Abstract
Virtual influencers (VIs) are a novel, increasingly popular and successful marketing tool in the digital marketing landscape, specifically in the domain of travel destination marketing. However, how social networking site (SNS) users respond to VI marketing in the travel sector remains underexplored. To address this gap, we systematically identified, empirically tested, and compared nine theories to understand the drivers of social networking site users’ visit intentions toward VI-promoted destinations. Using an online survey, we collected responses from 419 active SNS users and analyzed our data using partial least squares structural equation modeling, followed by a qualitative study (n = 18). Social power theory and parasocial interaction theory exhibited the highest explanatory power. Our research contributes to travel literature through deepening our understanding of the mechanisms by which VIs can influence visit intention, which is important for travel professionals’ understanding of VIs’ use in attracting a demographic that immerses themselves in social media.
Keywords
Introduction
In 2022, the Korean Tourism Organization introduced its virtual influencer (VI), Lizzie Yeo (@lizzie.days), to promote travel to South Korea (Tan, 2022). Kyra (@kyraonig), the first VI in India with almost a quarter of a million followers, was introduced in January 2022, and “she” showcases historical and contemporary travel destination places in India on her Instagram account (Jain, 2022). From these examples, it is clear that VI-based marketing, as a subset of the already very large influencer-based market, is an important element of how travel destinations market themselves online. There is growing evidence that increased investment is pouring into influencer marketing, and more recent trends show marketers have started using virtual agents and artificial intelligence, leading to the development of this new type of influencer, the VI (da Silva Oliveira & Chimenti, 2021; Xie-Carson, Benckendorff, & Hughes, 2023). VIs are computer-generated images of fictitious characters, which, with the help of AI, are embedded through graphics technology on social media platforms (Lou et al., 2023; Mrad et al., 2022). Thomas and Fowler (2021) defined a VI as “a digitally created artificial human associated with Internet fame and using software and algorithms to perform tasks like humans” (p. 12). Employment of VIs offers travel organizations the opportunity to create the “perfect” influencer for their target markets (Meng et al., 2025; Xie-Carson, Magor et al., 2023). Despite the increasing role of VIs in marketing travel destinations, there is very scant evidence from research on how VIs exert influence on potential travelers.
Our study addresses significant gaps in the literature. First, Seçilmiş et al. (2022) argued that there remains a lack of understanding on how behavioral intention, in a travel content, is driven by social media, and suggest that more theory-based models are required. Second, a range of different theories have been used to explain VI marketing and there is a need to consider which of theory is more pertinent to explain the impact of VIs on travel behavior. Third, human emotional responses to VI marketing are overlooked, particularly in travel literature. The Pleasure Arousal Dominance (PAD) model (Mehrabian & Russell, 1974) is well known and accepted and although Zhang et al. (2024) reported the effect of the arousal dimension on behavioral responses to a virtual endorser in a streaming context, the impact of the pleasure and dominance dimensions remains under-explored in the scholarship. To address the above-mentioned gaps, using a cross-sectional survey technique followed by qualitative interviews, we gathered empirical evidence about social networking site (SNS) users’ behavioral intention toward VI-promoted places and analyzed this data using a range of theories.
While there is emerging VI literature in tourism and travel settings, these studies have focused on issues such as SNS users’ engagement with VIs (Xie-Carson et al., 2024; Xie-Carson & Benckendorff, 2024), destination evaluation (F. Li & Zhou, 2023a), the role of VIs on travelers’ risk and trust perception (Ameen et al., 2024) and matching influencer type with destination type (Meng et al., 2025). Despite this scholarship, academic literature still lacks a comprehensive understanding of the key VI-related factors that influence travel behaviors, such as visit intention. In synthesizing and adding to this debate, our study advances knowledge about VIs from a behavioral perspective by adopting a multiple-model comparison method based on the method developed by Venkatesh et al. (2003).
Our study makes several significant contributions to existing scholarship. First, our study broadens the body of knowledge on influencer marketing in the travel industry by comparing nine theories to identify the key factors associated with VIs that contribute to SNS users’ visit intention. By doing this, we also respond to calls made by Ameen et al. (2024) through showcasing various significant factors of VI marketing toward consumer reactions (i.e., visit intention) based on quantitative and qualitative evidence. Second, according to our understanding, we are one of the very few studies that have utilized the consumer acceptance of technology (CAT) model (Kulviwat et al., 2007) in the tourism-based VI marketing literature to analyze travel behavior. This model concentrates on the integration of both affective (PAD) and cognitive (usefulness and ease of use) responses of humans to new stimuli (Nasco et al., 2008). Very few studies (cf. Cheng & Huang, 2022; S. Li et al., 2018; C. Wang et al., 2019) have applied the affective dimension (PAD) of human emotions to understand travel behaviors. We consider VIs to be a new technological stimulus and probe SNS users’ affective and cognitive emotional responses to destinations promoted by VIs. Specifically, our study sheds light on this scholarship by providing empirical evidence of the appropriateness of the CAT model in the VI travel context. Third, although the theory of reasoned action (TRA; Fishbein & Ajzen, 1975) and theory of planned behavior (TPB; Ajzen, 1991) have been extensively applied in the management and marketing literature, due to their high predictive ability, to study individual’s behavioral intentions, influencer marketing-based studies have rarely employed these two theories to investigate travel decision-making behavior, such as visit intention. Lastly, past partial least squares structural equation modeling (PLS-SEM) based model comparison studies (Hasan et al., 2019, 2020; Shiau & Chau, 2016), only reported the significant predictors of their outcome variable of interest. By applying Necessary Condition Analysis (NCA) along with PLS-SEM, we not only find the significant factors but also uncover the necessary conditions to achieve the desired level of SNS users’ visit intention toward VI-promoted destinations, constituting a methodological contribution to the model comparison approach.
Social Media Influencer Marketing in Travel Tourism
The integration of social networking sites (SNSs) as a marketing tool has become a prominent feature of the travel and tourism industry strategy (Liu et al., 2023) to promote their products and services to their target consumers and trigger travel. Social Media Influencers (SMIs), who are content creators, are considered online celebrities, and with an engaged online fan base, they seem to have substantial power to generate followers’ buying behaviors (Kapoor et al., 2022). In tourism and destination marketing campaigns, endorsing and partnering with SMIs have become powerful marketing tools for influencing travelers’ behaviors (X. Xu & Pratt, 2018). From the consumer perspective, travel-based SMIs increasingly curate their travel-related information on social media in an enjoyable and easy-to-consume manner, that potential travelers would find difficult to source elsewhere without significant research. This phenomenon has led to a growing body of research, and Table 1 lists a sample of very recent studies of SMI-related research in the travel industry.
Brief Overview of Recent Studies on Influencer Marketing in the Tourism and Travel Context.
