The Role of Augmented and Virtual Reality in Shaping Retail Marketing: A Meta-Analysis

Abstract

The Fourth Industrial Revolution has brought advanced technologies such as augmented reality (AR) and virtual reality (VR), transforming consumer behavior in retailing and arousing the interest of scholars in studying customer responses to these technologies in retail settings. However, owing to variations in specific contextual factors, the results of related research have been mixed, which impedes retailers’ capacity to gain a systematic understanding of the formulation of well-informed marketing decisions in the context of AR and VR retailing. To address these gaps, this systematic review synthesizes extant empirical evidence with 1099 effect sizes from 111 published studies with 136 datasets and 547,415 sample sizes. This study is based on well-established theories, including the technology acceptance model, the customer journey theory, and the unified theory of acceptance and use of technology, which are extended to create a more comprehensive framework that is adapted for the customer journey in AR and VR retailing. Our findings reveal significant and positive correlations for all the proposed constructs, including the experience; intrinsic, extrinsic, hedonic, and utilitarian factors; and customer experience, attitude, intention, and loyalty, and verify the significant moderating effects for technology and product types. From a management perspective, our findings provide a systematic understanding of enhancing retailers’ integrated sustainable marketing strategies in the context of AR and VR retail and propose a forward-looking research agenda.

1. Introduction

In recent years, there has been a growing interest in the use of augmented reality (AR) and virtual reality (VR) technologies in various industries, such as retailing, education, healthcare, real estate, hotel and tourism, and arts and culture. Among them, in the context of retailing environments, AR and VR are leveraged to enhance consumer experience through product presentations, marketing communications, and overall sales experience. For example, AR overlays virtual objects in the natural environment, which improves consumer experience. In contrast, VR creates an immersive virtual space for consumers so that they can engage in a virtual store to shop. In addition, retailers are using AR and VR technology to contribute to sustainability by reducing the environmental impact of physical shops and large inventories, minimizing returns and waste, and promoting responsible consumption by offering virtual shopping experiences.

AR and VR retailing-related studies aim to identify the influencing factors behind customers’ purchasing decision-making processes. However, these studies have reported mixed results due to variations in specific contextual factors highlighted by Palmatier [1], de Oliveira Santini et al. [2], Luceri et al. [3], and Vieira et al. [4]. For example, the effect of perceived risks on purchase intentions has yet to be precisely determined. Herz and Rauschnabel [5] indicated that the perceived risks (e.g., healthy risks and privacy concerns) negatively correlate with hedonic and utilitarian benefits, presence, attitudes, and purchase intentions. In contrast, Pillai et al. [6] reported a positive correlation between insecurity (i.e., privacy concerns) and shopping intentions. Such expositions are unsatisfactory because they need to be more conclusive and accordant. Furthermore, previous studies have predominantly relied on a single sample, product, or technology type; thus, the moderating effects have yet to be investigated, as Babić Rosario et al. [7] noted. For example, while meta-analyses have focused on customer behavior through AR (e.g., [4,8]), they have only focused on AR without integrating it into broader application scenarios or synthesizing it with VR technologies. Current studies have shown the rationale for integrating AR and VR research in a meta-analysis (e.g., [9,10]). AR and VR have been increasingly used to support marketing campaigns by creating engaging and appealing shopping experiences for consumers. Despite this, there has been a surprising lack of synthesis in the empirical research on customer purchasing behavior in the context of AR- and VR-enhanced retail environments. The inconsistent results from previous research, driven by specific contextual factors, impede retailers’ ability to gain a systematic understanding for making comprehensive marketing decisions in AR and VR retailing, underscoring the urgent need for a meta-analysis to provide convincing and comprehensive empirical evidence [1,3]. Therefore, our systematic review and meta-analysis synthesizes and generates the current empirical results and aims to (i) reveal a more convincing bivariate correlation between antecedents and consequences and test the moderating effects in a proposed comprehensive framework (which will be detailed and elaborated upon in the subsequent sections), (ii) identify the business opportunities and challenges in AR and VR retailing, and (iii) suggest an agenda for future research.

To achieve the research objectives, this study takes the form of meta-analytic structural equation modeling (MASEM), which combines a meta-analysis (MA) and structural equation modeling (SEM), to test the hypothesis of the comprehensive model [3,11] and investigate the factors that affect the customer journey and drive customer purchasing in retailing. Glass [12] first proposed the concept of a meta-analysis, intending to answer the original questions via an improved statistical method or addressing new questions with the primary data used in previous empirical research [11,12]. As the trend toward evidence-based research grows, a cluster of studies provides higher reliability than a single study does, reducing inconsistent and inconclusive results [1,7]. Moreover, Viswesvaran and Ones [13] proposed the MASEM approach in social science, in which the estimated true score correlations between constructs of interest are established through an MA, and then SEM is applied to estimate the correlation matrix. While the existing research has commonly adopted either MA or SEM, combining both methods has several attractive features. First, in previous studies, the MASEM helps establish connections between two constructs without a complete bivariate correlation [14,15,16]. As many studies can include only some of the variables of interest [17], MASEM allows scholars to address research questions that cannot be addressed by SEM on the basis of primary data [11,18]. Second, by integrating samples from pooled studies, a meta-analysis fulfills the sample size requirements (typically greater than 200) necessary for stable estimation results in SEM, surpassing the limitations of single results from primary research [16]. Finally, MASEM presents a potential solution for discrepant results in existing empirical studies [18], as small samples can lead to vastly different results when investigating the same phenomenon [11].

First, our research is committed to delivering broadly relevant conclusions. Theoretically, it addresses existing gaps by introducing a comprehensive framework that explores the relationships between antecedents and consequences, as well as potential moderating effects, within AR and VR retailing. Second, this study illuminates systematic approaches for retailers to develop comprehensive marketing strategies tailored to AR and VR marketing. It further enhances our understanding of both linear and nonlinear customer journeys in this context. Additionally, our research expands upon the foundational work of Vieira et al. [4], deepening our insights into retail activities facilitated through AR and VR.

The rest of this paper is organized as follows. First, we propose the conceptual framework and hypotheses. Then, we describe the research procedures involving data collection, coding, and analysis. Next, we discuss the results of the MASEM and moderator analysis. We also conclude this research, illustrate the contributions and limitations, and propose a future research agenda.

2. Conceptual Framework and Hypotheses Development

AR and VR are similar in that they aim to enrich and enhance customers’ experience [19]. For example, AR uses touch control to heighten customers’ perceived value within a multisensory retail context [20], whereas VR can simulate sensory experiences, such as smells, through a computer-brain interface [21]. However, they differ in terms of the experienced subject. AR applies virtual products to the natural body of the customer, whereas VR allows the avatar to experience virtual products [19]. Moreover, AR is a more realistic technology [22], as it combines virtual elements with reality and augments the perception of reality rather than substituting reality [23].

2.1. Augmented Reality (AR)

AR embodies the interactive incorporation of virtual elements within the users’ authentic surroundings [19,24]. Its primary objective is to enrich the user’s engagement throughout the purchasing journey [4,19] and to facilitate visualization and interaction with virtual products within real-world contexts [25]. Rauschnabel et al. [26] underscored the pivotal significance of AR, emphasizing its indispensable role in contemporary marketing and strategic frameworks. The inception of AR dates back to the 1990s, when it was pioneered by Thomas and David [27]. However, only in recent years has AR become readily accessible on various smartphones owing to significant technological advancements [28].

A recent survey examining AR applications suggests a broadening interpretation of AR, defining it as the seamless integration of virtual entities, regardless of type, into real-time environments [29]. Consequently, the essence of AR lies in its ability to merge physical and virtual realms, epitomizing the convergence of reality and digital augmentation [28]. In essence, real-time superimposition is the fundamental characteristic of AR [30].

Existing AR applications can be categorized into four distinct types: shopping, informational, entertainment, and social media [31]. The burgeoning field of AR marketing has garnered considerable attention among marketing scholars, evolving into a distinct subdiscipline within the broader marketing domain, delineated as AR marketing [26]. Rauschnabel et al. [26] defined AR as the strategic amalgamation of AR experiences, either independently or in conjunction with affiliated cues or media, to achieve holistic marketing objectives by generating value for brands, stakeholders, and society while simultaneously considering ethical and moral implications. Owing to its seamless integration capabilities across smartphones and other mobile devices, AR has the capacity to seamlessly infiltrate consumers’ daily routines [20]. In an era characterized by fierce competition for consumer attention, AR technology offers businesses novel avenues to captivate audiences [32]. By enriching the perceptual landscape with virtual objects, AR elevates consumers’ perceptions of reality [33]. Retailers leverage AR technology to heighten consumer engagement with branded merchandise and enhance the in-store shopping experience [34]. For example, Yang et al. [35] investigated the application of AR in the beauty hairdressing industry and reported that spatiality significantly affects behavioral intention.

Scholars have suggested that AR technologies are catalysts for enriching the customer experience, facilitating informed decision-making processes through AR-enabled services [36]. For example, AR’s unique ability to visually superimpose objects onto real-time environments heightens consumers’ ability to mentally simulate the consumption of depicted food items, thereby bolstering desire and purchase intent [30]. Hoffmann et al. [37] elucidate two primary functions of AR within the realm of retailing: augmenting self-exploration through virtual try-on experiences (e.g., apparel, eyewear, cosmetics) and augmenting physical environments (e.g., furniture). This delineation underscores AR’s versatile utility in transforming the retail landscape, offering immersive consumer experiences that bridge the gap between virtual and physical realms.

2.2. Virtual Reality (VR)

VR is a sophisticated simulation system that offers users an immersive artificial experience within a three-dimensional environment [24]. Coined by Jaron Lanier, the CEO of VPL Research, Inc., in 1989 [38], the term VR encompasses both the technological hardware—such as computers, headphones, head-mounted displays, and motion-sensing gloves—and the virtual environments themselves. These environments vary widely, from stitched 360° photographs and interactive 360° videos to virtual world environments and immersive headset-based VR [39]. Steuer [38] conceptualized VR as “a real or simulated environment in which a perceiver experiences telepresence”.

VR applications are characterized by three distinct aspects: virtual worlds, automated virtual environments, and product simulations [40]. The technology employs wearable devices, such as headsets, to occlude the sensory inputs of the physical world, thereby creating a more engaging and innovative shopping environment by immersing users in a vivid and interactive three-dimensional world [41]. Notably, Suh and Lee [42] were pioneers in identifying the beneficial impacts of VR on customer learning, particularly with products that require high experiential interaction. Their research highlighted the advancements from web-based VR, which primarily utilize visual and auditory modalities, to contemporary VR headsets that also incorporate haptic feedback, significantly enhancing the multisensory experience.

