model can deliver sustained positive outcomes for brands.

2.3. Source Credibility

Source credibility refers to how the positive characteristics of the information transmitter impact the information receiver’s acceptance level of that information [14]. According to Ohanian [14], attractiveness, expertise, and trustworthiness are the primary drivers of an individual’s perception of source credibility. Here, attractiveness refers to the ability of the endorser to attract the target audience through their appearance and personality traits; the level of expertise reflects the endorser’s professionalism, experience, or skills in the product/brand-related field; and trustworthiness refers to people’s perception of the endorser’s honesty, reliability, and integrity [14].

This model has been widely used in communication and marketing research, especially in celebrity endorsement and the recently emerging field of influencer marketing research. It is worth mentioning that the impact of endorser–product fit on source credibility has also received scholarly attention. For example, Koernig and Boyd [30] pointed out that when the fit between the endorser and the product is higher, the perceived expertise and trustworthiness, as well as the level of advertising effect, are higher, and the negative impact is relatively reduced. Similarly, in their research, Mishra et al. [31] also emphasized the impact of the fit between brand personality and celebrity endorser personality on the credibility of the endorser and advertising dimensions.

A good match can also bring benefits to celebrity endorsements, as it helps increase the celebrity endorser’s perceived attractiveness, expertise, and trustworthiness. In the live-streaming e-commerce context, research has shown that the celebrity–product fit has a more profound impact on perceived trustworthiness than on perceived attractiveness [32]. Empirical studies involving human endorsers consistently found that a higher endorser–product fit can enhance endorsement effects. However, there is scant research thoroughly examining its effects in the context of virtual endorsers. Based on this, we propose the following hypotheses:

H1.1. 

The virtual endorser–product fit positively impacts the attractiveness of the endorser.

H2.1. 

The virtual endorser–product fit positively impacts the expertise of the endorser.

H3.1. 

The virtual endorser–product fit positively impacts the trustworthiness of the endorser.

Further investigation reveals that communicators with higher source credibility can positively influence consumer attitudes and behaviors [17,18], making those with such characteristics ideal brand communicators [33]. Against the backdrop of digital transformation, virtual endorsers, with their visually realistic attributes and anthropomorphic features demonstrated in interactions with followers, are widely employed in advertising and social media domains to captivate consumers. They not only shape their identity through social media but also bestow significance upon products [7]. Intriguingly, their value might even be defined by the products they endorse [6].

Despite the extensive attention virtual endorsers have garnered, there is still a paucity of academic research on the source credibility of virtual endorsers. For example, although a study by Yang et al. [12] focused on the impact of the affinity, expertise, and relevance of virtual endorsers on their trustworthiness, it did not explicitly include the structure of attractiveness. In social psychology research, attractiveness is a foundational concept for constructing social relationships [34], and it holds similarly pivotal importance in marketing. Sources possessing attractiveness tend to wield greater persuasive power, thereby influencing the subjective attitudes of information recipients [35].

In addition, while the studies by Torres et al. [36] and Kim and Park [37] explored the attractiveness of virtual endorsers, they overlooked the impact of expertise and trustworthiness. In the realm of marketing, credibility consistently plays a paramount role. The credibility of human endorsers can also influence consumer attitudes towards brands and products [38]. Thus, it can be postulated that in an online milieu, the credibility attributes of highly anthropomorphized virtual endorsers will influence consumers’ attitudes towards products. Given this context and integrating the model of endorser–product fit with source credibility, this research seeks to deeply examine the perceived fit of virtual endorser–product, attractiveness, expertise, and trustworthiness in relation to the attitude towards endorsed products. Consequently, we posit the following hypotheses:

H1.2. 

The attractiveness of the virtual endorser has a positive effect on product attitude.

H2.2. 

The expertise of the virtual endorser has a positive effect on product attitude.

H3.2. 

The trustworthiness of the virtual endorser has a positive effect on product attitude.

H4. 

The virtual endorser–product fit has a positive effect on the advertising product attitude.

2.4. Source Credibility as a Mediator of the Relationship between Endorser–Product Fit and Product Attitude

In the study of the relationship between endorsers and advertising effects, the credibility features of endorsers have been confirmed as key influencing factors [14,39]. This research utilizes Ohanian’s [14] source credibility model to explore the mediation mechanism of the impact of the virtual endorser–product fit on product attitude. It is inferred that if consumers perceive low source credibility from the endorser, the endorser will lose their ability to influence consumers’ attitude towards the product (or the effect will not occur). For example, when Stephen Curry endorses basketball shoes, due to his athlete status, his endorsement represents a high degree of endorser–product fit, which might lead to a positive product attitude from consumers. Conversely, if he endorses sugary snacks, the mismatch with the athlete image could lead to less positive attitudes towards the product.

Regarding the mediating effect of source credibility features, only a few studies have conducted investigations. While some studies have supported the direct effect [19,39], others have not [27]. Notably, research by Siemens et al. [40] found that the influence of product–celebrity fit on consumers’ advertising attitude is mediated through the overall credibility. Another study indicated that this relationship is only partially mediated [19]. Moreover, Schouten et al. [17] compared the promotional effectiveness of models, actors, and social media influencers, and found that the type of endorser has no effect on expertise or trustworthiness with endorser–product fit, but endorser–product fit indirectly affects expertise and trustworthiness.