VI marketing has been increasingly embraced in the tourism and travel sectors. In terms of similarities between VIs and human influencers, both influencers can develop their own follower networks by generating captivating digital content on SNSs while showcasing their interactive abilities with SNS users (Franke et al., 2023). However, there are some distinctions. Sensory ability has been found to be a differential factor between VIs and human influencers in terms of endorsement, indicating that VIs are perceived to have lower sensory capability than human counterparts (H. Li et al., 2023). In contrasting VI versus human influencer marketing efficacy in destination endorsement, Meng et al. (2025) found that consumers responded more favorably to VI marketing for endorsing cultural destinations (i.e., historical infrastructure), while human influencer endorsements were more effective for natural destinations (i.e., natural landscapes). Virtual endorsers led to more positive destination evaluations for destinations predominantly viewed as competent, whereas for destinations primarily stereotyped as warm, the endorsement effect was more favorable when humans endorsed them (F. Li & Zhou, 2023a). To the best of our knowledge, few studies (cf. Ameen et al., 2024; M. Choi et al., 2023; F. Li & Zhou, 2023a; Meng et al., 2025; Xie-Carson et al., 2024; Xie-Carson, Magor et al., 2023) have explored the influence of VI marketing in the tourism setting with minimal focus on visit intention. To address this gap, this study aims to identify the impact of VI marketing-related factors on users’ visit intention by applying a multiple-model comparison approach.
Virtual Influencers: What Do We Know?
As an emerging research category in the influencer marketing literature, academic papers on VIs became more numerous at the beginning of 2020. Early studies described the significant attributes of VIs and their impact on consumer behavior. Among those VIs’ attributes, anthropomorphism and attractiveness have received substantial attention. da Silva Oliveira and Chimenti (2021) identified five categories: attractiveness, authenticity, controllability, scalability, and anthropomorphism, as significant VI attributes. Moustakas et al.’s (2020) findings suggest that the humanization of the VI characters and their appealing storytelling ability can be considered a success factor for brands adopting VIs as a marketing strategy. Following an exploratory research approach, Mrad et al. (2022) investigated the relationship between VIs and their followers’ perceptions of them. Their results reveal that the anthropomorphic human-like attributes of VIs mainly shape the connection between VIs and their followers. Ahn et al. (2022) examined consumers’ attitudes toward one famous VI, Lil Miquela, who sponsored posts on social media and found that her anthropomorphic attributes and perceived level of attractiveness contributed to improving consumers’ positive response toward endorsed brands.
H. Kim and Park (2023) indicated the level of a VI’s attractiveness did not directly affect SNS users’ intent to purchase VI-endorsed products. However, users’ brand attachment and mimetic desire were found to mediate the positive relationship between VI’s attractiveness and purchase intention. Alboqami (2023) applied qualitative comparative analysis to understand the factors driving online consumers’ trust in VIs. His findings imply that the configuration of several factors, namely attractiveness, credibility dimensions, and congruence between the VI, the consumer, and the product, may lead to forming a higher trust level in VIs. When compared to human influencers, VIs were found to be less effective in endorsing a brand because they were not perceived to be credible (Ozdemir et al., 2023) though they received higher levels of engagement (H. Li et al., 2023) and generated higher levels of novelty (Franke et al., 2023). Deng and Jiang (2023) analyzed the level of discontentment and anxiety felt by SNS users based on images posted by human influencers versus VIs. Their results highlight that the images of both human influencers and VIs reflected a higher level of appearance anxiety among the treatment group compared to the control group, who were shown scenic images.
A growing number of studies have noted the differential effectiveness of VIs versus human influencers in understanding consumer behavior. For example, consumers tend to identify more with humans than VIs, and while VIs’ recommendations are seen to be more useful with regard to utilitarian products, human influencers’ advice appears to be more useful for hedonic products (Belanche et al., 2024). Regarding social media engagement, VIs received more engagement (i.e., likes, comments) than humans from consumers, but consumers showed less positive emotions toward VIs. In terms of brand endorsement, brand attitude and purchase intention were lower for VI endorsement than human counterparts (H. Li et al., 2023). For sensory experience, both VIs and human influencers were believed to possess similar distal sensory capabilities; however, VIs were perceived to have lower proximal sensory capabilities than humans, resulting in lower purchase intention among consumers when VIs endorsed proximal sensory products (X. Zhou et al., 2024). Table 2 outlines recent studies on VI marketing.
Selected Studies on VIs.
Theories Underpinning Behavioral Intention Toward Influencers
There are a variety of theories in the extant literature that have been used to underpin research that considers VI from a behavioral intention perspective. In deliberating on the theories to include in our paper, we used Venkatesh et al.’s (2003) method to assess and compare different behavioral intention-based theories. We focused on those behavior-focused theories pertinent to explaining the VI phenomenon from consumer behavior viewpoint. In past model-comparison studies, researchers tended to select theories based on their own judgment (Davis et al., 1989; Huh et al., 2009), though Hasan et al. (2019) and Moody et al. (2018) do offer some guidance. Thus, we used the following four indicators to choose our competing models.
Scopus and Google Scholar citations: we considered models with a higher number of citations in both Google Scholar and Scopus databases. We acknowledge the number of citations does not necessarily reflect a research model’s quality. Still, a model with a higher number of citations indicates its significance and perhaps its efficacy.
Building on Human Influencer studies: Given the relatively paucity of pure VI studies, we reviewed key theories from the human SMI domain for inclusion as they may have relevance for the VI context.
Pertinency to the VI Perspective. VIs are a relatively new and emerging trend in the influencer marketing domain; therefore, we referred to those research models potentially suitable for our study.
Relevance to the Travel Context: Given that this study is in a travel context, we also reviewed literature on behavior-based models in this domain.
Based on this we considered nine theories for comparison because past scholarship suggests these chosen models illustrate individual behavioral aspects suitable to our research aim. Table 3 provides our justification for the inclusion of each theory with the four headings and a commentary. While the nine theories were selected in early 2023, we have included selected recent references to show the current relevance of these theories for VI scholarship.
Indicators for Model Selection.
C. Wang et al. (2019) and Lehto et al. (2008) considered the “pleasure-arousal-dominance (PAD)” theory to study travel intention and “PAD” has been included in the CAT model as an affective component.
Behavioral intention-based research models have remained one of the dominant research streams in the SMI marketing literature. Magrizos et al. (2021) employed TPB to explain consumers’ purchase behaviors toward influencers’ personally branded products. To understand people’s motivation to follow an influencer on social media, the various forms of user gratification needs (Lee et al., 2022) are a popular approach. Source credibility theory (SCT) has remained one of the dominant theoretical lenses for studying consumer behavior toward SMI marketing (Vrontis et al., 2021). Additionally, in line with social power theory, SMIs exhibit different forms of influence over their followers to strengthen desired behavioral outcomes (Cheung, Leung, Aw, & Koay, 2022). In influencer marketing, congruence is crucial because it develops and fosters a higher level of association between the influencer and the target audience (D. Y. Kim & Kim, 2021). The CAT model can also explain users’ behavioral responses to VIs because its dimensions have been found to be critical in comprehending consumer reactions toward new technological objects (Nasco et al., 2008). Parasocial interaction theory (PSI) has also been extensively applied to understand how individuals establish pseudo-intimate interactions with media celebrities, such as television celebrities (Stern et al., 2007) or SMIs (Tolbert & Drogos, 2019). The theory of anthropomorphism (ANTHRO) has also been found relevant in exploring consumer behaviors (Moriuchi, 2021). Based on the extant research, we have considered nine theories: TRA, TPB, uses and gratifications theory, SCT, social power theory, self-congruence theory, CAT, PSI, and ANTHRO. The following paragraphs discuss each of these theories in relation to the context of this study.