Recent research has underscored the potential advantages of employing VR technology within retail contexts [43]. VR, facilitated by head-mounted displays, offers users a comprehensive immersive experience within a three-dimensional environment. Notably, VR has been identified as more cost-effective than smartphone devices [21]. According to Herz and Rauschnabel [5], VR technology can closely mimic real-life settings, significantly enhancing customer immersion. Their empirical findings suggest that VR can improve the accuracy of predicting product adoption rates. Additionally, VR shopping experiences have been shown to augment consumer satisfaction and foster retail brand loyalty [43]. VR also enables retailers to craft interactive consumer experiences, bridging physical distances effectively [40]. However, Herz and Rauschnabel [5] cautioned that consumers’ willingness to adopt VR is contingent upon experiencing a robust sense of avatar (the sensation of embodying another person) and virtual presence (the sensation of being in another location). Notably, the absence of one of these elements may negatively affect consumer engagement with VR technology.

2.3. Theoretical Background and Conceptual Framework

Multiple theories have been proposed in the academic literature to explain customer behavior. Previous studies have focused primarily on examining customer behavior via a single theoretical model, such as the technology acceptance model (TAM) (e.g., [6,24,29,44]), the theory and planned behavior model (e.g., [35,45]), the unified theory of acceptance and use of technology (UTAUT) [46], customer journey [26,47], the theory of consumption values (TCV), prospect theory [45] and flow theory [40]. The TAM and UTAUT have been widely used models for new technology reception [4]. Although these studies have shed light on various factors that impact customer behavior, there is a pressing need for a more comprehensive and intricate framework to elucidate the interrelationships among the different driving forces of customer purchasing and their influence on the customer journey in retailing.

In marketing, the primary objective is to satisfy customers’ needs and deliver value at a profit [48]. The customers are the final purchasers of the product and service. They undergo a long decision process when considering a purchase [49], which depends on both the customer and the product. This decision-making process encompasses the customer journey, and businesses must possess a profound understanding of and the ability to manage the entirety of the customer experience throughout their shopping expedition [50]. Within this context, scholars have identified distinct stages within the customer’s purchasing decision process (e.g., [26]), which correspond to three main phases: pre-purchase, purchase, and post-purchase (e.g., [47,51,52]). However, customer journeys can be nonlinear and involve emotional, cognitive, and behavioral responses. Customers may bypass the traditional linear process by transitioning directly between the pre- and post-purchase stages [50]. To adapt to these nonlinear customer journeys, Farah et al. [49] suggest that retailers should endeavor to create experiential opportunities that enable the transformation of linear customer journeys into nonlinear ones.

Figure 1 presents a conceptual framework depicting the antecedents and outcomes of the customer journey in retailing on the basis of the work of Luceri et al. [3]. The framework integrates various important variables as antecedents that impact the purchasing decision process, drawing from existing theories. Additionally, the customer decision process within the customer journey is considered the primary outcome. To account for study-level variations, moderators are included in the analysis. The subsequent sections will elaborate on the outcomes, antecedents, and moderators of the framework. Additionally, a summary of the 14 constructs can be found in

Table A1. Definition of key constructs.

Construct	Description	Common Aliases	Source	Expected Effect
Consequences
Customer Experience (CX)	The subjective and internal response of customers who have direct (e.g., purchasing, service, and use) or indirect (e.g., WOM recommendations, criticisms, and advertisements) contact with companies [53].	User experience, rapport experience, flow experience		Positive
Attitude (ATT)	The extent to which a person has a favorable or unfavorable evaluation or assessment of the behavior under discussion [56].	Attractiveness, satisfaction, wow effect, trust	[56]	Positive
Intention (INT)	Thought to capture the motivational factors that affect the behavior [56].	Willingness to pay	[56]	Positive
Loyalty (LOY)	Customers become or remain loyal to a brand [26].		[26]
Antecedents
Perceived system quality (PSQ)	The perceived quality of the technology system, including augmented quality, display fidelity, and immersion degree.	Vividness, immersive, display fidelity, virtuality, media richness, sensory pleasantness, anthropomorphism, interface design, aesthetic quality, perceived store prestige, graphic style, esthetic	[3]	Positive
Presence (PRE)	The increased intimacy, social interaction [60], and the sense of being together [61].	Tele-presence, socializing, customization, and personalization, consumer engagement	[60,61]	Positive
Perceived novelty (PN)	The extent to which each response was rated unique, new, and different [63].	Curiosity, innovativeness; inspiration	[63]	Positive
Perceived ease of use (PEU)	Customers’ beliefs about the ease of using the information technology and whether the effort they put into the usage is higher than the performance improved by using it [65].	Complexity [93]	[65]	Positive
Perceived usefulness (PU)	Customers’ beliefs about the usefulness of information technology enhancing their performance or not [65].		[65]	Positive
Informativeness (INF)	Customers search for information about the product, aiming to increase their confidence.	Perceived informativity, WOM, branding, advertisement, media richness
Perceived enjoyment (PE)	The degree to which an activity using an information system is enjoyable [67].	Pleasure, predicted happiness, emotional engagement	[67]	Positive
Hedonic values (HV)	The extent of pleasure derived from the multi-sensory, fantastical, and emotional aspects of the product usage experience [68,69]	Entertainment, hedonic benefits, hedonic gratification	[68,69]	Positive
Perceived risk (PR)	Perceived risk with adopting interactive technologies in metaverse retailing, including perceived product risk and privacy issues.		[94,95,96]	Negative
Utilitarian values (UV)	The functional values from using technology [5].	Pragmatic quality, functional value, utilitarian gratification	[5]	Positive

in Appendix A.

2.3.1. Consequences of the Customer Shopping Journey

This section discusses the consequences of the customer shopping journey on the basis of the conventional customer journey. Their bivariate relationships with respect to the customer decision process are established through the proposed hypothesis. These consequences are customers’ responses [4] to using AR and VR while shopping.

Customer experience (CX) denotes the subjective and internal response of individuals who engage directly (e.g., through purchasing, receiving services, or product usage) or indirectly (e.g., through word-of-mouth recommendations, criticisms, or advertisements) with companies [53]. Businesses recognize that the key to achieving success lies in delivering exceptional shopping experiences to their customers [54]. This is because a remarkable customer experience has the potential to influence customers’ purchase intentions [55]. For example, the ability to physically interact with products or to experience them through AR or VR devices heightens the customer experience, exerts a positive influence on their attitudes, and facilitates their decision-making process.

Attitude (ATT) towards the behavior, which refers to the extent to which a person has a favorable or unfavorable evaluation or assessment of the behavior under discussion [56]. It holds significant importance as an outcome variable within the customer journey [45] and is the best predictor for assessing customer intention for groceries [54]. Furthermore, attitude plays a crucial role and is a predictor of intention and loyalty, especially in a retail environment. Our first hypothesis posits the following relationship:

H1. 

Customer experience (CX) significantly and positively affects attitude (ATT).

Intention (INT) captures the motivational factors that affect behaviors [56]. It refers to the first time a user considers using or adopting a technology or a channel [3]. In the context of this study, intention refers to customers’ consideration of purchasing in a retail setting using AR or VR technologies. Additionally, intention also serves as a predictor of customer loyalty in a retail setting. We present the second hypothesis:

H2. 

Attitude (ATT) significantly and positively affects intention (INT).

Loyalty (LOY) refers to customers becoming or remaining loyal to a brand [26], including continued buying, recommendation behavior, and brand preferences [57]. Previous research has demonstrated the positive effects of customer experience on loyalty through repeated purchases [58]. Furthermore, the nonlinear nature of customer journeys, such as skipping directly from the pre-purchase phase to the post-purchase phase, suggests that customer experience can directly influence loyalty [50]. Therefore, we put forward the following two hypotheses:

H3. 

Intention (INT) significantly and positively affects loyalty (LOY).

H4. 

Customer experience (CX) significantly and positively affects loyalty (LOY).

2.3.2. Antecedents of the Customer Shopping Journey

In this section, ten common antecedents are identified and categorized into five factors: experience factors, intrinsic factors, extrinsic factors, hedonic factors, and utilitarian factors. These antecedents are hypothesized to be related to the consequences of the customer shopping journey.

Experience factors

Perceived system quality (PSQ) refers to the perceived quality of the technology system, including the augmented quality, display fidelity, and immersion degree. Customers pay attention to the augmentation quality [25], and a higher AR or VR perceived quality can lead to spatial aesthetics, which increase customer purchasing intention [59]. Thus, we raise the hypothesis as below:

H5. 

Perceived system quality (PSQ) significantly and positively affects customer experience (CX).

Presence (PRE) refers to increased intimacy, social interaction [60], and a sense of being together [61]. Pillai et al. [6] reported that social interaction enhances the customer shopping experience. Real-time multisensory social interaction is central, allowing users to create value and engage in social presence and activities [62]. Hence, we put forward the following hypothesis:

H6. 

Presence (PRE) significantly and positively affects customer experience (CX).

Perceived novelty (PN) refers to the extent to which the experience is rated as unique, new, and distinct [63]. Yuan et al. [64] reported that when consumers perceive AR media technology as novel and unfamiliar stimuli, their allocation of cognitive resources to AR may intensify, thereby overcoming distractions in the AR context and experiencing a flow state. In AR or VR marketing, innovativeness offers customers novel experiences. Therefore, we propose the seventh hypothesis:

H7. 

Perceived novelty (PN) significantly and positively affects customer experience (CX).

Intrinsic factors

The intrinsic factors have two critical observation indicators in the TAM, which were also adopted by Luceri et al. [3].

Perceived ease of use (PEU) refers to customers’ beliefs about the ease of using the information technology and whether the effort they put into the usage is greater than the level of performance improvement by using it [65]. Perceived ease of use indirectly influences the actual use intention of the technology [44]. Based on Davis [65], we bring forward the following hypothesis:

H8. 

Perceived ease of use (PEU) significantly and positively affects attitude (ATT).

Perceived usefulness (PU) refers to customers’ beliefs about whether the usefulness of information technology enhances their performance [65], which directly influences the actual use intention of the technology [44]. We propose the following hypotheses:

H9. 

Perceived ease of use (PEU) significantly and positively affects perceived usefulness (PU).

H10. 

Perceived usefulness (PU) significantly and positively affects attitude (ATT).