When examining the effects of perceived source credibility by consumers regarding virtual endorsers, less research has been conducted. The study by Yang et al. [12] showed that the expertise and trustworthiness of virtual influencers as CSR messengers played a mediating role between the type of endorser and CSR involvement, while attractiveness did not show this effect. This may be because the information recognized by consumers is more about CSR than the product itself. Another study by Jain et al. [13] found that in the green sustainability movement, the trustworthiness of virtual endorsers as environmental communicators played a mediating role in influencing recycling attitudes. These two studies are very important for our research because they both considered the mediating effect of the credibility dimension of virtual endorsers, which is our primary concern. However, we believe that the unique attractiveness of visual authenticity and the expertise ability based on AI are unique characteristics of virtual endorsers; therefore, the dimensions of attractiveness and expertise cannot be ignored.

Based on this, we conducted a comprehensive discussion of whether the dimensions of attractiveness, expertise, and trustworthiness of virtual endorsers will play a mediating role in combination with the matching hypothesis. Should source credibility function as a partial mediator, it could also serve as an intermediary in mediating the impact of endorser–product fit on product attitudes. This aligns with the observation made by Zhao et al. [41], where many studies investigating mediation effects often determine that the mediator’s influence is not solely partial but coexists with a direct effect. Hence, the following hypothesis is proposed:

H5. 

Perceived source credibility features of attractiveness, expertise, and trustworthiness play a mediating role in connecting endorser–product fit with product attitude.

2.5. Moderating Effect of Product Type

Companies use virtual endorsers to promote various products. However, the persuasive effect of virtual endorsers’ advertisements might vary according to different product types. Therefore, we speculate that product type might have a significant impact on the persuasive effect of virtual endorsers.

We categorize products as utilitarian or hedonic goods based on their consumption purposes [42]. Utilitarian goods are products for which performance and functionality play a key role in product selection, offering consumers tangible value [8]. In contrast, hedonic goods are products that can bring consumers pleasure, satisfaction, and other emotional responses [42]. Utilitarian goods often include products where functionality and performance are critical, such as laptops, mobile phones, and stationery. Products meeting consumers’ sensory and emotional needs, like travel, aromatherapy, ice cream, and beverages, are considered hedonic goods [43]. However, it should be noted that the way a product is promoted might influence consumers’ perceptions of its type [44]. For example, if marketers emphasize the durability and size range of a pair of jeans, consumers may perceive it as a utilitarian product. However, if the design of the jeans is highlighted, consumers might view it as a hedonic product. Therefore, a product might possess both utilitarian and hedonic attributes.

Voss et al. [45] argue that products typically encompass both hedonic and utilitarian aspects. Based on this view, we can distinguish products in the dimensions of hedonism and utilitarianism. For hedonic products, consumers pay more attention to the product’s external symbolic attributes; thus, they are more inclined to choose hedonic products that match these attributes [46]. Some research posits that for products offering hedonic benefits (like perfume or massages), people may adopt a transformative buying pattern: that is, seeking attractive appearances or exciting sensations [43]. Consumers’ valuation of hedonic goods depends on how well the image attractiveness conveyed by the salesperson resonates with them [47]. Therefore, when the endorser in the advertisement fits the hedonic product’s image attractiveness, consumers’ attitudes towards the product will be more positive.

In contrast, when consumers choose utilitarian products, they are often cognitively driven [48], paying more attention to product functionality. Past studies have also shown that consumers’ perceptions of the abilities of advertising models significantly influence product sales [49]. The professional abilities of advertising models and salespeople can be transferred to the product, making consumers feel that the product’s functions and performance are superior [8,50]. Therefore, we predict that consumers’ perceptions of a virtual endorser’s expertise and trustworthiness will impact the persuasive effect of utilitarian product advertisements. Accordingly, we establish the following hypotheses:

H6a. 

For hedonic products, the attractiveness of a virtual endorser has a greater impact on product attitude than utilitarian products.

H6b. 

For utilitarian products, the expertise and trustworthiness of a virtual endorser have a greater impact on product attitude than hedonic products.

Based on the above theories and discussions, we propose a model for virtual endorser research, as shown in Figure 1.

3. Materials and Methods

3.1. Data Collection Procedure

To ensure the representativeness of the experimental stimuli, this study systematically screened video/image advertisements related to virtual endorsers available on the internet. Given the similar cultural background and product attribute characteristics, we selected the renowned virtual influencer, Imma, as the subject of our research. There were two primary reasons for choosing Imma. Firstly, she is active on multiple social platforms, with over a million followers, and has significant influence; in addition, her cultural background matches that of our survey participants. Secondly, she collaborates with numerous brands, endorsing a wide range of products from niche goods to luxury items. Additionally, she is one of the few virtual endorsers online who can represent both hedonic and utilitarian products.