Behavioral intention, the extent of the effort an individual intends to exercise toward engaging in a specific behavior, is considered a significant determinant of an individual’s actual behavior, and the higher the level of intention a person holds toward performing the behavior, the greater the likelihood the specific behavior will be followed. The theory of reasoned action (TRA; Fishbein & Ajzen, 1975) is one of the most applied theories, and it postulates that attitudes and subjective norms determine individual intent toward an action or object (Ajzen & Fishbein, 1980). The later theory of planned behavior (TPB) extended TRA to include perceived behavioral control (Ajzen, 1991). TPB has been regarded as one of the prominent theories in explaining human behavior across different research contexts. For instance, grounded in TPB, Woosnam et al. (2022) showed U.S. travelers’ attitudes, subjective norms, and behavioral control were significant predictors in explaining their willingness to travel to endangered places.
Uses and gratifications theory (UGT) is considered one of the prominent sociological approaches that elucidates an individual’s underlying motivations and processes in deliberately choosing a particular media platform to satisfy their needs (Katz et al., 1973; Katz & Foulkes, 1962). UGT postulates a person’s media behavior is goal-oriented (i.e., they select a media channel to satisfy their gratification needs; Zadeh et al., 2023). According to UGT, whenever audiences use a particular media to fulfill their gratification needs, they tend to form a positive behavioral attitude toward the specific media channel, resulting in continuous usage of that media channel (C. Xu et al., 2012). With the emergence of the Internet and SNSs, researchers have used UGT to investigate the determinants of consumer behavior in online channels. In more recent times, UGT has been employed to explain cutting-edge digital phenomena, such as virtual tourism (Geng et al., 2024) and VIs (Lou et al., 2023).
Past research studies have underscored the importance of the source credibility theory (SCT) in diagnosing individuals’ behavioral reactions to information sources (Jang et al., 2021; Lou & Yuan, 2019). Hovland and Weiss (1951) first applied the notion of source credibility as a theoretical lens based on the premise that individuals tend to be more persuaded and triggered to action when they perceive the information source as more credible. Credibility generally measures any given information source’s degree of believability from the receiver’s frame of mind (Yılmazdoğan et al., 2021). Marketing and communication research streams have widely used SCT to examine various phenomena. Notably, in the marketing literature, source credibility is applied to study and measure the potential of celebrity endorsement (Ayeh, 2015; Lim et al., 2017) and to evaluate travel intention among Instagram users in the travel influencer marketing domain (Yılmazdoğan et al., 2021).
According to Raven (2008), social power reflects an individual’s or a group’s capacity to involve other individuals or groups in actions that lead to change in other individuals’ thought processes and behaviors. Social Power Theory (SPT) posits that a person or a group of persons with social power can influence other persons’ behavioral and psychological perspectives (Raven et al., 1998). French and Raven (1959) identified five important bases of social power (expert, referent, reward, coercive, and legitimate) that can stimulate individuals’ future behavior. SMIs, with their respective expertise and brand endorsements, use social power to trigger consumer behavior aspects on social media platforms (P. Wang et al., 2020). P. Wang and Huang (2022) have explored the impact of four dimensions of social power (expert power, informational power, referent power, and reciprocity power) to capture online consumer engagement and buying behavior on social media. Moreover, Mehraliyev et al. (2021) employed SPT to examine the psychological interaction dynamics between the expert power of influencers and U.S. tourist individuals in the light of online travel review platforms.
The CAT model (Kulviwat et al., 2007) has been identified as one of the pivotal models for explaining consumers’ behavioral reactions toward a new technology (Kulviwat et al., 2014). The CAT model consists of cognitive elements (perceived usefulness, perceived ease of use, and relative advantages) and affective components (pleasure, arousal, and dominance) to account for individual behavioral responses in accepting and adopting new technology and technological products (Nasco et al., 2008). Past literature suggested that integrating cognitive and affective factors is crucial in discerning consumers’ multifaceted behavioral reactions to technology, which results in consumers either accepting or rejecting it (Kulviwat et al., 2007, 2014). Nasco et al. (2008) and Kulviwat et al. (2007) reported that both the cognitive and affective aspects of the CAT model are highly relevant to consumer adoption behaviors, and the inclusion of these constructs in their study notably strengthened the model’s explanatory power. García-Milon et al. (2021) suggested two affective components, pleasure and arousal, experienced from using a smartphone were significant determinants of tourists’ likeliness to use their smartphone to make purchases in the tourism context. Also, Purwandari et al. (2022) highlighted the significance of the three affective components in explaining influencers’ travel recommendations.
The theory of self-congruence (CONG) has commonly been applied to illustrate the impact of a consumer’s self-image and concept on their behavioral activities (Sirgy, 1985). Self-congruence theory posits that consumers tend to favor products or brands that generally harmonize with their self-image (Sirgy, 1982). This theory is an extension of self-concept theory (Sirgy, 1982) in which an individual exhibits two forms of concepts: the actual self and the ideal self (Zhu et al., 2019). The actual self represents how consumers perceive themselves (me as I am), and the ideal self represents how consumers want to perceive themselves (the perfect me; S. M. Choi & Rifon, 2012; X. Xu & Pratt, 2018). The condition of self-congruence greatly influences a person’s self-concept: a person’s actual and ideal self-concept mixes and matches with the image of a brand or a personality (Malär et al., 2011). In the SMI context, self-congruence reflects the perceived congruity and similarity between a consumer’s self-image and an influencer’s image (Zogaj et al., 2021) and has been found to influence people’s behaviors (X. Xu & Pratt, 2018). In the travel context, visitor image congruity emerged as an important predictor of travel intention for both prospective and repeat visitors (Maghrifani et al., 2022).
The concept of anthropomorphism (ANTHRO) refers to the assignment of the characteristics of human beings to non-human agents, namely a product, a good, a pet, or a brand (Epley et al., 2007). Anthropomorphism usually facilitates human connections and engagements with anthropomorphized non-human entities by bringing on interactivity between human and non-human agents and triggering individuals’ decision-making processes. Several studies have been conducted to understand how individuals respond to anthropomorphized objects. For instance, Dabiran et al. (2024) and J. Yang et al. (2023) found VI’s anthropomorphic attribute as a critical element in driving consumer behavior toward VI endorsement.
Parasocial interaction theory (PSI) reflects the unidirectional media-generated communications between users and media celebrity personalities initially introduced by Horton and Richard Wohl (1956). They described this as a one-sided, non–face-to-face, human-to-human relationship lacking reciprocity, where the person who believes in that relationship develops and carries away this interaction. This concept is interpreted as a paradoxical and delusory one-to-one human connection in which no actual relationship exists between the parties (C. Whang & Im, 2021). Currently, research focusing on the role of parasocial interaction has pivoted from traditional media personas to SMIs (Aw & Chuah, 2021; Labrecque, 2014). PSI has been detected between VIs and SNS users (Block & Lovegrove, 2021).
Methods
Empirical Studies on Model Comparison
Model comparison models are developed when there are a number of competing empirically validated models to explain a phenomenon in the interest of finding the best of these models to illuminate a research problem (Hsiao & Tang, 2014). Table 4 provides details of extant research that uses this approach.
Summary of Past Model Comparison Studies.