Extrinsic factors

Informativeness (INF) Customers search for information about the product, aiming to increase their confidence. Customers utilize mental imagery to address the lack of product information [66]. Word-of-mouth (WOM), advertising, branding, and product experience can enhance perceived product information in AR-supported retailing and positively affect attitudes. Hence, we put forward the hypothesis below:

H11. 

Informativeness (INF) significantly and positively affects attitude (ATT).

Hedonic factors

Perceived enjoyment (PE) refers to the degree to which an activity using an information system is enjoyable [67]. Scholars usually define hedonic benefits as perceived enjoyment [37]. This study describes PE as the degree of perceived enjoyment for customers using AR or VR. We raise the following hypothesis:

H12. 

Perceived enjoyment (PE) significantly and positively affects intention (INT).

Hedonic value (HV) refers to the extent of pleasure derived from the multisensory, fantastical, and emotional aspects of the product usage experience [68,69]. Multisensory AR (e.g., [20]) and VR (e.g., [49]) provide more customer responses with hedonic motivation and enhance the shopping experience and purchasing intention [70]. Therefore, we present the below hypothesis:

H13. 

Hedonic values (HV) significantly and positively affect intention (INT).

Utilitarian factors

Perceived risk (PR) refers to the perceived risk of adopting interactive technologies in retailing, including perceived product risk and privacy issues. For example, when customers try on products via AR in their houses, the pictures of their houses will be transferred to the cloud processor [25], which leads to privacy issues. Following Luceri et al. [3], perceived risks, such as privacy concerns and technology immaturity, decrease usage intention and frequency. Based on this, we come up with the following hypothesis:

H14. 

Perceived risk (PR) significantly and negatively affects intention (INT).

Utilitarian value (UV) refers to the functional value from using technology [5]. For example, AR beauty apps satisfy customers’ utilitarian motivations and values by providing practical and trustworthy knowledge and information [71]. Hence, we propose the last hypothesis as follows:

H15. 

Utilitarian values (UV) significantly and positively affect intention (INT).

In summary, this study develops fifteen hypotheses that aim to test the effects of the factors (antecedents) on the customer decision process (consequences). We also aim to investigate the relationships among the constructs in the customer journey.

2.3.3. Moderators

In addition to synthesizing various studies, meta-analyses can examine the differences among subsets by analyzing the moderating effects [11]. Given that the proposed relationships may be contingent upon additional factors (e.g., [3]), we also investigate the potential moderators that might reduce the heterogeneity across studies [2] and apply our conceptual model across different contexts [1]. In this study, we consider the technology context type, product type, publication year, journal ranking, and research subject as moderators. Additionally, the coding and explanation of the moderator category can be found in

Table A2. Moderators.

Moderator	Category	Dummy Variables	Description
Tech type	AR VR and others	AR = 1 VR and others = 0	Whether participants experience through VR or AR or other technologies.
Product type	Non-digital Digital	Non-digital = 1 Digital and others = 0	Whether participants experience digital or non-digital or no products.
Year	2021 and after 2021 Before 2021	2021 and after 2021 = 1 Before 2021 = 0	Whether this paper published before 2021 or not.
Journal ranking	Group 1 (Relative higher ranking) Group 2 (Others)	Relative higher ranking = 1 Others = 0	Whether this paper published in higher ranked journal (i.e., ABDC-A and above, ABS-3 and above) or not.
Sample type	Only student Not only and non-student	Only student = 1 Not only and non-student = 0	Whether the participants are only students or not only and non-students.

in Appendix A.

Technology context type (tech type). Retailing includes various retail-related activities through interactive virtual-reality technologies (e.g., AR and VR). The primary application of the brand-new topic in AR (e.g., [28,72,73]) and VR (e.g., [5,46]) has been discussed in early research. Hence, we discuss three technology context types: AR, VR, and others.

Product type. As the product type influences the transaction process and channel choices in the digital environment [3], it serves as a potential moderator for our proposed conceptual framework. With respect to the virtual-reality properties, the product attributes in retailing can be categorized into digital products (e.g., virtual property) and nondigital products (e.g., daily necessities).

Publication year (year). Consumer behavior is influenced by prevailing technological conditions and is subject to change over time [3]. Given the rapid technological development and changes over time, this study aims to examine whether there are differences in the shopping journey on the basis of the publication year. As 2021 is commonly regarded as the first year of the era, we divide the studies into two periods: before 2021 (excluded) and after 2021 (included).

Journal ranking. To reduce the file drawer problem and publication bias [3,74], we include the rankings of the journals. According to the Australian Business Deans Council (ABDC) ranking and Academic Journal Quality Guide (AJG) list released by The Association of Business Schools (ABS), we are interested in two ranking groups: the higher ranking group (ABS 3 and above; ABDC A and above) and others.

Research subject (sample type). Previous empirical research has relied on a single sample, and different demographic characteristics may influence the outcomes of a shopping journey. Previous studies have investigated retailing on the basis of student groups (e.g., [70]), as they consider new technological innovation features [74]. Hence, we divide the research subjects into two groups: only students and not only nonstudents.

Overall, we synthesized existing theories and generated a total of 14 constructs. On the basis of these constructs, we propose 15 hypotheses to examine the bivariate correlation between the antecedents and consequences in the customer shopping journey (see Figure 2 for the conceptual framework for an empirical test). Additionally, we identified five moderators that may influence the relationships within our proposed conceptual framework.

3. Method

Scholars have employed the use of meta-analyses and their higher-level applications (e.g., meta-regression and MASEM) to consolidate and generate a comprehensive framework that can address the existing conflicting and inconclusive empirical findings (e.g., [1,3,4,7,75,76,77]). We analyze various empirical studies that use the MASEM by adapting the steps suggested by Viswesvaran and Ones [13] to conceptualize the MASEM and convert them into three main steps: data collection, coding, and analysis.

3.1. Data Collection

The data collection process followed the PRISMA guidelines [78,79] (see Figure 3), which are also widely used in the previous meta-analysis literature (e.g., [80,81]). We conducted a comprehensive literature search via the Web of Science (WoS) database, which is a widely used academic platform (e.g., [82]). We used the keywords “metaverse” OR “virtual reality” OR “augmented reality” (abstract) and “consumer” OR “customer” AND “purchase*” (abstract). Additionally, we manually searched other databases (e.g., EBSCO) to ensure comprehensive coverage. The following inclusion criteria were applied to filter the search results: language (English, N = 2428), document type (article and meeting, N = 1034), database (WoS core collection, N = 1010), publication year (2004 to 2023, N = 993), and research area (business economics, communication, psychology, behavioral science, geography, N = 442). Conference papers were also included to minimize the file drawer problem following Rosenthal [74] and to reduce publication bias. After the titles and abstracts were examined, irrelevant papers were excluded. The remaining research papers were downloaded, and those without full-text availability were excluded (N = 218). The full texts of the remaining papers were then reviewed, and studies that did not report correlation coefficients (e.g., Pearson’s $r$ or standardized $\beta$) were excluded (N = 111). The publications or manuscripts were included in the meta-analysis only if they satisfied the criteria above and reported at least one effect size for the correlation between the proposed constructs. In total, 111 studies were included in the final selection, meeting the minimum sample requirement of MASEM suggested by Jak et al. [17], in which analyzing a sample of thirty studies was sufficient to obtain correct standard errors and parameter estimates. Detailed information on the selected literature can be found in

Table A3. Literature List for MASEM.