This study used purposive sampling to collect data in South Korea. South Korea has a high social media penetration rate, with 82% of the population using social media platforms [51], and virtual endorsers mainly operate on social media platforms. Therefore, Korean users were considered suitable subjects for this study. The survey primarily involved participants from the technology-reliant Millennials (born 1981–1995) and Generation Zero (born after the mid-1990s), who live in an era of digitalization and globalization, and possess a unique affinity for and engagement with technology, entertainment, and social media [52]. This approach aligns with the principles of purposive sampling, as we aim for the sample to represent the user group most likely to interact with virtual influencers. The survey began by asking respondents if they were familiar with virtual endorsers, and only continued if they responded that they were.

To test the effectiveness of product type on the persuasive power of virtual spokespersons, the study, based on [8,43], selected utilitarian and hedonic products as experimental stimuli for separate tests. Two versions of the survey, featuring identical questions, were administered. Respondents were randomly assigned to either the hedonic or utilitarian product survey link and only needed to answer questions about one type of product. This way, we could compare the results, which is a method widely used in between-group design [53]. Specifically, we provided two survey links. One link was for a hedonic product (an advertisement video for a certain brand of soda (https://www.youtube.com/watch?v=qX4s6A9GggA, accessed on 4 August 2023)), and the other was for a utilitarian product (an advertisement video for a certain brand of mobile phone (https://www.youtube.com/watch?v=RQyWfi5dFsQ, accessed on 4 August 2023)). Respondents were required to watch the video link first and then answer the corresponding questions. The survey ran from 4 March 2023 to 24 March 2023, and was conducted online. Among all the collected questionnaires, we excluded 38 for insincere and incomplete responses, resulting in 376 valid questionnaires for subsequent data analysis.

3.2. Measures

The measurement indicators of this study are based on previous research. To ensure greater alignment with the content of this research, meticulous modifications and optimizations were made to the original questionnaire, further enhancing the feasibility and practical applicability of the questions. A five-point Likert scale, ranging from strongly disagree (1) to strongly agree (5), was used. Specifically, the scales for measuring attractiveness, expertise and trustworthiness referred to Ohanian’s [14] research. The scale for endorser–product fit was adapted from Till and Busler’s [16] research. The scale for measuring product attitude was derived from Schouten et al.’s [17] research. The details of these items are shown in

Table 1. The questionnaire with all of the final items.

Constructs	Items
Attractiveness	ATT1: The endorser has a strong attractiveness.
	ATT2: The endorser has a very beautiful face.
	ATT3: The endorser is very lively.
Expertise	EXP1: The endorser has expertise in her field.
	EXP2: The endorser has product experience.
	EXP3: The endorser has extensive product knowledge.
Trustworthiness	TRU1: The endorser is an honest person
	TRU2: The endorser is trustworthy.
	TRU3: The endorser is a reliable source of information.
Endorser Product Fit	EPF1: The characteristic of the endorser is consistent with the attributes of the product that she promotes and sells.
	EPF2: The product attributes that the endorser promotes and sells are highly appropriate for her.
	EPF3: The pairing of the endorser with the product is natural.
Product Attitude	PA1: I think the products or services recommended by the endorser are good.
	PA2: I have a positive impression of the products or services recommended by the endorser.
	PA3: I like the products or services recommended by the endorser.
	PA4: I have a positive attitude towards the products or services recommended by the endorser.

.

4. Data Analysis and Results

4.1. Descriptive Analysis

We compiled relevant information about the respondents, including their gender, age, education level, occupation, income status, etc., in order to gain a more comprehensive understanding of the research sample and its characteristics. Among the respondents, 186 were male (accounting for 49.5%) and 190 were female (accounting for 50.5%). The majority of survey participants were in the age range of 20 to 39 years old (59.9% of the sample) and belonged to the Millennial generation. Approximately 40.7% of the respondents held a bachelor’s degree. The sample shows that 33.8% of the respondents were company employees, while a noteworthy 83.8% reported a monthly income (disposable income) of less than KRW 3 million. This shows that the consumer group interested in virtual influencers is mainly composed of a younger generation with lower income, as shown in

Table 2. Demographic characteristics.

Variables	Category	Total N = 376
		Frequency	Percentage (%)
Gender	Male	186	49.5%
	Female	190	50.5%
Age	Below 20	42	11.2%
	20–29	83	22.2%
	30–39	142	37.7%
	40–49	65	17.3%
	Above 50	44	11.6%
Education Level	Middle school or below	26	6.9%
	High school	69	18.4%
	Undergraduate or bachelor	153	40.7%
	Postgraduate or above	88	23.4%
	Other	40	10.6%
Occupation	Student	67	17.5%
	Company Employee	127	33.8%
	Govt. Employee	61	16.2%
	Freelancer	75	19.9%
	Self-employed	32	8.5%
	other	14	3.7%
Monthly Income (KRW)	Below 500,000	86	22.9%
	500,001–1,000,000	74	19.7%
	1,000,001–2,000,000	75	19.9%
	2,000,001–3,000,000	80	21.3%
	3,000,001–4,000,000	37	9.8%
	Above 4,000,000	24	6.4%

.