TRA = theory of reasoned action; TPB = theory of planned behavior; TAM = technology acceptance model; DTPB = decomposed theory of planned behavior; C-TAM-TPB = combined model of TAM and TPB; MM = motivational model; EDT = expectation–disconfirmation theory; MPCU = model of personal computer utilization; IDT = innovation diffusion theory; SCT = social cognitive theory; SQ = service quality theory; SE = self-efficacy theory; DOI = diffusion of innovations; VAM = value-based adoption model; CAT = consumer acceptance of technology model; CBOP = contextualized BOP (bottom-of the-pyramid) model.
Davis et al. (1989) empirically compared TRA and the technology acceptance model (TAM) to explain university students’ behavioral intentions toward computer technology at the organizational level. Chau and Hu (2001) compared three well-established behavioral intention-based models to predict healthcare professionals’ intention to use telemedicine technology. Venkatesh et al. (2003) assessed eight prominent theoretical models of users’ behavioral adoption to determine users’ intention to adopt information technology in organizational settings. Using five well-known intention-based models, Hsiao and Tang (2014) undertook a model comparison method to study college students’ behavioral aspirations to accept electronic textbooks. Shiau and Chau (2016) conducted multiple-model comparison studies to understand Taiwanese students’ behavioral intent to accept cloud computing classrooms by comparing six behavioral research theories. Giovanis et al. (2019) empirically tested four theoretical models to predict potential users’ adoption behavior of mobile banking services in Greece. Hasan et al. (2019) undertook a comparison study by considering seven adoption intention-based behavioral models to understand low-income consumers’ behavioral adoption of mobile banking technology. In the VI marketing context, we believe the model comparison approach will help better understand current scholarship on behavioral intention and the determinants of users’ behavioral aspects toward VIs.
Questionnaire Development
In this study, we developed an online-based structured questionnaire consisting of each selected theoretical model’s constructs, all of which are reflective in nature. We adapted each construct’s reliable and valid measurement items from relevant prior studies and then slightly modified the items in accordance with our VI research context. The list of the adapted measurement items under each construct and their sources is presented in Appendix 1.
To check the consistency and clarity of all measurement items under each construct, we first approached several frequent SNS users and academic scholars in the digital marketing discipline to pretest our questionnaire. On the basis of their feedback, the survey questionnaire was modified to enhance its understandability and transparency. To ensure the further quality of the questionnaire and verify the reliability and validity of the measures, we conducted a pilot survey with the relevant individuals (n = 31) who are active users of SNSs. The result of the pilot test confirmed the measures are valid and reliable, reflecting the survey questionnaire was understandable and interpretable from the respondents’ perspectives.
All SNS users may not be acquainted with VIs; therefore, we first described what VIs are by providing a definition, vignettes, and links to VIs’ social media accounts (e.g., Lil Miquela). Then, if they had never heard of VIs, we asked them to spend some time on these VIs’ accounts. Afterward, respondents answered demographic questions, such as their gender, age, education level, employment, and nationality, and we inquired about their social media usage frequency and which social media platforms they use. Survey respondents were informed their responses would remain anonymous and confidential. We then asked the participants to respond to the constructs of the selected models. Responses were scored on a 7-point Likert scale, ranging from strongly disagree (1) to strongly agree (7). We included two attention-checking questions across the survey questionnaire to evaluate the participants’ engagement with the study.
Participants and Survey Administration
This research study focused on comprehending SNS users’ (our target population) behavioral intent toward VIs. The ethics committee of the lead researcher’s university granted ethical approval for the study. We used an online marketplace called Prolific to recruit the relevant survey participants. It has also been affirmed that data collection via online marketplaces is considered more credible and reliable than data collected on traditional platforms (Legendre et al., 2020). Moreover, Prolific was deliberately chosen for this study because it offers researchers prompt and effortless access to the desired study populations, maintaining a strict measure for ensuring data quality (Kapoor et al., 2022).
The participants were recruited during March 2023 and were selected based on three prescreening criteria: (1) at least 18 years of age, (2) active user of social media, and (3) adequate proficiency in the English language. A total of 437 participants who met these criteria participated in this online survey. Because SNS users have easy access to online websites, this research directed the participants to a Qualtrics web-based questionnaire platform to collect user responses and to confirm their participation agreement on the informed consent form. Based on the attention-checking questions and incomplete completion of the survey questionnaire, of the total respondents, 419 responses were found suitable for the data analysis. The survey participants on Prolific Academic were allowed 60 min for survey completion with a remuneration payment of GBP 6.00 (USD 7.50).
Prior to collecting data for our study, we determined the minimum sample size requirement by performing the a priori power technique (f2 value = 0.15, power value = 0.90, and probability α error = .05) using G*Power software (Faul et al., 2007). The software yielded a minimum recommended sample size of 166; our final sample size (n = 419) significantly exceeded this requirement. Afterward, we also ran a post hoc analysis on G*Power to confirm the power of our sample n = 419, generating a power level of 0.99 (>0.90 of priori power) and establishing the adequate statistical power of our sample size.
Out of 419 responses in our study, 261 respondents reported themselves as male (62.3%), and 153 were reported as female (36.5%). Four respondents selected non-binary gender (1.0%), and one respondent (0.2%) did not reveal a gender identity. Regarding the age distribution, 32% belonged to the 18 to 24 years-old group, with 39.6% reporting an age between 25 and 34, with the remainder (119) in the 35 to 74 years old age category. In terms of social media usage, 282 respondents (67.3%) reported they use social media several times a day, 109 use (26%) daily, and only 16 use (3.8%) several times a week. Out of the 419 respondents, only 43 reported themselves as followers of a VI.
Common Method Bias
The potential for common method bias (CMB) issue was minimized by adopting the recommended procedural approach by Podsakoff et al. (2012). First, a plain and straightforward English language tone was followed to draft the entire questionnaire so respondents would smoothly grasp the questionnaire. Careful writing and arrangement of the survey questionnaire was prioritized, and pretesting was conducted to exclude any vague or unknown terms or words. At the very beginning of the questionnaire, a consent form was attached, in which the study’s research purpose was described in detail. In this consent form, we further confirmed that the study would be conducted as only an academic research project, not for commercial purposes; no personal information about respondents would be retained; and the survey questionnaire had no right or wrong responses. Furthermore, in the survey questionnaire flow, the predictor and outcome variables were placed in separate sections to reduce the probability that the survey respondents could gauge the potential causal-effect connection between the variables.
We also employed statistical methods to address potential CMB in our study. First, we performed a collinearity test among the latent constructs of each selected model separately and found the variance inflation factor (VIF) values of all constructs under each model were below the suggested cut-off value of 3.30 (Kock, 2015). Also, we followed Lindell and Whitney’s (2001) recommended statistical procedure by considering a theoretically unrelated marker variable. In our case, we measured attitude toward drinking coffee. Then, we ran a correlation test among the marker variable (MV) and all other latent variables under each model individually; this revealed no correlation value between MV and any other variables in any model was greater than 0.30. Finally, in each model, we compared the value of R2 before and after the inclusion of the MV on our endogenous variable (i.e., visit intention), and no significant difference was observed in the R2 value of the endogenous variable for each model from partialing out our MV. Therefore, our statistical procedures also confirmed that CMB is not likely to be an issue in this study.