ID	Title	Author	Year	Journal
1	The personalization-privacy paradox: Consumer interaction with smart technologies and shopping mall loyalty	Ameen, Hosany & Paul [94]	2022	Computers in Human Behavior
2	Role of augmented reality in changing consumer behavior and decision making: Case of Pakistan	Kazmi et al. [72]	2021	Sustainability (Switzerland)
3	The mediation effect of marketing activities toward augmented reality: the perspective of extended customer experience	Chen et al. [97]	2022	Journal of Hospitality and Tourism Technology
4	Retail consumers’ behavioral intention to use augmented reality mobile apps in Pakistan	Saleem et al. [73]	2022	Journal of Internet Commerce
5	Mechanism linking AR-based presentation mode and consumers’ responses: A moderated serial mediation model	Han et al. [96]	2021	Journal of Theoretical and Applied Electronic Commerce Research
6	Enhancing brick-and-mortar store shopping experience with an augmented reality shopping assistant application using personalized recommendations and explainable artificial intelligence	Zimmermann et al. [98]	2022	Journal of Research in Interactive Marketing
7	The effect of augmented reality on purchase intention of beauty products: The roles of consumers’ control	Whang et al. [36]	2021	Journal of Business Research
8	The adoption of virtual reality devices: The technology acceptance model integrating enjoyment, social interaction, and strength of the social ties	Lee, Kim & Choi [99]	2019	Telematics and Informatics
9	How augmented reality media richness influences consumer behavior	de Amorim et al. [32]	2022	International Journal of Consumer Studies
10	Augmented reality’s perceived immersion effect on the customer shopping process: Decision-making quality and privacy concerns	Sengupta & Cao [100]	2022	International Journal of Retail and Distribution Management
11	Extending the technology acceptance model to explain how perceived augmented reality affects consumers’ perceptions	Oyman, Bal & Ozer [101]	2022	Computers in Human Behavior
12	Shopping in the digital world: Examining customer engagement through augmented reality mobile applications	McLean & Wilson [102]	2019	Computers in Human Behavior
13	Virtual reality in new product development: Insights from prelaunch sales forecasting for durables.	Harz, Hohenberg & Homburg [103]	2022	Journal of Marketing
14	Blending the real world and the virtual world: Exploring the role of flow in augmented reality experiences	Brannon Barhorst et al. [104]	2021	Journal of Business Research
15	The impact of representation media on customer engagement in tourism marketing among millennials	Willems, Brengman & Van Kerrebroeck [39]	2019	European Journal of Marketing
16	Enhancing the online decision-making process by using augmented reality: A two country comparison of youth markets	Pantano, Rese & Baier [105]	2017	Journal of Retailing and Consumer Services
17	The effects of augmented reality mobile app advertising: Viral marketing via shared social experience	Sung (Christine) [106]	2021	Journal of Business Research
18	Understanding the diffusion of virtual reality glasses: The role of media, fashion and technology	Herz & Rauschnabel [5]	2019	Technological Forecasting and Social Change
19	How mobile augmented reality applications affect continuous use and purchase intentions: A cognition-affect-conation perspective	Qin, Osatuyi & Xu [107]	2021	Journal of Retailing and Consumer Services
20	User interfaces and consumer perceptions of online stores: The role of telepresence.	Suh & Chang [108]	2006	Behavior and Information Technology
21	“Yes, we do. Why not use augmented reality?” Customer responses to experiential presentations of AR-based applications	Hsu, Tsou & Chen [71]	2021	Journal of Retailing and Consumer Services
22	Enhancing the sneakers shopping experience through virtual fitting using augmented reality	Rhee & Lee [109]	2021	Sustainability (Switzerland)
23	Mobile augmented reality in electronic commerce: Investigating user perception and purchase intent amongst educated young adults	Haile & Kang [110]	2020	Sustainability (Switzerland)
24	Augmented reality advertising via a mobile app	Sung, Han & Choi [111]	2022	Psychology and Marketing
25	Telepresence, time distortion, and consumer traits of virtual reality shopping	Han et al. [112]	2020	Journal of Business Research
26	Beyond the gimmick: How affective responses drive brand attitudes and intentions in augmented reality marketing	Zanger, Meißner & Rauschnabel [113]	2022	Psychology and Marketing
27	Virtual try-on: How to enhance consumer experience?	Bialkova & Barr [114]	2022	2022 IEEE Conference on Virtual Reality and 3D User Interfaces Abstracts and Workshops (VRW), Virtual Reality and 3D User Interfaces Abstracts and Workshops (VRW), 2022 IEEE Conference on, VRW
28	A virtual market in your pocket: How does mobile augmented reality (MAR) influence consumer decision-making?	Qin, Peak & Prybutok [115]	2021	Journal of Retailing and Consumer Services
29	An adoption framework for mobile augmented reality games: The case of Pokémon Go	Rauschnabel, Rossmann & tom Dieck [116]	2017	Computers in Human Behavior
30	Perception is reality… How digital retail environments influence brand perceptions through presence	Cowan et al. [117]	2021	Journal of Business Research
31	Nostalgia beats the wow-effect: Inspiration, awe and meaningful associations in augmented reality marketing	Hinsch, Felix & Rauschnabel [33]	2020	Journal of Retailing and Consumer Services
32	Effects of physical, non-immersive virtual, and immersive virtual store environments on consumers’ perceptions and purchase behavior	Lombart et al. [118]	2020	Computers in Human Behavior
33	Gathering pre-purchase information for a cruise vacation with virtual reality: the effects of media technology and gender	Martínez-Molés et al. [119]	2022	International Journal of Contemporary Hospitality Management
34	Measuring the content characteristics of videos featuring augmented reality advertising campaigns	Feng & Xie [93]	2018	Journal of Research in Interactive Marketing
35	Shopping with augmented reality: How wow-effect changes the equations!	Arghashi [120]	2022	Electronic Commerce Research and Applications
36	The influence of flow experience in the augmented reality context on psychological ownership	Yuan et al. [64]	2021	International Journal of Advertising
37	Does virtual reality attract visitors? The mediating effect of presence on consumer response in virtual reality tourism advertising	Lo & Cheng [121]	2020	Information Technology and Tourism
38	A holistic analysis towards understanding consumer perceptions of virtual reality devices in the post-adoption phase	Dehghani et al. [122]	2022	Behavior and Information Technology
39	Discernible impact of augmented reality on retail customer’s experience, satisfaction and willingness to buy	Poushneh & Vasquez-Parraga [123]	2017	Journal of Retailing and Consumer Services
40	Augmented reality shopping application usage: The influence of attitude, value, and characteristics of innovation	Jiang, Wang & Yuen [124]	2021	Journal of Retailing and Consumer Services
41	Exploring store atmospherics of FMCG brands flagship stores with an immersive 180-degree dome-shaped display	Duong et al. [125]	2022	Journal of Global Scholars of Marketing Science
42	Can augmented reality satisfy consumers’ need for touch?	Gatter, Hüttl-Maack & Rauschnabel [126]	2022	Psychology and Marketing
43	How nostalgic feelings impact Pokémon Go players—integrating childhood brand nostalgia into the technology acceptance theory	Harborth & Pape [127]	2020	Behavior and Information Technology
44	Effects of perceived interactivity of augmented reality on consumer responses: A mental imagery perspective	Park & Yoo [128]	2020	Journal of Retailing and Consumer Services
45	Is augmented reality technology an effective tool for e-commerce? An interactivity and vividness perspective	Yim, Chu & Sauer [19]	2017	Journal of Interactive Marketing
46	Cognitive, affective, and behavioral consumer responses to augmented reality in e-commerce: A comparative study	Kowalczuk, Siepmann, née & Adler [129]	2021	Journal of Business Research
47	How augmented reality apps are accepted by consumers: A comparative analysis using scales and opinions	Rese et al. [130]	2017	Technological Forecasting and Social Change
48	How mobile augmented reality digitally transforms the retail sector: Examining trust in augmented reality apps and online/offline store patronage intention	Kang et al. [131]	2022	Journal of Fashion Marketing and Management
49	Consumer response to virtual CSR experiences	Lee, Zhao & Chen [132]	2021	Journal of Current Issues and Research in Advertising
50	Relationships between the “Big Five” personality types and consumer attitudes in Indian students toward augmented reality advertising	Srivastava et al. [133]	2021	Aslib Journal of Information Management
51	Augmented reality filters on social media: Analyzing the drivers of playability based on uses and gratifications theory	Ibáñez-Sánchez, Orús & Flavián [134]	2022	Psychology and Marketing
52	User perceptions of 3D online store designs: An experimental investigation	Krasonikolakis et al. [135]	2021	Information Systems and e-Business Management
53	How 3D virtual reality stores can shape consumer purchase decisions: The roles of informativeness and playfulness	Kang, Shin & Ponto [136]	2020	Journal of Interactive Marketing
54	The impact of experiential augmented reality applications on fashion purchase intention	Watson, Alexander & Salavati [137]	2020	International Journal of Retail and Distribution Management
55	Immersive multisensory virtual reality technologies for virtual tourism: A study of the user’s sense of presence, satisfaction, emotions, and attitudes	Melo et al. [138]	2022	Multimedia Systems
56	Augmented reality interactive technology and interfaces: A construal-level theory perspective	Chiang (Luke), Huang & Chung [139]	2022	Journal of Research in Interactive Marketing
57	Augmented reality (AR) app use in the beauty product industry and consumer purchase intention	Wang, Ko & Wang [140]	2022	Asia Pacific Journal of Marketing and Logistics
58	A model of acceptance of augmented-reality interactive technology: The moderating role of cognitive innovativeness	Huang & Liao [44]	2015	Electronic Commerce Research
59	A hierarchical model of virtual experience and its influences on the perceived value and loyalty of customers	Piyathasanan et al. [61]	2015	International Journal of Electronic Commerce
60	Examining the antecedents and consequences of perceived shopping value through smart retail technology	Adapa et al. [52]	2020	Journal of Retailing and Consumer Services
61	Privacy threats with retail technologies: A consumer perspective	Pizzi & Scarpi [95]	2020	Journal of Retailing and Consumer Services
62	Consumer-computer interaction and in-store smart technology (IST) in the retail industry: The role of motivation, opportunity, and ability	Roy, Balaji, & Nguyen [141]	2020	Journal of Marketing Management
63	Shopping intention at AI-powered automated retail stores (AIPARS)	Pillai, Sivathanu & Dwivedi [6]	2020	Journal of Retailing and Consumer Services
64	Anthropomorphism and augmented reality in the retail environment	Van Esch et al. [142]	2019	Journal of Retailing and Consumer Services
65	Chatbots in retailers’ customer communication: How to measure their acceptance?	Rese, Ganster & Baier [29]	2020	Journal of Retailing and Consumer Services
66	Touch it, swipe it, shake it: Does the emergence of haptic touch in mobile retailing advertising improve its effectiveness?	Mulcahy & Riedel [143]	2020	Journal of Retailing and Consumer Services
67	Me or just like me? The role of virtual try-on and physical appearance in apparel M-retailing	Plotkina & Saurel [144]	2019	Journal of Retailing and Consumer Services
68	Understanding the virtual tours of retail stores: How can store brand experience promote visit intentions	Baek et al. [145]	2020	International Journal of Retail and Distribution Management
69	Building e-commerce satisfaction and boosting sales: The role of social commerce trust and its antecedents	Lin, Wang & Hajlj [146]	2019	International Journal of Electronic Commerce
70	A study on the reciprocal relationship between user perception and retailer perception on platform-based mobile payment service	Lee, Ryu & Lee [147]	2019	Journal of Retailing and Consumer Services
71	Augmented reality in smart retailing: A (n) (A) symmetric approach to continuous intention to use retail brands’ mobile AR apps	Nikhashemi et al. [148]	2021	Journal of Retailing and Consumer Services
72	Can a retail environment be simulated by photographs?	Willems, Doucé & Petermans [149]	2021	Journal of Marketing Management
73	Interactivity, inspiration, and perceived usefulness! How retailers’ AR-apps improve consumer engagement through flow	Arghashi & Yuksel [150]	2022	Journal of Retailing and Consumer Services
74	How augmented reality (AR) experience affects purchase intention in sport E-commerce: Roles of perceived diagnosticity, psychological distance, and perceived risks	Uhm et al. [151]	2022	Journal of Retailing and Consumer Services
75	UTAUT in metaverse: An “Ifland” case	Lee & Kim [46]	2022	Journal of Theoretical and Applied Electronic Commerce Research
76	How close do we feel to virtual product to make a purchase decision? Impact of perceived proximity to virtual product and temporal purchase intention	Poushneh [152]	2021	Journal of Retailing and Consumer Services
77	Consumer switching behavior to an augmented reality (AR) beauty product application: Push-pull mooring theory framework	Nugroho & Wang [153]	2023	Computers in Human Behavior
78	Using augmented reality to reduce cognitive dissonance and increase purchase intention	Barta, Gurrea & Flavián [154]	2023	Computers in Human Behavior
79	Consumer engagement via interactive artificial intelligence and mixed reality	Sung (Christine) et al. [155]	2021	International Journal of Information Management
80	Living the experience before you go… but did it meet expectations? The role of virtual reality during hotel bookings	McLean & Barhorst [156]	2022	Journal of Travel Research
81	Smarter real estate marketing using virtual reality to influence potential homebuyers’ emotions and purchase intention	Azmi et al. [157]	2022	Smart and Sustainable Built Environment
82	Augmented reality generalizations: A meta-analytical review on consumer-related outcomes and the mediating role of hedonic and utilitarian values	Vieira, Rafael & Agnihotri [4]	2022	Journal of Business Research
83	Augment yourself through virtual mirror: The impact of self-viewing and narcissism on consumer responses	Baek, Yoo & Yoon [158]	2018	International Journal of Advertising
84	The effect of augmented reality experience on loyalty and purchasing intent: An application on the retail sector	Eru, Topuz & Cop [159]	2022	Sosyoekonomi
85	Technologically empowered? Perception and acceptance of AR glasses and 3D printers in new generations of consumers	Ponzoa et al. [160]	2021	Technological Forecasting and Social Change
86	Beyond presence: Creating attractive online retailing stores through the cool AR technology	Zhang et al. [161]	2023	International Journal of Consumer Studies
87	The effect of augmented reality in mobile applications on consumers’ online impulse purchase intention: The mediating role of perceived value	Trivedi et al. [162]	2022	Journal of Consumer Behavior
88	Can consumers’ gamified, personalized, and engaging experiences with VR fashion apps increase in-app purchase intention by fulfilling needs?	Oiyan Lau & Chung-Wha (Chloe) Ki [163]	2021	Fashion and Textiles
89	WOW, the make-up AR app is impressive: A comparative study between China and South Korea	Butt et al. [164]	2022	Journal of Services Marketing
90	Branding in the time of virtual reality: Are virtual store brand perceptions real?	Pizzi, Vannucci & Aiello [165]	2020	Journal of Business Research
91	An analysis of the impact of personality traits towards augmented reality in online shopping	Lixăndroiu et al. [166]	2021	Symmetry
92	As if the product is already mine: How augmented reality improves the digital product presentation	Alt, Esch & Krause [167]	2020	Transfer: Zeitschrift für Kommunikation & Markenmanagement
93	Innovation and promotion activities in the internet to increase sales volume of music product using augmented reality technology	Kusumawati, Purnamasari & Sardiyo [168]	2013	Egitania Sciencia
94	Consumer behavior in an augmented reality environment: Exploring the effects of flow via augmented realism and technology fluidity	Chen & Lin [169]	2022	Telematics and Informatics
95	Value-based adoption of augmented reality: A study on the influence on online purchase intention in retail	Erdmann, Mas & Arilla [170]	2021	Journal of Consumer Behavior
96	Effects of diffusion of innovations, spatial presence, and flow on virtual reality shopping	Lu & Hsiao [171]	2022	Frontiers in Psychology
97	Effects of mobile augmented reality apps on impulse buying behavior: An investigation in the tourism field	Do, Shih & Ha [172]	2020	Heliyon
98	When brands come to life: Experimental research on the vividness effect of virtual reality in transformational marketing communications	Kerrebroeck, Brengman & Willems [173]	2017	Virtual Reality
99	The virtual reality leisure activities experience on elderly people	Jeng, Pai & Yeh [174]	2017	Applied Research in Quality of Life
100	Feeling the service product closer: Triggering visit intention via virtual reality	Yuce et al. [175]	2020	Sustainability (Switzerland)
101	The power of affection: Exploring the key drivers of customer loyalty in virtual reality-enabled services	Yan et al. [176]	2022	Frontiers in Psychology
102	An empirical study on the impact of online travel consumers’ brand loyalty: The mediating effect of flow experience	Xie & Yuan [177]	2021	E3S Web of Conferences
103	The impact of augmented reality on overall service satisfaction in elaborate servicescapes	Gäthke [178]	2020	Journal of Service Management
104	The effects of perceived quality of augmented reality in mobile commerce—an application of the information systems success model	Yoo [179]	2020	Informatics
105	Consumers’ self-congruence with a “Liked” brand	Wallace, Buil & de Chernatony [180]	2017	European Journal of Marketing
106	Influence of augmented reality product display on consumers’ product attitudes: A product uncertainty reduction perspective	Sun et al. [181]	2022	Journal of Retailing and Consumer Services
107	A new reality: Fan perceptions of augmented reality readiness in sport marketing	Goebert & Greenhalgh [182]	2020	Computers in Human Behavior
108	Augmented reality marketing: How mobile AR-apps can improve brands through inspiration	Rauschnabel, Felix & Hinsch [183]	2019	Journal of Retailing and Consumer Services
109	The impact of “e-atmospherics” on physical stores	Poncin & Ben Mimoun [184]	2014	Journal of Retailing and Consumer Services
110	A new reality: Exploring continuance intention to use mobile augmented reality for entertainment purposes	Hung, Chang & Ma [185]	2021	Technology in Society
111	On m-commerce adoption and augmented reality: A study on apparel buying using m-commerce in Indian context	Manchanda & Deb [186]	2021	Journal of Internet Commerce