4.2. Data Analysis

The proposed model was estimated and evaluated using Structural Equation Modeling (SEM), with the analysis being conducted in two stages: firstly, an analysis of the measurement model, and then an inspection of the structural model [54]. Only structures with acceptable factor loadings, composite reliability, convergent validity, and discriminant validity were used for the structural model. Data analysis was conducted using SPSS 24.0 and Smart-PLS 4.0 software.

Table 3. Construct reliability and validity.

Constructs	Loadings	Cronbach’s Alpha	CR	AVE
Attractiveness	0.865	0.766	0.864	0.680
	0.812
	0.795
Expertise	0.815	0.787	0.876	0.702
	0.855
	0.842
Trustworthiness	0.823	0.784	0.874	0.689
	0.858
	0.826
Endorser Product Fit	0.815	0.797	0.881	0.712
	0.866
	0.849
Product Attitude	0.783	0.791	0.864	0.614
	0.804
	0.764
	0.784

shows that in the measurement model, factor loadings exceeded the acceptable threshold of 0.70, showing a sufficient level of reliability [55]. The structures’ alpha CR exceeded the recommended 0.70 threshold [54], which supports the reliable internal consistency of the structures. All research structures’ AVE values exceeded the recommended 0.50 threshold [56]. The results affirm the reliability and validity of the data gathered in this investigation.

To assess the discriminant validity of the measurement model, we utilized the Heterotrait–Monotrait (HTMT) criteria proposed by Fornell and Larcker [56] and Henseler et al. [57]. According to the criteria laid out by Fornell and Larcker [56], as shown in

Table 4. Discriminant validity (Fornell–Larcker criterion).

	ATT	EXP	TRU	EPF	PA
ATT	0.824
EXP	0.519	0.838
TRU	0.434	0.433	0.836
EPF	0.583	0.573	0.505	0.844
PA	0.600	0.588	0.511	0.635	0.784

, the results demonstrate that the square root of Average Variance Extracted (AVE) surpassed the correlation values between the constructs examined in this study and those in previous research. The HTMT criteria presented by Henseler et al. [57], as shown in

Table 5. Discriminant validity (Heterotrait–Monotrait ratios).

	ATT	EXP	TRU	EPF	PA
ATT
EXP	0.661
TRU	0.546	0.548
EPF	0.737	0.723	0.635
PA	0.762	0.745	0.647	0.800

, indicate that discriminant validity was successfully established among the research constructs, as all HTMT values were below the threshold of 0.85. Consequently, this study met the requirements for demonstrating discriminant validity.

Following the validation of the measurement model’s reliability and validity, we proceeded to examine the structural model in order to substantiate the proposed hypotheses. First, to confirm multicollinearity, a Variance Inflation Factor (VIF) greater than five implies potential multicollinearity issues among the dimensions [58]. The VIF values of this study ranged from 1 to 1.850, suggesting no multicollinearity existed among the research dimensions. Furthermore, R2 and Q2 were estimated to predict the model. R2 precisely predicts the variance explained by the constructs, with a higher R2 (ranging from 0 to 1) indicating better predictive accuracy [59]. The R2 for attractiveness in this study was 0.340, the expertise was 0.328, the trustworthiness was 0.255, and the product attitude was 0.544, demonstrating that the structural model had a strong explanatory power.

Stone–Geisser’s Q2 was used to validate the model’s prediction. A Q2 ≥ 0 indicates that the model has predictive relevance, and the larger the Q2, the stronger the predictive relevance [59]. The results show a Q2 value of 0.223 for attractiveness, 0.228 for expertise, 0.172 for trustworthiness, and 0.327 for product attitude, indicating predictive relevance between endorser–product fit, source credibility, and product attitude.

Finally, we evaluated the goodness of fit of the model [60]. SRMR and NFI are commonly used indicators for PLS-SEM, with an SRMR below 0.08 suggesting a good model fit [61]. An NFI above 0.9 indicates a good model fit [60]. The SRMR value of this study’s model evaluation was 0.062. Although the NFI value of 0.805 is less than 0.9, the difference is not significant. DULS and DG are both below the 99% percentile of bootstrap differences (Hi99), indicating a very good fit of the measurement model [61]. Therefore, the model in this study is overall reasonably fit (see

Table 6. Overall measurement model fit.

	R2	Adjusted R2	Q2	Model Fit	Value (SM)	HI99
Attractiveness	0.340	0.338	0.223	SRMR	0.062	0.088
Expertise	0.328	0.326	0.228	DULS	0.531	0.606
Trustworthiness	0.255	0.253	0.172	DG	0.224	0.298
Product Attitude	0.544	0.539	0.327	NIF	0.805

).

We followed the direct relations and conditional mediation procedures recommended by early researchers [54]. The evaluation of the structural model adopted the bootstrapping method to determine the size of the path coefficients. All path coefficient results are presented in Figure 2 and

Table 7. Hypotheses testing results.