Reliability and Validity
The reliability and validity of each construct under each theoretical model were assessed by applying SMART PLS version 4.0, and all measures are in Appendix 2 and the models tested in Appendix 3. Cronbach’s alpha (α) coefficient and composite reliability (CR) values were employed to measure the internal consistency and reliability of the constructs under each model. All CR values of the constructs were found above the prescribed value of 0.70, indicating the constructs’ reliability (Bagozzi & Yi, 2012). Additionally, all Cronbach’s alpha (α) coefficient values for all constructs were observed to be above the suggested cutoff value of .70. Thus, the constructs under each chosen theoretical model were evidenced as reliable and internally consistent. Convergent validity was exhibited by diagnosing whether the items under each construct loaded onto their respective constructs. All average variance extracted (AVE) scores in the study were greater than the threshold score of 0.50, implying convergent validity is established (Hair et al., 2010). Following Fornell and Larcker’s (1981) approach, we also calculated the discriminant validity test of all constructs by taking into consideration the square root value of the AVE of each construct and compared whether that value is greater than the correlation between other constructs (Lowry & Gaskin, 2014). The results indicate the square root value of the AVE of each construct exceeds the respective intercorrelated constructs, confirming that discriminant validity exists between the constructs. We also adopted a heterotrait-monotrait (HTMT) ratio to manifest the discriminant validity further and found that no values of the HTMT ratio exceeded the threshold value of 0.90 (Henseler et al., 2015).
Data Analysis: Empirical Comparison of the Models
Our research aim has been grounded in understanding VI-related factors that affect SNS users’ intention to visit. Thus, in accordance with the research aim, we adopted the PLS-SEM technique to perform statistical analysis; this technique is deemed appropriate when the research aim has been developed on predictive models (Shmueli et al., 2016). Henseler et al. (2012) found utilizing the PLS-SEM technique is beneficial when comparing and evaluating theoretical models. Prior model comparison studies have also applied the PLS-SEM method (Hasan et al., 2019; Shiau & Chau, 2016). In this study, we compared nine theoretical models to discover the determinants of SNS users’ visit intention in the context of VI-related factors by considering the adjusted R2 value (variance explanation), model’s predictive power (PLSPredict and CVPAT) and the theoretical understanding of path coefficients. In addition, we also applied NCA to understand the necessary logic of the PLS-SEM findings.
Structural Model: Explained Variance (Adjusted R2)
Table 5 demonstrates that the nine theoretical models’ explained variance value (using adjusted R2) ranged from 38.10% to 59.80% variation in explaining SNS users’ travel intention toward VI-promoted destinations. The strongest adjusted R2 value is found by employing SPT (59.80%), closely followed by PSI (59.50%).
Model Comparison.
p < .01. *p < .05.
Conversely, ANTHRO was found to have the lowest adjusted R2 value of 38.10%, and CONG theory exhibits the second lowest adjusted R2 value of 44.00%, which is followed by SCT with adjusted R2 value of 46.80%. Moreover, with adjusted R2 value of 55.70%, the CAT theory presents relatively higher explanatory power compared to TRA (55.50%) and TPB (55.40%) though the differences are minor. We also considered two indicators for model fit indices: standardized root mean square residual (SRMR) and normed fit index (NFI). For SRMR, a value of 0.05 or less and for NFI, a value of 0.90 or more indicates a good model fit (Dash & Paul, 2021). Based on the SRMR and NFI values (Table 5), except for UGT, all other models are a good fit.
Assessment of Model Predictive Capabilities
The study has also considered the PLSPredict (Shmueli et al., 2019) and CVPAT (cross-validated predictive ability test; Sharma et al., 2023) techniques to determine the model’s predictive power. According to PLSPredict analysis, if the Q2 predict value of the target construct’s (i.e., visit intention) indicators is positive and the difference between PLS-RMSE (root mean squared error) and linear model (LM)-RMSE is negative for all indicators, the model would have strong power. For CVPAT analysis, if the PLS model produces a significant and lower average loss than that of IA (indicator average) and LM prediction benchmark, the model exhibits strong predictive capacity. Table 6 shows that both SCT and CAT models possess strong predictive power and prediction capabilities.
Model’s Predictive Power Assessment.
p < .01. *p < .05.
Note. By considering adjusted R2, PLS Predict, and CVPAT measures from Tables 5 and 6, it can be concluded that social power theory (highest adjusted R2 value, strong predictive power, and some form of predictive capability) and consumer acceptance of technology (CAT) theory (moderately high adjusted R2 value, strong predictive power, and strong predictive capability) are preferable to other models. RMSE = the root mean squared error; IA = indicator averages; LM = linear model.
Necessary Condition Analysis (NCA)
Along with the PLS-SEM technique, this study also used the NCA approach to determine the necessary variables (within the nine theories) to attain the outcome variable (visit intention) in the given data set (Dul, 2016). This approach is regarded as complementary to the PLS-SEM results (Richter et al., 2020). We identified the necessary factors associated with VIs under each theory through NCA to achieve visit intention based on the effect size (d) >0.10 and p < .05 criteria. The NCA results (Table 7) reveal that, except for the attractiveness of VIs in SCT, all the other factors under each theory are necessary conditions for visit intention.
NCA Results.
Insights From Qualitative Study
To corroborate and complement our quantitative findings and enhance our understanding of SNS users’ behavioral responses to VI-promoted destinations, a supplementary qualitative study using semi-structured interviews has been conducted. This additional qualitative study helps us to obtain new insights about VIs apart from the quantitative findings (Creswell, 2014). The main aim of this qualitative study was to capture SNS users’ motivating factors for visiting places promoted by VIs. The author team recruited relevant interview participants from their professional social media networks, such as LinkedIn, which was followed by adopting the snowball sampling strategy to secure more potential interviewees. This snowball strategy is deemed a suitable approach when access to the target population is not effortless and straightforward in nature (Hair et al., 2007). Respondents were recruited based on the following indicators: they had to be fluent in English and frequent daily users of major SNSs such as Instagram, Facebook, TikTok and Snapchat, and they had to be active followers of human influencers on SNSs with a minimum age of 18 years old. Afterward, 18 interviewees took part in this qualitative phase during June and July 2024. The majority of the participants were male (n = 14), while four of them were female, and all interviewees were between 19 and 40 years old. Interview participation was entirely voluntary, and participants were made aware that they could withdraw their participation at any point in time. The interviews were conducted online and face-to-face in Microsoft Teams and recorded after obtaining explicit consent from the interviewees. Interviews lasted between 19 and 48 min, with an average of approximately 29 min. One week before the interview, participants received emails containing an explanation of what a VI is, the study background, and a few links to the popular VIs’ social media accounts to familiarize themselves with VIs if they were not already followers of VIs.
Interviewee participants were asked about their overall perception of VIs and what would be important in determining their intention to visit VI-promoted places. Our interview guidelines were developed in such a way as to ensure the capture of open-ended responses from the interview participants. This process helped us gather rich and in-depth information about participants’ perceptions of VIs and their endorsements in the travel context. By following the thematic analysis guidelines provided by Braun and Clarke (2006), several key factors were identified to support and extend our quantitative results (please see Table 8).
Key themes from the Qualitative Study.
Discussion
This study sought to ascertain the factors associated with VIs that trigger SNS users’ travel intentions and determine which of the nine theories better explains this relationship. Our results suggest that all the chosen theories provide reasonable explanatory power to account for users’ travel intentions. By exhibiting higher adjusted R2 values than their competing theories, SPT and PSI theories better explain travel intention.