in Appendix A.

3.2. Data Coding

The minimum required data for MASEM include the sample size and correlation coefficient between two variables across independent studies [18,69] and each moderator variable in each sample for the moderator analysis [17]. Our coding process was adapted from Luceri et al. [3]. The coding process can be found in

Table A4. Data coding.

General Topic	Specific Coding Items	Notes
General study information	Paper ID
	Title
	Author
	Publication year
	Journal/Proceedings Journal ranking	Referring to ABS and ABDC ranking list
Dataset characteristics	Sample size Product types Sample types
Model characteristics	Independent variable (IV)	Corresponding to the construct
	Dependent variable (DV)	Corresponding to the construct
Effect size information	Effect size	Pearson’s product-moment correlation coefficient $r$ [3]; standardized regression $\left(\beta \right)$ coefficients [85]
	Page number	The page number of effect size

in Appendix A. Two independent coders, consisting of an author and a research assistant, individually coded the 30 previously selected papers. To assess the intercoder reliability, we calculated Cohen’s kappa coefficient as suggested by Schamp et al. [76]. The obtained coefficient was $0.85>0.80$, indicating an almost perfect strength of agreement, and the $p \mathrm{v}\mathrm{a}\mathrm{l}\mathrm{u}\mathrm{e}=0.000<0.05,$ further confirms the strong reliability of the measurement [76,83]. After resolving any discrepancies, the first coder (i.e., the author) completed the coding for the remaining papers independently.

The effect size is a quantitative measure utilized to address a specific problem of interest [84]. It serves as an indicator of the strength of the relationship between two variables. In the MA, the data employed consists of summary statistics, specifically the effect size statistics, reported in each individual study, rather than the raw data [11]. The magnitude of the effect size varies depending on the variables involved [11]. Among the various effect size measures employed by researchers, the correlation coefficient stands as one of the most commonly used [18], such as Pearson’s product–moment correlation coefficient, denoted as $r$ (e.g., [3]), and standardized regression coefficients, symbolized as $\beta$ (e.g., [85]). We regarded $r$, as recommended by Luceri et al. [3] and Vieira et al. [4], as the appropriate effect size measure. In cases where studies did not report Pearson’s $r$ matrices, we transformed the $\beta$ coefficients into Pearson’s $r$ via the following conversion formula: $r=\beta +0.05\lambda ⟺r=\left\{\begin{array}{c}\beta +0.05, \beta \ge 0\\ \beta , \beta <0\end{array}\right.$ (where $\lambda =1$ if $\beta$ is nonnegative and $\lambda =0$ if $\beta$ is negative) [3,85]. Moreover, some studies (e.g., [3,18]) transferred Pearson’s $r$ to Fisher’s $z$ prior to the analysis. This transformation is primarily employed to correct for nonlinearity in Pearson correlation coefficients and to normalize their sample distribution [18]. However, applying Fisher’s $z$ transformation to the raw correlation coefficients may introduce some unforeseen implications in their interpretation [86]. The disparity between $r$ and $z$ is minimal and does not sufficiently support the advantages of Fisher’s $z$ transformations [86]. Schulze [11,86] recommended $r$ over $z$ when assuming a random effect model. Hence, our study employed the original Pearson’s $r$ for model fitting, as we adopted the random effect model. Additionally, in a correlation matrix, the ideal number of bivariate correlations among $p$ variables is given by $p\times (p-1)/2$ [17]. In our study, we analyzed 14 variables, so ideally, we should code 91 bivariate correlations between the constructs.

3.3. Data Analysis

In the process of data analysis, we utilized the metaSEM package in R Studio version 4.1.3, which is a user-friendly tool created by Cheung [18] and is specifically designed for conducting a MASEM study.

To examine the relationships between variables across a range of independent studies, we utilized a two-stage MASEM (TSSEM), which is a random effects technique [17]. In the first stage, we generated a pooled correlation coefficient matrix [11] as a result of the MA, which served as the input of the subsequent stage, where we fitted the SEM [11,83].

To account for heterogeneity, we explored the analysis of study-level variables (i.e., subgroup, moderator). While it is advantageous to explicitly test these study-level moderators, it is important to note that some subgroups may have a limited number of studies [11]. In our study, we considered five moderators, as mentioned previously. However, it is worth mentioning that the metaSEM package in R Studio can handle only numeric binary variables. Therefore, we encoded the study-level variables as zero-one dummy variables.

4. Results

This section presents the findings of the two-stage MASEM analysis and the examination of the moderating effects. We obtained a total of 136 datasets from 111 studies, resulting in 1099 observed effect sizes. On average, each dataset (i.e., study) contributed eight effect sizes, ranging from 2 to 36. The overall sample size aggregated across all the studies was 547,415, representing the cumulative number of individual participants or observations from the 111 studies that contributed to the observed effect size. The sample size of this meta-analysis surpasses the minimum sample size requirement for SEM (i.e., exceeding 200), thereby ensuring the robustness and reliability of our evidence-based results compared with those of the individual studies. Furthermore, all bivariate relationships encompassed at least one effect size, fulfilling the essential prerequisite for MASEM. Additionally, the total sample sizes for each of the proposed bivariate correlations also met the aforementioned minimum requirements. The total sample sizes and effect sizes per proposed bivariate correlation can be found in

Table A7. Total sample size and number of effect size per bivariate correlation.