Hypothesis	Path	Path Coefficient	T Statistics	Result
H1.1	Endorser Product Fit → Attractiveness	0.583	12.990 ***	Supported
H2.1	Endorser Product Fit → Expertise	0.573	11.809 ***	Supported
H3.1	Endorser Product Fit → Trustworthiness	0.505	10.203 ***	Supported
H1.2	Attractiveness → Product Attitude	0.249	5.104 ***	Supported
H2.2	Expertise → Product Attitude	0.229	4.443 ***	Supported
H3.2	Trustworthiness → Product Attitude	0.165	3.833 ***	Supported
H4	Endorser Product Fit → Product Attitude	0.276	4.833 ***	Supported

Notes: *** p < 0.001.

.

The results show that the endorser–product fit had a positive and significant impact on attractiveness (β = 0.583, t = 12.990, p < 0.001), expertise (β = 0.573, t = 11.809, p < 0.001), and trustworthiness (β = 0.505, t = 10.203, p < 0.001). Moreover, attractiveness (β = 0.249, t = 5.104, p < 0.001), expertise (β = 0.229, t = 4.443, p < 0.001), and trustworthiness (β = 0.165, t = 3.883, p < 0.001) had a positive and significant impact on product attitude. The endorser–product fit exerted a positive direct effect on product attitude (β = 0.276, t = 4.833, p < 0.001). Therefore, hypotheses H1.1, H2.1, H3.1, H1.2, H2.2, H3.2, and H4 were supported.

4.3. Mediating Effect of Source Credibility

The determination of a mediating effect relies chiefly on the statistical significance of the path coefficients originating from the independent variable, extending to both the mediating variable and the dependent variable. Additionally, it is essential that the path coefficients stemming from the mediating variable to the dependent variable also exhibit statistical significance. In essence, the independent variable exerts an indirect influence on the dependent variable through the intermediary action of the mediating variable. When this indirect effect surpasses the direct effect from the independent variable to the dependent variable, the existence of a mediating effect can be confirmed [62].

In this study, we employed bootstrapping analysis to confirm the mediating effects of source credibility—attractiveness, expertise, and trustworthiness. Bootstrapping techniques were initially introduced by Preacher and Hayes [63] to test the indirect effects more accurately among variables. When determining the confidence intervals for indirect relationships, obtaining precise results is crucial. If the bootstrapped confidence intervals do not include zero, the existence of a mediating effect can be confirmed [63]. Furthermore, this study incorporated the recommendations put forth by Nitzl et al. [64], employing the variance accounted for (VAF) as a measure to assess the magnitude of the mediating effect. As delineated by Nitzl et al. [64], a VAF value below 0.2 signifies the absence of mediation, a VAF value falling within the range of 0.2 to 0.8 suggests partial mediation, and a VAF value exceeding 0.8 points indicates complete mediation.

The results show that attractiveness, expertise, and trustworthiness all partially mediate the relationship between the endorser–product fit and product attitude (VAF = 0.345, CI = [0.085, 0.210]; VAF = 0.322, CI = [0.071, 0.200]; and VAF = 0.232, CI = [0.390, 0.100], respectively). Hence, H5 is accepted. Furthermore, among the three sub-dimensions of credibility, the perceived attractiveness of the endorser has the greatest impact on product attitude. The trustworthiness of the endorser has a relatively lower influence on consumer product attitude. The verification results of the mediating effects are presented in

Table 8. Moderated indirect effect.

Indirect Effect of Moderator					Confidence Interval
Paths	β	STDEV	T Statistics	VAF	2.50%	97.50%
Endorser Product Fit → Attractiveness →Product Attitude	0.145 ***	0.032	4.552	0.345	0.085	0.210
Endorser Product Fit → Expertise → Product Attitude	0.131 ***	0.033	3.951	0.322	0.071	0.200
Endorser Product Fit → Trustworthiness → Product Attitude	0.083 ***	0.018	3.492	0.232	0.039	0.100

Notes: *** p < 0.001.

.

4.4. Moderating Effect of Product Type

To test the impact of product type (hedonic vs. utilitarian) on the effect of virtual influencer endorsements, the dataset was divided into two subsets based on product type and Partial Least Squares (PLS) analysis was performed. Figure 3 shows the results of the PLS analysis for the split samples. By statistically comparing the path coefficients of the impact of endorser credibility (attractiveness, expertise, and trustworthiness) on product attitude in the structural model for hedonic products with the corresponding path coefficients in the structural model for utilitarian products, the existence of interaction effects was evaluated. The statistical comparison was conducted using the following steps [65]:

$$t=\frac{Pat{h}_{sample1}-Pat{h}_{sample2}}{\sqrt{\frac{\left(m-1\right)}{(m+n-2)}\times S{E}_{sample1}^2+\frac{\left(n-1\right)}{\left(m+n-2\right)}\times S{E}_{sample2}^2}\times \sqrt{\frac{1}{m}+\frac{1}{n}}}$$

where $Pat{h}_{sample i}$ is the path coefficient for subset i, m and n are the sample sizes of the product type datasets; $S{E}_{sample i}$ is the standard error of the path in the structural model for product type i; and t is the t-statistic with m + n − 2 degrees of freedom. If the T-value is +1.96 (p = 0.05), then the difference in path coefficients is significant at a 5% level.