In our study, drawing on SPT, VIs’ informational, reciprocity, and referent power appear to be significant predictors of users’ travel intention. This is in line with P. Wang and Huang (2022) and P. Wang et al. (2020), who reported similar results in relation to online consumer behavior. Once online consumers receive something positive from SMIs, they may form an obligatory feeling to reciprocate (Casaló et al., 2020). Likewise, VIs may establish personal attachments with online users by disseminating relevant destination information, which would lead to users developing a commitment to reciprocate the favor to VIs by considering and visiting those destinations (Wong & Wei, 2023). Also, our qualitative findings assert that the absence of reciprocal interaction between VIs and online users may not be helpful in developing behavioral reactions toward VIs (Table 8). On social media, consumers tend to consider influencers’ accounts and digital content as a source of information on products that help them make their purchase decisions (Djafarova & Rushworth, 2017). Yi et al. (2021) indicated by consuming content information on SNSs, tourism enthusiasts could develop a better understanding of attractive destination places. Furthermore, our interview participants feel VIs can reflect their information power by easily disseminating information about a destination to their audience (Table 8). Therefore, by providing pertinent information and references about the destination places, VIs can enhance users’ overall perception of their promoted tourism places (Xie-Carson, Magor et al., 2023). It is then likely to improve users’ willingness to visit those places. Moreover, visiting tourist places with information from a VI may reduce the level of individuals’ tourism information-seeking efforts.
The TRA and TPB models support that attitude toward VIs and, also, subjective norms are powerful visit-intention predictors. These findings are consistent with past research evidence, which led researchers to argue that consumers’ attitudes toward influencers favorably affect several aspects of individual behavioral response, such as event attendance intent (Sun et al., 2021). In the context of this study, our findings imply that SNS users would intend to visit VI-endorsed places whenever they develop a positive and favorable attitude toward VIs. In addition, subjective norms in relation to VIs have a strong influence on visit intention, demonstrating that others’ opinions about VIs are critical for SNS users to develop a positive propensity to visit VI-promoted places. Our qualitative results (Table 8) also confirm that VI recommendations would influence interviewees if they were discussed within their networks, as subjective norms emerged as an important theme. Interestingly, perceived behavioral control does not affect visit intention in our study. A probable reason for this non-significant effect could be that visiting or checking a VI’s social media content does not require serious technical knowledge; merely having an active social media account allows online users to visit any influencer’s account on SNSs.
In our study, three factors (attractiveness, expertise, and trustworthiness of VIs) under SCT demonstrate significant positive effects on visit intention. Prior literature also found impactful evidence of source credibility on consumer behaviors (Lim et al., 2017; Lou & Yuan, 2019). In addition, an influencer’s level of trustworthiness plays a critical role in affecting consumer behavioral attitudes; research confirmed that a media persona’s trustworthiness is one of the effective means to bolster the level of confidence in the consumer’s mind about the endorsed product (S. S. Kim et al., 2013). Yılmazdoğan et al. (2021) explored the effect of source credibility of Instagram influencers on travel intention by drawing samples from Instagram users in Turkey. They reported that the expertise and trustworthiness dimension of source credibility, both directly and indirectly (via parasocial interaction), affected Instagram users’ travel intentions. VI’s trustworthiness is further detected as a critical factor in our qualitative findings (Table 8), endorsing that interviewees would be influenced when they find VI’s recommendations and shared information about destinations as trustworthy. Seçilmiş et al. (2022) found the expertise and content attractiveness of travel influencers were the critical determinants of travel intention. However, our findings suggest that if a VI recommends visiting a travel destination perceived as attractive, expert, and trustworthy, SNS users will most likely be enticed to explore that place as a destination.
In our study, we used CAT theory, which has pleasure, arousal, and dominance as its affective factors and perceived usefulness and ease of use as its cognitive factors (Kulviwat et al., 2014; Nasco et al., 2008). The findings reveal pleasure, perceived usefulness, and ease of use are significant predictors of visit intention, and that arousal and dominance do not significantly impact intention. CAT theory is rarely applied in SMI marketing studies, although it possesses the considerable potential to study different aspects of consumer behaviors. Purwandari et al. (2022) modeled pleasure, arousal, and dominance as mediating factors to unravel the connection between the influencer-follower experience and behavioral intention to follow the influencer’s travel recommendations. Their results demonstrated the significant effect of pleasure and dominance on recommendation intention via the influencer’s commitment. In the context of travel behavior, Lehto et al. (2008) documented perceived pleasure’s strong impact on visit intention to a beach location, which is consistent with our findings that experiencing pleasure from VIs is a strong motivator for visit intention. Moreover, our findings suggest that the perceived usefulness of visiting a VI’s social media account will help develop positive visit intentions among online users. Belanche et al. (2024) found VI’s usefulness critical to consumers’ behavioral intention toward product recommendations. The same evidence is reflected in our qualitative findings (Table 8), where interviewees indicated that if they identified VI’s content as useful and relevant, they would likely be influenced as the theme of VI’s usefulness appeared as significant. Thus, SNS users may find checking VI’s content helpful in developing a positive perception of the VI-promoted destination, and from that, they will be eager to visit the destination. In contrast, arousal and dominance dimensions appear to be non-significant predictors of visit intention. These results correspond to the findings of Kumar et al. (2021) and Verkijika and De Wet (2019). The possible explanation is that SNS users may not feel an adequate level of arousal and dominance emotion evoked from VIs that could contribute to their visit intention.
Our findings support that SNS users’ actual and ideal self-congruence, with VIs demonstrate a powerful effect on their intention to visit VI-endorsed places. Past literature reported consumers’ self-congruence as significantly impacting consumer behavior (Malär et al., 2011; Zogaj et al., 2021). Likewise, when it comes to visiting destinations, the results indicate if an online user finds a similarity with VIs with reference to a VI’s lifestyle and travel history, online users may be persuaded to travel and visit the same places. When an online user identifies VIs to reflect their ideal self-image, the online user would most likely be prompted to mirror the VI’s travel journey.
We also detected a positive relationship between PSI with VIs and visit intention, which is congruent with past empirical findings. Yılmazdoğan et al. (2021) examined the direct and mediating role of PSI with an influencer on visit intention. Their results reflect that parasocial interaction with an influencer impacts intention and fully mediates the source credibility dimensions’ effect on intention, a finding endorsed by Um (2023). PSI with a destination-based TV documentary program enhanced the travel intention of young Chinese viewers (Bi et al., 2021). VIs are able to develop PSIs with their followers (Block & Lovegrove, 2021); thus, interacting with VIs would evoke a higher level of PSI among their followers, leading to increased intention to visit VI-endorsed places.
In this study, we also measured the direct impact of the anthropomorphism of VIs on visit intention and detected positive statistical evidence of the relationship in alignment with the past findings. Park et al. (2021) strove to evaluate the impact of anthropomorphism in understanding customers’ experiences in the hospitality industry about employing robot services in China. The authors analyzed online customer-generated reviews and comment-related data in relation to robot services and reported that human-like service robots tend to please customers whenever they perform a satisfactory service delivery. A recent study on VIs reported that VIs’ anthropomorphism significantly affects their social and physical attractiveness in the SNS users’ view, which also partially influences users’ attitudes toward VI-endorsed brands (Ahn et al., 2022). Our interview findings (Table 8) also corroborate that the anthropomorphic attribute of VIs would affect the interviewee’s perception of VIs, as the importance of anthropomorphism emerged as a critical factor.