	CX	ATT	INT	LOY	PSQ	PRE	PN	PEU	PU	INF	PE	HV	PR	UV
CX	CX	13	20	3	16	14	13	8	7	9	12	7	6	6
ATT	8931	ATT	53	7	29	33	15	20	17	24	32	9	12	8
INT	515,329	523,534	INT	1	38	35	13	19	26	34	38	13	15	10
LOY	893	3708	338	LOY	4	4	1	1	1	2	1	1	3	1
PSQ	8474	13,329	12,697	3149	PSQ	27	14	15	16	21	20	9	5	6
PRE	5744	11,993	10,649	3060	9083	PRE	7	11	11	22	23	6	9	5
PN	3768	4806	4372	338	3478	3337	PN	8	5	13	13	4	5	3
PEU	2694	5944	5504	266	2745	4730	2636	PEU	26	12	17	2	2	1
PU	2490	4704	7316	266	3330	4345	2339	7264	PU	14	21	1	4	1
INF	2511	5827	7564	757	5162	6547	3593	2629	3386	INF	23	9	5	5
PE	9089	15256	16,051	319	10,479	8653	5021	6682	7093	6572	PE	2	8	2
HV	507,453	507,518	3500	338	1897	2541	988	307	253	2486	357	HV	3	7
PR	1835	5566	4881	2616	3169	5548	2555	1470	1750	1070	3644	979	PR	2
UV	507,308	507,326	2351	338	1387	1858	843	162	253	1251	285	2091	949	UV

Note: The value in the lower triangle matrix represents the total sample size per bivariate relationship. The value in the upper triangle matrix represents the number of effect size (i.e., the correlation coefficient) per bivariate relationship. CX = consumer experience, ATT = attitude, INT = intention, LOY = loyalty, PSQ = perceived system quality, PRE = presence, PN = perceived novelty, PEU = perceived ease of use, PU = perceived usefulness, INF = information, PE = perceived enjoyment, HV = hedonic value, PR = perceived risks, UV = utilitarian value.

in Appendix A.

4.1. TSSEM Stage 1 Results

In the first stage of the TSSEM, we generated a pooled correlation matrix of the 14 constructs. The results of the correlational analysis and generated pooled correlation matrix can be found in

Table A8. Pooled correlation matrix between bivariate constructs based on random effects model.

	CX	ATT	INT	LOY	PSQ	PRE	PEU	PU	INF	PE	PN	HV	PR	UV
CX	1
ATT	0.468	1
INT	0.356	0.626	1
LOY	0.346	0.424	0.599	1
PSQ	0.540	0.415	0.411	0.459	1
PRE	0.436	0.515	0.411	0.465	0.474	1
PEU	0.319	0.346	0.364	0.506	0.434	0.419	1
PU	0.396	0.401	0.322	0.588	0.476	0.482	0.396	1
INF	0.465	0.637	0.467	0.557	0.530	0.482	0.338	0.547	1
PE	0.471	0.494	0.501	0.607	0.458	0.446	0.422	0.373	0.652	1
PN	0.583	0.574	0.466	0.229	0.582	0.499	0.398	0.493	0.599	0.554	1
HV	0.536	0.485	0.450	0.435	0.536	0.484	0.597	0.537	0.285	0.456	0.563	1
PR	0.146	0.146	0.270	0.453	0.538	0.296	0.141	0.330	0.351	0.302	0.227	0.060	1
UV	0.490	0.499	0.535	0.436	0.542	0.567	0.591	0.529	0.287	0.449	0.387	0.603	0.015	1

Notes: CX = consumer experience, ATT = attitude, INT = intention, LOY = loyalty, PSQ = perceived system quality, PRE = presence, PN = perceived novelty, PEU = perceived ease of use, PU = perceived usefulness, INF = information, PE = perceived enjoyment, HV = hedonic value, PR = perceived risks, UV = utilitarian value.

in Appendix A.

Table 1. MASEM stage 1 random effect results.

Bivariate Correlation	K	Cumulative N	r	Estimate	S.E.	95% CI		Z-Value	Significance	df.	I2 (Q Statistic)	FsN
						Lower	Higher
CX-ATT	13	8931	0.460	0.468	0.053	0.364	0.571	8.844	***	12	0.980	4871
ATT-INT	53	523,534	0.627	0.626	0.022	0.583	0.670	28.076	***	52	0.956	818,229
INT-LOY	1	338	0.610	0.599	0.034	0.532	0.665	17.723	***	0	0.000	115
CX-LOY	3	893	0.346	0.346	0.131	0.090	0.602	2.646	**	2	0.974	20
PSQ-CX	16	8474	0.544	0.540	0.061	0.421	0.660	8.841	***	15	0.990	7378
PRE-CX	14	5744	0.431	0.436	0.045	0.348	0.525	9.662	***	13	0.960	6747
PN-CX	12	3768	0.314	0.319	0.084	0.155	0.482	3.814	***	11	0.984	762
PEU-ATT	20	5944	0.401	0.401	0.037	0.330	0.473	10.960	***	19	0.955	17,736
PEU-PU	26	7264	0.543	0.547	0.035	0.479	0.615	15.756	***	25	0.972	61,993
PU-ATT	17	4704	0.638	0.637	0.044	0.551	0.723	14.526	***	16	0.978	22,518
INF-ATT	24	5827	0.503	0.494	0.039	0.419	0.570	12.792	***	23	0.971	34,807
PE-INT	38	16,051	0.472	0.466	0.035	0.397	0.535	13.267	***	37	0.979	93,887
HV-INT	13	3500	0.452	0.450	0.060	0.333	0.568	7.538	***	12	0.980	3535
PR-INT	15	4881	0.262	0.270	0.072	0.130	0.410	3.771	***	14	0.981	1167
UV-INT	10	2351	0.536	0.535	0.058	0.422	0.648	9.265	***	9	0.975	3162

Significant codes: ‘***’ < 0.001, ‘**’ < 0.05. K = number of effect size, cumulative N = cumulative samples size, r = average Pearson correlation coefficient, S.E. = estimation of standard errors, df. = degree of freedom, Q = Cochrane’s Q.

displays an overview of the dataset we collected from the existing literature suggested by Luceri et al. [3], including the bivariate relationships, the number of datasets that reported the corresponding relationships (K), and the cumulative sample size corresponding to the relationships (cumulative N). Since there are no variables that are independent of each other, indicating the absence of explanatory variables in this study, we computed the average effect sizes (correlations) for each bivariate relationship. We also calculated the average Pearson correlation coefficient ($r$), the estimation with standard errors (S.E.), and the degree of freedom ($df.=K-1$). Notably, the number of datasets (K) ranges from 1 (INT-LOY) to 53 (ATT-INT), with an average of 15, and only two bivariate relationships (CX-LOY and INT-LOY) have fewer than 10. Importantly, the cumulative sample sizes for all 15 bivariate relationships met the minimum sample size requirement for SEM (i.e., exceeding 200).

Additionally, we calculated the 95% confidence interval (CI) and the z value. The confidence intervals for all the correlations were greater than zero, indicating that the effects of the proposed correlations were significantly different from zero, as suggested by Zhao et al. [77]. The results indicated that all the correlations were significant and positive at the $p<0.05$ level, indicating significant and positive relationships between the proposed constructs. However, the correlations for CX-LOY, PN-CX, and PR-INT were relatively low. In contrast, the majority of bivariate correlations ranged from 0.400 to 0.700, indicating a moderately strong correlation between the antecedents and consequences. Among the antecedent constructs, perceived system quality had the highest correlation with customer experience, perceived usefulness had the strongest correlation with attitude, and utilitarian value had the most significant correlation with intentions, with $r>0.4$ and $p<0.001$, as suggested by Luceri et al. [3].

In addition, we used Cochrane’s Q test to calculate the degree of heterogeneity $I^2$ for the heterogeneity test. The general result showed that the Q statistic for the homogeneity of effect sizes was 28,245.65 ($df.=1008,$ p-value = 0 < 0.05). A significant Q value indicates that there is indeed heterogeneity in the correlation matrix [11]. Specifically, in addition to the correlation of INT-LOY (i.e., $I^2=0$), which has only one effect size, all other bivariate relationships showed substantial heterogeneity, i.e., greater than 90%. A large portion of the variance was at the study level for all the correlations [11]. Hence, as Luceri et al. [3] and Jak [11] suggested, the examination of study-level variables (i.e., moderators) was considered for all bivariate correlations with heterogeneity greater than 90%. Moreover, according to Rosenthal [74], file drawer problems exist in which only 5% of the significant results (e.g., $p<0.05$) with type I errors have been published in journals, whereas 95% of the nonsignificant results (e.g., $p>0.05$) remain in the file drawer. Hence, we also calculated the fail-safe number (FsN) to test for publication bias (e.g., [3,4,74]). The fail-safe N (FsN) is calculated as follows: $FsN={\left({\displaystyle \frac{\sum \overline{Z}}{1.645}}\right)}^2-k \left(\alpha <0.05\right)$, where $\alpha$ is the significance level [74]. As the FsN of almost all (excluding CX-INT) correlations were much larger than $5k+10$, the relationships passed the tolerance level suggested by Rosenthal [74], which means that there is little or no publication bias in the collected literature [3].

4.2. TSSEM Stage 2 Results

In the second stage, we fitted the structural equation model.

Table 2. MASEM stage 2 random effect.

			95% CI
Hypothesis	Path	Estimate	Lower	Higher
1	CX→ATT	0.683	0.622	0.747	Support
2	ATT→INT	0.675	0.639	0.712	Support
3	INT→LOY	0.646	0.571	0.724	Support
4	CX→LOY	0.132	0.025	0.236	Support
5	PSQ→CX	0.623	0.545	0.703	Support
6	PRE→CX	0.604	0.544	0.665	Support
7	PN→CX	0.584	0.490	0.680	Support
8	PEU→ATT	0.147	0.057	0.230	Support
9	PEU→PU	0.451	0.383	0.519	Support
10	PU→ATT	0.554	0.464	0.647	Support
11	INF→ATT	0.546	0.486	0.606	Support
12	PE→INT	0.402	0.338	0.467	Support
13	HV→INT	0.467	0.383	0.553	Support
14	PR→INT	0.405	0.292	0.519	Not Support
15	UV→INT	0.511	0.427	0.596	Support

presents an overview of the TSSEM stage 2 results. The OpenMx status1 was “0”, indicating that the model optimization was satisfactory. The goodness-of-fit indices suggested that the model had a moderate fit compared with previous meta-analyses. First, as the 95% CI of all the parameters (effects) did not include zero, the parameters estimated by the A-matrix and S-matrix were significantly different from zero [11]. Second, the chi-square ${\chi }^2 (df=76)=4683.00$ with $\mathit{p-}\mathrm{v}\mathrm{a}\mathrm{l}\mathrm{u}\mathrm{e}=0.000<0.05$, which means that the exact fit was rejected [11]. The chi-square/degree of freedom was not in the range of [1,3], as ${\chi }^2/df=61.61$ is larger than 3. However, compared with previous meta-analyses (e.g., Hogreve et al. [87]: ${\chi }^2 (df=25)=2819, {\chi }^2/df=112.7608$), our model showed a moderately good fit according to these two indices. Moreover, the similarity indices, such as the comparative fit index (CFI) and Tucker–Lewis index (TLI, also known as the nonnormed fit index, NNFI) [88] (CFI: 0.392, TLI: 0.272), were less than 0.9, which implies that the model fit was not good. However, the root mean square error of approximation (RMSEA) was 0.011 (95% CI [0.010, 0.011]), suggesting a close approximate fit, as suggested by Jak [11].