The results show that in the structural model for hedonic products, the path coefficient from attractiveness to product attitude is significantly stronger than in the structural model for utilitarian products (T = 30.512). This indicates that the attractiveness of the virtual endorser plays a more crucial role in explaining variance in product attitude when endorsing hedonic products.

In the structural model for utilitarian products, the path coefficients from expertise and trustworthiness to product attitude are significantly stronger than in the structural model for hedonic products (T = −22.960; T = −5.096) (see

Table 9. Hypothesis testing for moderation of product types for sub-sample of two tests.

Path	Path Coefficients (T Statistics)		Path Coefficients Diff (T Statistics)	Result
	Hedonic (N = 186)	Utilitarian (N = 190)
Attractiveness → Product Attitude	0.446 (6.505)	0.230 (3.348)	30.512	Hedonic > Utilitarian
Expertise → Product Attitude	0.215 (2.836)	0.381 (5.953)	−22.960	Hedonic < Utilitarian
Trustworthiness →Product Attitude	0.230 (4.669)	0.263 (3.571)	−5.096	Hedonic < Utilitarian

). This suggests that expertise and trustworthiness play a more crucial role in explaining variance in product attitude when the virtual endorser is endorsing utilitarian products. Thus, H6 is supported.

5. Discussion

Virtual endorsers, as an innovative endorsement approach, have demonstrated their immense potential in achieving brand product promotion and digital sustainable growth. This paper, grounded in the match-up hypothesis and source credibility theory, proposes an assessment model to measure consumer attitudes towards products promoted by virtual endorsers. It also explores the interaction between source credibility and product type, and how these factors collectively influence consumers’ attitudes towards products.

First, the congruence between the endorser and the product is of paramount importance for the long-term evolution of a brand. When there is alignment between the endorser and product characteristics, brand ethos, and the firm’s digital transformation objectives, it can amplify consumer perceptions of the endorser’s attractiveness, expertise, and trustworthiness, thereby enhancing their attitude towards the product. This finding aligns with the match-up hypothesis, which posits that fit between an endorser and an endorsed product enhances the endorsement’s effectiveness [18,39]. While these studies typically focused on human endorsers, our research extends this theory to virtual endorsers. This not only offers a fresh perspective for businesses to integrate digital technology into their business models but also promotes the synergy of sustainable development with digital transformation.

In addition, all three credibility features of the virtual endorser—attractiveness, expertise, and trustworthiness—have a positive impact on product attitudes. This finding can be understood in terms of how individuals are typically more easily attracted to, and more likely to pay attention to and remember, information provided by attractive and engaging individuals [66]. This study further confirms the positive impact of a virtual endorser’s attractiveness on consumer product attitudes [37]. In addition, consumers’ perceptions of the expertise and trustworthiness of a virtual endorser also positively influence their attitudes towards the product, which is consistent with previous studies [12,17,18]. When consumers perceive a virtual endorser as knowledgeable, competent, and trustworthy, they exhibit more positive attitudes towards the products endorsed by such a virtual endorser. As an emerging digital marketing tool, virtual endorsers not only help businesses balance environmental, social, and commercial value creation in the digital era, but also enable more environmentally friendly and efficient interactions with consumers. This innovative marketing strategy can reduce reliance on traditional resources, thereby supporting sustainable development goals.

Second, this study reveals that endorser–product fit has a positive impact on product attitudes through the source credibility model. This finding is consistent with research in the domain of human endorsers, underscoring the importance of matching endorsers with products [33,67]. Furthermore, our research indicates that the credibility model plays a partially mediating role in the relationship between the match-up hypothesis and product attitudes. This suggests that when the attributes of a virtual endorser highly align with the endorsed product’s attributes, consumers’ trust and fondness for the endorser can translate into favorability for the product [66]. As such, the attractiveness, expertise, and trustworthiness of the endorser are crucial in persuading consumers to develop favorable attitudes toward the product. Through these findings, this research furnishes a novel theoretical perspective in the field of digital branding, affirming the significance of virtual endorsers in the ‘match-up hypothesis → source credibility model → consumer product attitude’ continuum. This provides invaluable guidance for enterprises in their pursuit of sustainable growth in the digital age.

It’s noteworthy that, compared to human endorsers, consumers’ perceptions of credibility towards virtual endorsers exhibit certain disparities. Research on human endorsers suggests that an endorser’s expertise and trustworthiness have a more pronounced effect than the allure stemming from the endorser–product fit [32,39,67]. This study indicates that, among the three major sub-dimensions of credibility determining consumers’ attitudes towards products, the perceived attractiveness of the virtual endorser plays the most significant role, while its trustworthiness exerts a relatively weaker influence on product attitude. This can be partially attributed to CG technologies, which can greatly enhance the visual appeal of virtual endorsers. However, given that virtual endorsers do not genuinely exist and are predominantly oriented towards commercial objectives, consumers often harbor doubts regarding the authenticity of the products they endorse [6], which directly impacts consumer trust. These findings have significant implications for brand promotion strategies, suggesting that when employing virtual endorsers, businesses should deeply consider their application methods, ensuring a high degree of fit between the virtual endorsers, the brand, and its products. This is not only crucial for enhancing brand image and strengthening consumer purchase intentions but is also vital for achieving a company’s digital sustainable development.