In our study, SNS users’ gratification factors with VIs, namely escapism, trendiness, purposive value, and social interaction, were identified as strong influencers of visit intention, indicating that these elements could induce travel intention. Lou et al. (2023) suggested that social interaction strongly motivates online followers’ behavioral intention toward VIs because they can relate to a VI’s social values or personality. Moreover, social interaction was found to be a critical motivating factor in affecting online consumer engagement behavior with SMIs (Cheung, Leung, Yang, et al., 2022). Past research also argued purposive value is a predictor of users’ behavioral aspects in the online environment. For instance, within the social media environment, consumers’ valuing of co-creation behavior is significantly driven by their purposive values (Zadeh et al., 2023). The findings of Sung et al. (2022) indicate escapism experiences via an augmented reality setting shape consumers’ behavioral intentions. We can argue VIs, as virtual persons, may offer SNS users a channel for experiencing escapism from their daily life and may eventually evoke their willingness to visit the destination VIs promote on their social media accounts.
Theoretical Implications
Notwithstanding the increasing scholarly attention on leveraging VIs to promote tourism and travel destinations, research evidence in this field remains scarce (Xie-Carson et al., 2024; Xie-Carson & Benckendorff, 2024; Xie-Carson, Magor et al., 2023). To address this void in the scholarship, we explore the motivating factors that influence SNS users’ intention to visit VI-promoted places. Our study presents significant theoretical implications for the domain of VI marketing in the travel and tourism field. Firstly, our study provides a comprehensive understanding of why individuals would consider VI marketing efforts in the travel industry by exploring significant predictors of visit intention toward VI-promoted destinations, utilizing the well-understood model comparison approach. We also answer a call from studies which have suggested more research to capture the impact of VI marketing on travel behavior (Ameen et al., 2024; Meng et al., 2025). Therefore, to advance our knowledge of the effect of VI endorsement in the travel industry, we have identified several key factors associated with VI marketing that impact visit intention by considering nine theoretical models.
Secondly, by applying a qualitative interview approach after employing model comparison, we not only revalidate our quantitative results but also identify novel VI factors and offer new insights to better understand visit intention beyond corroborating quantitative findings. For instance, the interviewees pronounced VI’s collaboration with human influencers as a critical element in influencing SNS users’ intention to visit (Table 8). As consumers tend to form their evaluation of influencers based on their collaborative efforts (Thomas et al., 2024), we assert that the collaboration between VIs and human influencers will bring human aspects to the travel endorsement by compensating for VI’s shortcomings and reinforcing more positive consumer behavior. Cultural representation through VIs has also been deemed a significant factor in driving visit intention toward VI-promoted places (Table 8). Our findings also resonate with past literature on influencers. The famous Chinese VI, Ling, has remained distinctive among other VIs as her communications promote traditional Chinese culture to her audience nationally and globally (Luo & Kim, 2023). Hence, the qualitative results contribute to our nuanced understanding of VI marketing in travel literature by outlining VI-related novel factors that can influence travel behavior, which are not accounted for by the existing theories of our model comparison approach.
Lastly, we offer a methodological contribution to the academic literature by applying NCA, PLS-SEM and a qualitative interview-based approach to the travel research paradigm to identify the VI-related critical factors of visit intention. The PLS-SEM, often referred to as a standard multivariate analytical method, helps identify significant determinants of a certain outcome by examining the causal-predictive connections between antecedent and outcome factors (Hauff et al., 2024). Nonetheless, this method does not specify which factors are essential to obtain this outcome. NCA infers that necessary conditions are to be met to achieve a particular outcome; otherwise, the desired outcome will not occur (Dul, 2016, 2024). Importantly, non-significant antecedent factors of the PLS-SEM method can be necessary conditions for achieving that particular outcome factor (Hauff et al., 2024). In Table 9, we outline the implications of our PLS-SEM and NCA analysis. Moreover, the quantitative findings of the PLS-SEM-based model comparison method can be supported and strengthened by applying qualitative methods such as interviews. Therefore, the combined application of PLS-SEM, NCA and qualitative methods puts forward a novel methodological synergy within the context of the model comparison method that has not been previously applied in VI marketing, tourism and travel literature—hence representing the methodological contribution of this study.
Explanations of the combined use of PLS-SEM and NCA Results.
Note. NC = Necessary condition; sig. = significant determinant; non sig. = non-significant determinant.
Managerial Implications
The multiple-model comparison method used in this paper offers some notable implications for travel and tourism brand managers and marketing professionals through leveraging VI marketing. First, the attractiveness and anthropomorphic characteristics of VIs should be taken into account in travel marketing on SNSs. Marketing practitioners might also focus on enhancing the credibility perception of VIs by taking advantage of expertise and physical appearance attributes (Xie-Carson et al., 2024). Travel marketers could adopt the latest technologies to enhance both the glamorousness and humanlike attributes of VIs, aiming to make them more appealing and interesting to the online travel community. At the same time, the overall persona and storyline of a VI must be pertinent to the travel destination marketing context to motivate SNS users’ visiting propensity. Our qualitative findings (Table 8) reveal that cultural representation by VIs has appeared to be an influencing factor in VI recommendations. For example, destination marketing organizations can partner with or develop their own VIs to showcase specific cultural traditions of a travel place with a view of catering to a niche market, creating a relatable narrative of local cultural attractions to connect with potential travelers. Also, travel marketers may focus more on VIs’ clothing choices to showcase contemporary lifestyles, reflecting online audiences’ tastes and resembling the cultural essence of the respective travel destination.
Second, our results noted VI’s knowledge and information-centric aspects, such as expertise and informational power, are significant predictors of visit intention. Therefore, VI’s social media content should emphasize travel destination-oriented information, such as the destination’s history, pricing, transportation arrangements, and tourist-related amenities (Xie-Carson, Magor et al., 2023). Because SNS users nowadays tend to be more detail-focused, we suggest that travel marketers include in their VI posts other destination-based attributes, namely the well-known landmarks, high-quality destination images and videos, famous local cuisines, and additional useful information related to the destination; this may signal the VIs’ expertise to their audiences.
Third, tourism marketing practitioners should increase their focus on matching the congruence between VIs and online users. Practitioners must pay careful attention when selecting VIs as travel destination promoters, making sure the VI’s overall personality, posts, and lifestyle harmonize with the target travelers’ self-concept. To do that, tourism marketing professionals must first attempt to determine their target travelers and identify their actual and ideal self-concept by researching their way of living, values, tastes, and preferences. In addition, VIs’ storylines and social media content should be designed to display narratives resembling the potential travelers’ lifestyles and aspirations. This method of endorsing VIs, which should resonate with target tourists, will elevate target travelers’ visit intention.
Fourth, creators of VIs must display a transparent and explicit operational process of manifesting VIs to the audience to make VIs more believable and trusted. This would enable marketing managers to select those VIs considered most trustworthy and lead target tourists to have faith in VI-promoted travel content. This VI selection process may entail evaluating authenticity and clarity in creating and maintaining VI. As a result, collaborating with trusted VIs will allow tourism marketers to strengthen their marketing campaigns to be more credible, which may increase the likelihood of triggering visit intention.