Generally, all of the effects on the fifteen paths were significant. However, the direct effect of perceived risk on customer intention contradicted our initial hypothesis, which requires further explanation. The potential explanation for the differences in results can be the combination of the conceptual framework and methodology choices [7]. First, the meta-analysis we used aimed to synthesize the varying empirical results obtained by scholars who approached the research question from different perspectives and with different objectives. Meta-analyses aggregate effect sizes from multiple studies to provide a general understanding of the proposed correlations. Second, the customer is experience orientated. As AR or VR is an advanced technology, it initially attracts a younger demographic and video game players, leading to a unique and novel experience for these early adopters.

Figure 4 shows the path model with 95% confidence intervals for the proposed conceptual framework.

4.3. Moderate Effect Analysis

In this study, we also analyzed five potential moderating effects of the proposed correlation with more than ten effect sizes (i.e., excluding CX-LOY and INT-LOY). A summary of the moderating effects can be found in

Table A9. Moderator analysis.

	No Moderator		Journal Ranking		Tech Type		Product Type		Sample Type		Year
	Estimate (S.E.)	Pr (>|z|)	Estimate (S.E.)	Pr (>|z|)	Estimate (S.E.)	Pr (>|z|)	Estimate (S.E.)	Pr (>|z|)	Estimate (S.E.)	Pr (>|z|)	Estimate (S.E.)	Pr (>|z|)
CX→ATT	0.762 (0.073)	0.000	−0.137 (0.112)	0.221	−0.451 (0.023)	0.000	−0.121 (0.143)	0.397	−0.267 (130.833)	0.998	−0.204 (102.542)	0.998
ATT→INT	0.677 (0.022)	0.000	0.41 (0.147)	0.005	0.304 (0.132)	0.021	0.796 (0.158)	0.000	0.288 (144.516)	0.998	0.085 (77.176)	0.999
PSQ→CX	0.62 (0.055)	0.000	0.108 (0.082)	0.187	0.955 (0.055)	0.000	0.322 (0.144)	0.025	0.059 (195.702)	1.000	0.141 (65.362)	0.998
PRE→CX	0.655 (0.07)	0.000	−0.073 (0.066)	0.270	1.19 (0.064)	0.000	0.056 (0.133)	0.672	0.186 (127.551)	0.999	−0.041 (80.814)	1.000
PN→CX	0.444 (0.071)	0.000	0.084 (0.152)	0.579	−0.58 (0.127)	0.000	0.341 (0.111)	0.002	−0.309 (309.121)	0.999	0.473 (34.468)	0.989
PEU→ATT	0.098 (0.058)	0.090	0.511 (0.124)	0.000	0.609 (NA)	NA	0.941 (0.082)	0.000	0.42 (50.523)	0.993	0.782 (63.772)	0.990
PEU→PU	0.533 (0.036)	0.000	−0.582 (0.126)	0.000	−0.198 (0.03)	0.000	−0.394 (0.106)	0.000	0.774 (106.974)	0.994	−0.638 (123.16)	0.996
PU→ATT	0.588 (0.061)	0.000	0.051 (0.148)	0.732	0.901 (0.018)	0.000	0.806 (0.108)	0.000	−0.199 (175.597)	0.999	0.149 (152.502)	0.999
INF→ATT	0.588 (0.045)	0.000	0.413 (0.119)	0.001	0.581 (0.034)	0.000	0.831 (0.085)	0.000	1.023 (245.331)	0.997	0.545 (63.016)	0.993
PE→INT	0.461 (0.036)	0.000	0.806 (0.1)	0.000	1.051 (0.053)	0.000	0.985 (0.068)	0.000	0.501 (152.132)	0.997	1.019 (24.268)	0.967
HV→INT	0.444 (0.06)	0.000	0.934 (0.103)	0.000	0.9 (0.074)	0.000	1.091 (0.139)	0.000	−0.142 (133.209)	0.999	0.418 (36.772)	0.991
PR→INT	0.26 (0.073)	0.000	0.259 (0.108)	0.017	0.658 (0.105)	0.000	0.549 (0.095)	0.000	0.506 (189.361)	0.998	0.746 (26.741)	0.978
UV→INT	0.526 (0.058)	0.000	0.982 (0.17)	0.000	1.072 (0.08)	0.000	0.949 (0.1)	0.000	−0.498 (96.56)	0.996	0.417 (40.114)	0.992

Notes: S.E. = standard error, CX = consumer experience, ATT = attitude, INT = intention, PSQ = perceived system quality, PRE = presence, PN = perceived novelty, PEU = perceived ease of use, PU = perceived usefulness, INF = information, PE = perceived enjoyment, HV = hedonic value, PR = perceived risks, UV = utilitarian value.

in Appendix A.

The results for the no moderator model were significant for all thirteen paths, as the p-value was much less than 0.05. Moreover, three moderators had significant moderating effects—technology types, journal rankings, and product types—which indicates significant variation in the bivariate correlation under different contextual conditions. First, the technology type was the most influential moderator, as only one bivariate correlation was nonsignificant. Moreover, journal rankings and product types also had significant moderating effects on some proposed bivariate relationships. However, no significant differences were found for the different product types and publication years on the basis of the bivariate correlations. Our result differs from that of Luceri et al. [3] in that the publication year had nonsignificant moderating effects on all the proposed correlations. The potential reason is that the research and publication processes take a particular amount of time, but the division is based on the publication year (i.e., the first year of the metaverse, 2021), which ignores the research and review period. Hence, the publication year does not significantly affect bivariate relationships.

Overall, the results in this chapter indicate a significant and positive association between the proposed bivariate correlations and the moderating effects of some proposed correlations.

5. Conclusions

The primary objective of this study was to propose and evaluate a comprehensive conceptual framework of the customer journey in retailing, specifically through its initial application of AR and VR. To accomplish this aim, we employed meta-analytic structural equation modeling to integrate the current empirical research on customer behavior using AR and VR in this field. Interestingly, our results validate the positive associations among all the proposed correlations, providing support for most hypotheses with the exception of one. Notably, the consequence-to-consequence correlations (e.g., CX→ATT, ATT→INT, and INT→LOY) were found to be stronger than the antecedent-to-consequence (e.g., PSQ→CX, INF→ATT, and PR→INT) or antecedent-to-antecedent (e.g., PEU→PU) correlations. This suggests a robust connection within the customer journey. Additionally, customer experience was found to have a significant and positive influence on attitude, intention, and loyalty within a linear customer journey but a relatively weaker direct effect on loyalty within a nonlinear customer journey. Furthermore, we observed significant moderating effects of journal ranking, technology type, and product type. However, the sample type and publication year were found to have nonsignificant moderating effects on the proposed correlations. We have summarized the key findings and their implications in

Table 3. Summary of key findings and implications.

Key Findings	Managerial Implications
Antecedents
All the proposed correlations are significantly and positively correlated.	Retailers should consider the factors that influence customer decision-making.
The hedonic and utilitarian values have a significant and positive impact on customers’ intentions, with utilitarian values exerting a stronger influence than hedonic values.	It is crucial for retailers to consider the hedonic values and perceived enjoyment of their customers. However, rather than solely emphasizing one type of value, retailers should place equal emphasis on both hedonic and utilitarian values in order to attract a larger customer base and optimize profitability.
Perceived system quality has the greatest impact on customer experience.	To enhance customer experience, retailers should employ atmospherics to recreate the atmospherics of a physical shopping environment and create an immersive virtual shopping experience.
Informativeness and perceived usefulness are critical factors that affect customers’ attitudes.	Retailers can utilize WOM marketing, advertising, brand strategies, and interactive product design to enhance customers’ perception of product informativeness in AR- or VR-supported retailing.
The perceived risk significantly and positively affects the intention.	New or advanced technology brings both challenges and opportunities for retailers. It offers a unique and innovative experience for customers. Therefore, it is crucial for retailers to effectively navigate these challenges and transform them into opportunities to maximize their profits.
The perceived presence of customers plays a vital role in AR/VR retailing.	Retailers should develop interactive advertising and distribution strategies that aim to encourage and engage more users to actively participate in AR/VR retailing.
Consequences
The customer experience significantly and positively affects attitude, purchasing intentions, and loyalty (linear customer journey). Customer experience significantly and positively affects loyalty (non-linear customer journey).	Retailers should create experiences that facilitate turning linear customer journeys into non-linear ones.
Moderators
Journal ranking, technology, and product types show significant moderating effects.	Retailers should set different marketing strategies for different technology and product types.
Sample type and publication years have non-significant moderating effects on the proposed correlations.	In light of the global ageing trend, retailers should also consider digital inclusion for the oldest to maximize profits.

and present them in more detail in the subsequent subsections.

5.1. Theoretical Contributions

From a theoretical perspective, this study contributes to the literature by providing a comprehensive framework to examine the bivariate correlations among antecedents, consequences, and potential moderating factors in the domain of AR and VR retailing. More specifically, this study provides acceptable findings by synthesizing and generating empirical results from previous studies in this field and contributes to the field in three ways.

First, drawing on established and widely cited theories such as the TAM and UTAUT in the domain of AR and VR marketing, we examined the key determinants shaping customer shopping behavior. Our findings reveal significant positive correlations among the proposed variables. Notably, PU and PEU demonstrate robust and positive associations with attitude, which is consistent with the TAM as posited by Davis [65] and corroborated by recent research [3]. Moreover, this study identifies pivotal factors driving customer behavior within AR and VR shopping experiences.

Second, this research contributes valuable insights into the contextual boundaries of AR and VR retailing by expanding the range of moderating influences. Our study represents a more reliable and evidence-based methodology than single quantitative approaches do. Compared with the first meta-analysis conducted that focused solely on investigating customer behavior via AR (i.e., [4]), our study expands the scope to encompass more generic scenarios involving both AR and VR.