Third, we confirmed the moderating role of product type in the effectiveness of virtual endorsers, i.e., consumers demonstrate significant differences in attitudes towards hedonic and utilitarian products promoted by virtual endorsers. When virtual endorsers promote hedonic products, consumers are more easily swayed by the endorsers’ allure. Conversely, for utilitarian products, consumers place greater emphasis on the endorsers’ expertise and trustworthiness. This difference reflects the distinct decision-making psychology of consumers towards hedonic and utilitarian products, with hedonic products relying more on the overall ambiance of the advertisement, whereas utilitarian products emphasize professional and credible advice [6,50]. This research result responds to the call by Thomas and Fowler [20] for a deeper examination of the endorsement effects of virtual endorsers on different product categories. This provides valuable references for brands to formulate more effective advertising strategies using virtual endorsers.

6. Conclusions

This study aims to explore the effective mechanisms of virtual endorsers. We referred to the match-up hypothesis and source credibility model to investigate the impact of endorser–product fit and credibility features (attractiveness, expertise, and trustworthiness) on product attitudes, as well as the interaction between credibility features and product type. The research findings are as follows: Firstly, the virtual endorser–product fit positively impacts the credibility features, which in turn enhances consumers’ positive attitudes towards the advertised products. Secondly, the source credibility model serves as a partial mediator in the connection between endorser–product fit and product attitude. Among the influence relationships of the three sub-dimensions of credibility, attractiveness has the most significant impact on product attitude, while the impact of trustworthiness is relatively small. Thirdly, consumer attitudes toward products vary depending on the product type. This study provides a systematic perspective for a more comprehensive understanding of the role of virtual endorsers in product promotion within digital channels. Digital technology affects all aspects of human life, and the findings of this research offer valuable insights for businesses in the advertising and marketing field to utilize digital technologies in achieving sustainable business models.

6.1. Theoretical Implications

Firstly, this study proposes a new comprehensive framework by integrating the match-up hypothesis and source credibility model and applies it to virtual endorsers in the field of digital marketing. Considering that digital platforms have become the primary channel for communication between consumers and businesses [68], this theoretical innovation not only extends the application of digital technology to the advertising and marketing domain, adapting to the rapid development of technological changes, but also offers a new theoretical perspective for businesses to embed digital technology into their business models. These virtual endorsers not only emphasize the importance of digital tools in modern marketing but also highlight the potential of these tools in promoting sustainable development for businesses.

Secondly, this study emphasizes the significant role of credibility characteristics conveyed by virtual endorsers in shaping consumer attitudes towards products. This finding is of substantial importance, as it reveals a marked distinction between virtual endorsers and traditional human endorsement strategies. Traditional research has placed greater emphasis on the expertise [69] and trustworthiness [32] of endorsers as opposed to their attractiveness. However, in the current digital context, due to the unique nature and influence of virtual endorsers, attractiveness may prevail in certain scenarios. This is because the guiding role of visual information in digital channels for product identification and perception should not be overlooked [70]. The visual stimuli provided by virtual endorsers are particularly crucial in influencing consumer evaluations. This insight offers a new theoretical perspective, significantly guiding the use of virtual endorsers in the advertising industry’s digital transformation. It underscores the necessity of leveraging digital technology to adjust marketing strategies, redesign, and optimize business processes for creating sustainable business models. Through these theoretical contributions, this study not only provides direction for the digital business transformation of the advertising industry but also offers profound insights into understanding and leveraging the role of virtual endorsers in digital marketing.

Thirdly, this study further reveals the types of products suitable for endorsement by virtual endorsers, uncovering the moderating effects of hedonic and utilitarian products in this context. Although previous research has explored the role of AI recommenders or virtual agents [49,50] in consumer choices, the involvement of virtual digital characters in the endorsement arena, in particular, remains a relatively new topic. Prior studies have largely focused on technology, cosmetics, or luxury goods [6,71], with a relative lack of discussion on how to balance hedonic and utilitarian products to influence attitudes towards products endorsed by virtual endorsers. This finding provides a new theoretical perspective for understanding the integration and development between product attributes and digital technology, thereby expanding the discussion about the sources of virtual endorsers’ design. It also explores the attitudes towards the application of virtual endorsers in product design from a consumer perspective. This not only greatly enriches the existing literature but also provides important theoretical references for businesses on how to leverage digital advantages to integrate business processes across multiple levels to create sustainable value.

6.2. Managerial Implications

With the extensive application of AI and CGI in marketing, modern enterprises are rapidly employing these advanced technologies to enhance customer experiences and ensure sustainable competitiveness in the wave of digitalization. In this context, our research provides insightful implications for businesses that utilize virtual endorsers.