Fifth, tourism marketers may consider collaboration between VIs and human influencers as a strategic means to strengthen their promotional campaigns. A high realism setting for VIs, such as VIs with human friends in a single photo frame, shapes consumer behavior toward VI marketing (I. Kim et al., 2024). Thus, utilizing human and VI’s distinctive power and capabilities will offer marketers unique opportunities to generate more captivating and trustworthy endorsement content for travel enthusiasts. VIs are able to provide creative and appealing content with advanced computer graphical design, while human influencers can ensure the authenticity and realness aspect of the travel endorsement content to enhance individuals’ travel propensity.
Lastly, our results also report PSI with VIs will evoke visit intention among online users. While partnering with tourism brands, VI creators should reinforce the two-way communication process between online users and VIs to foster PSI by replying to and liking users’ comments and having online conversations with them. Moreover, managers should strategically design VI content to generate more engagement and interactions with and behavioral responses from audiences. By engaging online users in the VI content-creation process and having open question-and-answer sessions with them, tourism marketers may enhance the degree of PSI between VIs and users and ultimately positively influence their visit intention.
Limitations and Future Research
We acknowledge our study has some limitations, and these present directions for future research studies. First, most of the participants are younger (71.6% of our sample are under 35), though perhaps this is not surprising for a social media-based study. This group is more likely to be on social media and thus have the potential to interact with VIs. Second, the study is limited by comparing nine theoretical models to elucidate the VI phenomenon from the travel and tourism perspective. Future studies might consider using three to four theoretical models, proposing a combined view of those models, and empirically testing the resulting unified model. Third, only a small proportion of our sample (10.26%) already followed a VI. We mitigated this by providing links to popular VIs to educate the respondents on VIs and in the preamble to our study we carefully explained the concept of the VI. We asked those who followed VIs already which VIs they followed, and our exemplars provided were popular with those who followed VIs thus providing evidence that we had chosen relevant exemplars. Fourth, our statistical analysis employed the PLS-SEM method through SMART PLS software, in which NCA findings complement the results. Since applying a configurational approach (fsQCA) has become popular among scholars, we thus assume future studies may consider examining the configurational effects between VI-related factors and visit intention under different travel contexts. Fifth, we chose one outcome variable, visit intention. Alternative measures of engagement such as heightened awareness or engagement metrics could be used as outcome variables as they are becoming more commonplace is SMI studies. Sixth, our study had a 47.48% response from non-English speaking Europe, with a further 27.6% from English speaking Europe. This is most likely as our study is conducted in English as English is widely spoken in Europe and many European citizens live in English speaking countries in Europe. Future studies may consider drawing samples from less developed, developing, and developed countries to document the cross-cultural differences in the context of antecedents of behavioral intentions toward VIs. Future cross-cultural studies on VI marketing will also help improve the generalizability of the study findings and be essential for gaining a clearer understanding of this phenomenon. Finally, our study focused on the perceptions of SNS users. One further avenue would be to engage in experimental research to more clearly understand the boundary conditions on how different VI-related factors affect the travel decision-making process. Examples of such experiments would include manipulating VI type (humanlike VI vs. cartoon-like VI), displayed gender for humanlike VIs, level of anthropomorphic attributes (high vs. low anthropomorphic VI), and VI attractiveness (highly attractive vs. less attractive VI). Also, future studies might explore travel brands’ perspectives toward VIs to gain in-depth insights into the phenomenon. Researchers may also be interested in exploring the role of arousal and dominance evoked from VIs on pre-travel behavior and examining the moderating role of types of VIs (human vs. cartoon-like vs. anime-like) on these relationships. Finally, our qualitative findings reveal a set of novel factors associated with VI marketing that may influence travel behavior. Future researchers may consider these factors from our qualitative study (i.e., VI’s collaboration with human influencers, VI’s appealing content, cultural representation by VIs) in quantitative studies.
Conclusions
Consumer behavior toward influencer marketing has garnered significant interest among research scholars; however, visit intention engendered by VIs in the tourism industry remains relatively unexplored principally because of the novelty of VIs. Given VIs’ increasing popularity and higher level of brand endorsements, somewhat increased by the global pandemic when human models were less available in person to shoot campaigns, tourism marketers’ understanding of the salient attributes that activate SNS users’ behavior has become imperative for exploiting VI marketing in the tourism sector. Through adopting a model comparison approach, our study seeks to determine the key factors associated with VIs that were significant in explaining SNS users’ visit intention. Our study is one of the first studies that applied a model-comparison approach to identify the explanatory power of the nine selected theories to study the VI phenomenon within the tourism context. Our findings highlight that while all nine theories possess explanatory power in relation to visit intention, social power and parasocial interaction theories are potentially the most effective theories. Our findings produce a set of key factors related to VIs that augment visit intention that may act as principles for scholars and marketing practitioners to develop a more nuanced view of consumer behavioral actions toward VI, especially in relation to visit intentions.
Footnotes
Appendix
Discriminant Validity Uses and Gratifications Theory.
| Entertainment value | Escapism | Integrative gratification | Purposive value | Social interaction | Social presence | Trendiness | Visit intention | |
|---|---|---|---|---|---|---|---|---|
| Heterotrait-Monotrait ratio (HTMT) | ||||||||
| Entertainment value | ||||||||
| Escapism | 0.674 | |||||||
| Integrative gratification | 0.673 | 0.683 | ||||||
| Purposive value | 0.624 | 0.662 | 0.775 | |||||
| Social interaction | 0.453 | 0.685 | 0.770 | 0.758 | ||||
| Social presence | 0.628 | 0.746 | 0.678 | 0.692 | 0.688 | |||
| Trendiness | 0.773 | 0.654 | 0.734 | 0.743 | 0.572 | 0.685 | ||
| Visit intention | 0.584 | 0.663 | 0.654 | 0.724 | 0.663 | 0.627 | 0.693 | |
| Fornell-Larcker criterion | ||||||||
| Entertainment value | 0.928 | |||||||
| Escapism | 0.630 | 0.898 | ||||||
| Integrative gratification | 0.641 | 0.639 | 0.912 | |||||
| Purposive value | 0.599 | 0.622 | 0.735 | 0.873 | ||||
| Social interaction | 0.432 | 0.639 | 0.725 | 0.712 | 0.912 | |||
| Social presence | 0.598 | 0.697 | 0.646 | 0.658 | 0.655 | 0.883 | ||
| Trendiness | 0.706 | 0.590 | 0.676 | 0.687 | 0.527 | 0.629 | 0.859 | |
| Visit intention | 0.560 | 0.626 | 0.627 | 0.695 | 0.632 | 0.601 | 0.641 | 0.934 |
Acknowledgements
None.
Author Contributions
Ahmed Al Asheq: Conceptualization; Data curation; Formal analysis; Investigation; Methodology; Project administration; Software; Validation; Visualization; Writing—original draft; Writing—review & editing.
Rajibul Hasan: Conceptualization; Data curation; Formal analysis; Funding acquisition; Investigation; Methodology; Project administration; Software; Supervision; Validation; Visualization; Writing—original draft; Writing—review & editing.
Joseph Coughlan: Conceptualization; Data curation; Formal analysis; Funding acquisition; Investigation; Methodology; Project administration; Resources; Software; Supervision; Validation; Visualization; Writing—original draft; Writing—review & editing.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The authors would like to acknowledge funding from the Maynooth University School of Business Seed Funding Scheme.