Finally, this study provides evidence by shedding new light on a systematic understanding of how retailers can develop comprehensive marketing strategies in AR and VR retailing. By generating and synthesizing the fragment and inconsistent empirical findings from the existing literature, we offer a systematic understanding of how retailers make comprehensive marketing decisions. Moreover, our research extends beyond conventional linear customer journeys and explores the complexities of nonlinear customer journeys in retailing, leveraging the AR and VR technologies. Furthermore, the comprehensive framework we propose serves as a foundation for future research, enabling scholars to explore the complex dynamics and relationships. In conclusion, we will outline the managerial implications for retailers and propose a future research agenda to further advance the knowledge in this field.

5.2. Managerial Implications

This meta-analysis delves into the crucial factors of customer purchasing and their impact on customer decision-making. From a managerial perspective, this study also provides managerial implications for AR, VR, and even metaverse retailers in the forthcoming era of metaverse retailing based on AR and VR retailing. Considering that AR and VR technologies offer different experiences for the consumer to enhance their engagement and satisfaction in the customer journey, it is important to understand how they can influence consumer decisions in different ways. While both technologies contribute to immersive experiences, AR enhances real-world environments by overlaying the virtual object in the natural environment. At the same time, VR provides an immersive 3D virtual space for consumers, creating a more isolated and immersive experience. Given the nascent stage of the metaverse, our study examines the retailing activities facilitated by AR and VR technologies. Consequently, we offer valuable insights and practical recommendations for practitioners and retailers involved in metaverse-related activities.

In today’s marketing environment, especially in online retail or e-commerce, ensuring consistency with current practices is critical. An enhanced customer experience has been shown to significantly influence attitudes, intentions, and loyalty throughout the customer journey. Therefore, retailers must prioritize factors that facilitate purchasing decisions, emphasizing those antecedents that yield the most significant effects on desired outcomes. Our research reveals that all identified antecedent constructs exert a significant and positive influence on the customer decision process. Notably, constructs such as perceived system quality, perceived informativeness, perceived usefulness, and utilitarian value significantly affect customer experience, attitudes, and intentions.

First, perceived system quality has the most substantial effect on customer experience, indicating the pivotal role of the perceived quality of AR and VR in influencing customers’ decision process and perceived value. As the metaverse offers considerable opportunities for companies to cultivate customer relationships [89], retailers are supposed to employ atmospheric elements to craft engaging shopping experiences and provide an immersive virtual retail environment. For example, in VR retailing, the distinctive atmospheric store design possesses a unique attribute that can generate added value for customers and enhance their overall shopping experience. While AR can enhance physical store retailing or e-commerce, the virtual store created by VR requires a fully immersive approach to appeal more to those consumers seeking novelty and escapism so that the perceived system quality and atmospheric store design have added value in VR retailing. Through such value-added activities, retailers can gain a competitive edge by augmenting customers’ perceived value, consequently fostering higher purchase intentions and brand loyalty and facilitating long-term customer–vendor relationships [44]. For example, while advertisements are integral to the metaverse experience, users may experience conflicting emotions [90]. Leveraging the interaction between avatars and multiple layers of sensorimotor feedback in metaverse advertisements creates a profound sense of presence, which is preferred by customers over traditional formats [91]. Furthermore, metaverse retailers and brand managers can harness AR interactive technology to cultivate brand affinity and deliver practical virtual experiences within metaverse retail environments.

Second, perceived usefulness emerges as the primary driver of purchase intentions, surpassing the role of ease of use, which aligns with Davis’s [65] research. Therefore, metaverse retailers should focus on designing a shopping context that offers clear utility and value to customers. Additionally, the informativeness of the retail experience has a substantial effect on customers’ attitudes. Customers often rely on mental imagery to compensate for the lack of physical product information within AR- or VR-supported retail environments [66]. Leveraging word of mouth (WOM), advertising, brand reputation, and product experiences can enhance perceived product information, enriching the overall retail experience within the metaverse. Considering the unique characteristics of AR and VR, the way information is presented in these environments can be very different. For example, AR allows customers to visualize a product in a realistic environment, while VR can provide an immersive demonstration of a product’s functionality. Moreover, advertisements are essential to the experience, but users may have strongly contradictory emotions [90]. Metaverse advertisements leverage the interaction between avatars and multiple layers of sensorimotor feedback to create a sense of presence, and customers prefer these formats over traditional formats [91]. Furthermore, metaverse retailers and brand managers can leverage AR interactive technology to foster brand love and provide consumers with practical virtual experiences within metaverse retailing settings.

Third, our findings indicate that both hedonic and utilitarian values significantly influence customers’ intentions, with utilitarian values exerting a comparatively stronger impact than hedonic values do. Given that the metaverse primarily revolves around entertainment, retailers should prioritize addressing customers’ hedonic desires and crafting enjoyable shopping experiences. However, it is essential for retailers not to underestimate the significance of utilitarian values [3]. By comprehending and addressing both hedonic and utilitarian value, retailers can attract a larger customer base and optimize profitability.

Interestingly, contrary to our initial hypothesis, perceived risk was found to positively affect customer purchasing intention. This intriguing finding may be attributed to the novel experience that customers have with innovative retailing technologies, as well as the specific characteristics of our sample, which predominantly consisted of students who are more inclined to embrace and explore new experiences. Moreover, while new or advanced technologies present challenges for retailers, they also offer unique opportunities. It is imperative for retailers to address these challenges and harness the potential of innovative technologies to maximize profits.

Additionally, moderators such as product and technology types significantly influence the customer shopping journey. This suggests that the impact of AR, VR, or other metaverse technologies on customer behavior varies depending on the product being sold. Consequently, metaverse retailers must develop tailored marketing strategies that accommodate these differences. However, the sample type and publication year had nonsignificant moderating effects on the proposed correlations. Therefore, metaverse retailers should consider both younger and older customers, as they exhibit distinct preferences and biases within the metaverse retailing context, ensuring digital inclusion and maximizing profits amid the global aging trend.

Finally, it is important to note that the metaverse is still in its nascent stage, and the technology is not yet fully developed. Many organizations are leveraging the concept of the metaverse as a marketing tool to attract customers. However, customers may have high expectations of the metaverse but may only experience its primary applications. Customer satisfaction hinges on whether their expectations are met, and perceived value is determined by the disparity between the perceived product quality, service, and the price paid. Dissatisfaction among customers could lead to a decline in purchasing intention. Furthermore, our study delves into the nonlinear customer journey discussed by Grewal and Roggeveen [50]. Our results corroborate the significant and positive effect of customer experience on loyalty, underscoring the importance for metaverse retailers to enhance customer experience to facilitate their decision-making process.

5.3. Limitations

This study aims to enhance the understanding of customer behavior in AR and VR retailing. However, it is essential to acknowledge several limitations. First, the meta-analysis heavily relies on the existing literature, limiting the analysis of the correlations proposed in previous research, thus constraining the depth and breadth of the framework. For example, the exclusion of constructs related to purchase behavior due to the limited existing literature represents a notable constraint because understanding the gap between customer intention and actual behavior is crucial for a comprehensive understanding of retailing. Moreover, there may be issues of selection and publication bias. Despite following standard procedures for literature selection, some biases may still exist. For example, meta-analyses depend heavily on published papers, which often prioritize significant results, thus leading to the potential for publication bias. Although this study made efforts to minimize publication bias during data collection, and the initial findings suggest minimal bias in the literature reviewed, it is important to acknowledge the possibility that some degree of publication bias might still influence the results. Therefore, future studies should address this limitation to yield findings that are more generalizable and robust.

5.4. Future Research Agenda

A research agenda focusing on investigating customer behavior is recommended, taking into account the limitations of this meta-analysis. For example, the effect sizes derived from the existing literature often lack exploration into purchase behavior and post-purchase loyalty. Future studies could delve deeper into these areas to better understand consumer decision-making processes.

Furthermore, the future of retailing seems poised toward metaverse retailing, which integrates AR, VR, and other interactive technologies to provide a seamless and immersive virtual experience. Despite its promising potential, the current immaturity and technical constraints of the metaverse present challenges in analyzing the customer journey within this novel retail landscape. Future research should explore advancements in metaverse technologies and their influence on consumer behavior. One approach is to integrate MacInnes’s [92] business model, which examines the disruptive effects of technological advancements across four stages: (i) technical, addressing technology acceptance, sensory marketing, and crowd-sourcing; (ii) environmental, focusing on policy and regulation in metaverse retailing; (iii) revenue, considering the maturity and profitability phases for retailers in the metaverse; and (iv) sustaining, which explores sustainability and expansion strategies in metaverse retailing.

Table 4. Research agenda based on the business model.

Stages	Items	Research Agenda
Stage 1: Technical	Technology acceptance	To investigate the different degrees of technology acceptance in different purchase stages (i.e., pre-purchase, purchase, and post-purchase) of metaverse retailing. To explore the customer decision process through mixed reality.
	Sensory marketing	To investigate the influence of sensory marketing on customer purchasing decisions in metaverse retailing for hedonic and utilitarian motivation.
	Crowdsourcing or co-creation	To explore whether co-creation in the metaverse would enhance the customers’ presence in metaverse retailing.
Stage 2: Environmental	Policy and regulation	To explore the property and privacy security in metaverse retailing. To investigate how metaverse retailers can achieve carbon neutrality when utilizing omnichannel distribution.
Stage 3: Revenue	Avatar	To examine the user’s preference for building their avatar’s shape and customer preference for AR or VR in metaverse retailing.
	Advertisement Branding	To explore the influence of falsity experience in the metaverse on customers’ purchasing decisions. To investigate how retailers build a brand and sustain a competitive advantage in metaverse retailing. To investigate how brand affects customer purchasing intention and the continuous purchasing intention in metaverse retailing.
	Geography	To investigate the influence on customer attractiveness by different geographical factors in the virtual and real world. To investigate the influence of the retailing cluster in metaverse retailing. To investigate the cross-impact of geographical factors in metaverse retailing for non-digital products from the perspective of omnichannel.
Stage 4: Sustaining	Omnichannel	To investigate the realization of the omnichannel in metaverse retailing. To investigate the realization of the seamless/smooth experience in omnichannel retailing to shorten the customer journey.
	WOM	To compare the word-of-mouth (WOM) difference between social media influencers in e-commerce and metaverse retailing and explore a sustainable way for WOM.
	Digital inclusion	To investigate how metaverse retailers realize digital inclusion for elders.

outlines the envisioned development of the metaverse and potential business opportunities for retail.