Firstly, it is essential to recognize that accurately assessing and enhancing the match-up between virtual endorsers and products is a key factor influencing endorsement effectiveness. This strategy not only impacts the effectiveness of endorsements but also directly relates to the development of enterprises. With the enhancement of consumers’ online skills, awareness, and engagement [72], empowering them can foster innovation, enhance competition, and boost productivity. This not only promotes sales growth and market share on an economic level but also strengthens the connection between brands and consumers on a social level. Companies should utilize digital channels and virtual worlds to create bridges between virtual and real-life experiences, customizing personalized marketing strategies based on customer needs and feedback [4] (pp. 167–168). By leveraging the high adaptability of virtual endorsers, necessary adjustments can be made to ensure that their image and brand story align with product characteristics and consumer expectations. According to Henninger et al. [4] (pp. 168–170), virtual endorsers can also act as opinion leaders, disseminating messages about sustainable development on social platforms and influencing their followers. This provides consumers with opportunities for direct participation and interaction, making virtual endorsers effective tools for promoting social responsibility and positive values, thereby helping companies build a positive social image.

Secondly, when discussing the application of virtual endorsers in advertising, the key lies in understanding how to enhance the critical attributes of these digital personas to increase consumer attraction and engagement. As highly attractive virtual endorsers can draw consumer attention through their human-like features and audiovisual characteristics [25], this direct visual experience significantly affects consumers’ evaluation of products. The visual and interactive features of digital media not only facilitate product recognition and perception but also, by integrating background design and virtual endorser imagery, significantly enhance the visual presentation and sensory experience of the product. Additionally, empathetic responses evoked by visual stimuli can enhance tactile experiences [4] (pp. 149–150). Through virtual interaction technology, a more enriched and immersive experience can be provided, further increasing consumer attractiveness and engagement. At the same time, utilizing digital media channels and customized virtual endorsers can effectively convey a brand’s sustainable development strategies and value propositions.

On the other hand, the trustworthiness of virtual endorsers remains a crucial element of their success. Their active presence and enjoyable interactions on social media are particularly important for attracting younger users. Brands can utilize these digital endorsers to convey information related to their sustainable activities, enhancing emotional connections with consumers through innovative storytelling and interactive events [4] (pp. 153–156). They can provide personalized services to meet user needs anytime, anywhere, while also increasing consumer awareness of sustainable development and offering high-quality support. This depth of imagined social relationship can significantly enhance consumer trust [73]. Digital technology offers unique opportunities for co-creating value. By aligning technological investments with strategic objectives, virtual endorsers can become pivotal tools for disseminating information on sustainable development and educating consumers about sustainability through social media narrative strategies, as highlighted by Henninger et al. [4] (p. 156). This role is instrumental in aiding enterprises to adopt more sustainable production and consumption patterns, thereby contributing to the broader goal of sustainability in business practices.

Lastly, this study reveals the role of product types for virtual endorsers, proffering invaluable insights for brands’ strategic decisions in the digital era. Brands need to adjust their use of virtual endorsers according to the characteristics of different types of products. For products aimed at providing hedonic experiences, brands should focus on utilizing the attractiveness and charm of virtual endorsers to enhance consumer experience. Such strategies not only attract the target market but also build the brand image socially and strengthen emotional ties between users and the brand. For products with a stronger focus on functionality or practicality, the endorser’s expertise and trustworthiness become paramount. Advertising strategies need to concentrate on demonstrating the functionality and benefits of the product, thereby enhancing consumer trust in the product’s performance. This strategy of employing digital platforms to integrate actual product attributes with digital technology represents a novel pathway to sustainability. By shifting promotional activities to digital realms, it substantially reduces the resource wastage typically associated with physical promotions, thereby mitigating environmental impact. The results of this study can assist stakeholders and practitioners in understanding how to use digital tools to create sustainable value in the digitization process, aligning their business with other businesses, industries, and external environments to achieve more effective digital business strategies in various contexts.

6.3. Limitations and Future Research

Firstly, the applicability of this study might be limited as it only examined the virtual endorser Imma, and restricted hedonic products to soda drinks and utilitarian products to electronic devices like mobile phones. Therefore, research on other virtual endorsers and product types could enhance our understanding.

Secondly, while this study explored the mediating role of the source credibility model and the moderating role of product type, other important influencing factors might need to be considered in a complex market environment. For instance, consumers’ familiarity with a product or brand and consumers’ consumption habits may be an important factor, as consumers usually have a more positive attitude towards a product or brand with which they are more familiar [74]. If participants have low familiarity with the product, they might rely more on the endorser, namely the attractiveness of the virtual endorser, while watching the advertisement, which could affect the differential effect of product types. Therefore, further investigation into this is necessary.

Thirdly, this study examines the impact of credibility features as a whole. Future research could delve deeper into comparing consumers’ perceptions of various aspects of credibility related to virtual endorsers. Furthermore, in light of the existing literature, conducting a comparative study of the credibility of virtual endorsers and real endorsers is deemed necessary. Such an investigation would furnish valuable insights for businesses in their decision-making processes related to marketing.

Additionally, future research could explore other personalized features of virtual endorsers, such as gender, age, or ethnicity. The present study only used a female virtual endorser, and thus did not fully consider the impact of gender on endorsement effects. The gender of the model could influence the persuasiveness of the advertisement [75]; hence, subsequent research could consider the interaction between gender and product types.