A Conceptual Approach to Understanding the Customer Experience in E-Commerce: An Empirical Study

Abstract

This study aimed to identify the constructs related to customer experience that underpin e-commerce, as well as their interconnections, to develop a comprehensive conceptual model based on theories-in-use. A quantitative approach was employed through a survey of 441 respondents. Data analysis was conducted using partial least squares structural equation modeling. The research findings revealed that there are a total of 11 constructs: customer experience, customer satisfaction, customer loyalty, word-of-mouth, trust, perceived risk, security and privacy, web content, perceived price, perceived value, and service quality. Furthermore, twelve relationships were established between these constructs, which led to the development of a holistic conceptual model. The identified constructs and the relationships between them are hierarchized, which has practical implications for businesses. It allows them to concentrate on operational activities and formulate and implement strategies that are valued by consumers and supported by empirical evidence. The originality and value of this research lie in the conception and development of a comprehensive e-commerce model, which includes eleven constructs and twelve relationships. It also highlights the pivotal role of the customer experience.

1. Introduction

E-commerce has experienced significant growth over the past few years. From 2017 to 2022, sales through e-commerce retail have almost doubled, from USD 1,624,327 million to USD 3,511,876 million [1], and e-commerce retail percentage sales are expected to grow from 7.4% to 23% of total retail sales from 2015 to 2027 [2]. In the year 2020, there was a significant increase in both time series due to the occurrence of the pandemic. The share of e-commerce retail sales in total retail sales increased by 4 percentage points, and the value of e-commerce retail sales increased by USD 654,075 million. In many locations, e-commerce was the only available sales channel for businesses due to the imposition of confinement and social distancing. To the question posed by Sheth [3], “Impact of COVID-19 on consumer behavior: Will the old habits return or die?” we have a conclusive response, one that reiterates that e-commerce retail sales are here to expand and, as a channel, have grabbed share from other channels. Admittedly, the pandemic disrupted consumer behavior, and this is one of the main reasons to rethink consumer behavior. However, other reasons are relevant as well. In particular, the prevalence of constructs tied to consumer behavior, of which customer experience (CX) is the most prominent [4].

Integrated into the marketing school of thought of consumer behavior [5], consumer behavior has historically been a relevant dimension of the marketing field. The 1960s saw the emergence of the first formal consumer behavior models (CBM) with names like Ernest Dichter, John Howard, George Katona, Engel, and Nicosia [5]. Friedman [6] claims that conceptual CBMs are neither obsolescent nor outdated. In his work, he presents five of them that continue to be indispensable references, precisely the models of (1) Andreasen, (2) Nicosia, (3) Howard–Sheth, (4) Engel, Kollat, and Blackwell, and (5) Bettman. There is no denying the relevance of the classical CBM, since they are the foundation of the newer ones, with the most common model being the five phases–need recognition, information search, evaluation of alternatives, purchase decision, and post-purchase behavior [7]. Some authors also indicate a broader scope of CBM by including references such as [8]: AIDA [9], the hierarchy of effects [10], and the hierarchy of sequence [11], among others.

With the advent of the Internet, there has been an evolution in the vast body of literature on consumer behavior. New terms have been introduced, most notably the terms customer journey [12], customer experience [13], and customer engagement [14]. Technology adoption and acceptance models have also been widely applied to e-commerce. Beginning with the work of Fishbein and Ajzen (Fishbein, 1963; Fishbein, 1967; Fishbein & Ajzen, 1975), research on this subject has a long history. There are several relevant models, namely the Theory of Reasoned Action (TRA) [15], the Theory of Planned Behavior (TPB) [16], the Reasoned Action Model [17], the Technology Acceptance Model (TAM) [18,19], and the Unified Theory of Acceptance and Use of Technology (UTAUT) [20,21]. Designed to explain acceptance and adoption of new technologies, the models average in predicting usage intentions, attitudes, or behaviors [21,22,23,24]. Its conceptual limitations nonetheless rest in determining relationships with the most important constructs, specifically customer satisfaction (CS), customer loyalty (CL), and word-of-mouth (WOM). The CS construct is particularly striking from the previous list because of the number of studies that have confirmed its positive correlation with business performance. Pooser and Browne [25] found that CS reduces costs and increases profitability; Yeung and Ennew [26] stated that CS can affect financial performance and previously found that CS has a positive financial effect, but the direct effects are generally small. De Mendonca and Zhou [27] claim CS positively impacts long-term corporate profitability. Hallowell [28] provided a testament to the theory that CS has a relationship with CL, which in turn has a relationship with profitability. From a different perspective, Ye et al. [29] concluded that intentionally lowering CS hurts long-term profitability. Other authors have highlighted the relationship between CS and CL and business performance while emphasizing that CL also influences business performance [30]. The impact of CS on financial business performance is provided in detail by Sun and Kim [31]. Using the Customer Satisfaction Index (CSI), they found that CS is reflected in Profit Margin (PM), Return on Assets (ROA), Return on Equity (ROE—proxies of a firm’s profitability), and Market Value Added (MVA—proxy of firm value). Slightly similar results, though less comprehensive, were also documented by Eklof et al. [32] and by Eklof et al. [33]. Also, Wetzels et al. [34] are proponents of the view that a CX strategy creates business value by ultimately driving revenue and profit, having proven this empirically. However, Guo et al. [35] contended that the relationship between CS and business performance is time-lagged, meaning past satisfaction has a positive effect on current profitability, and similarly, past profitability affects current satisfaction. Matzler et al. [36] likewise demonstrated the time-lag effect and confirmed the existence of a positive association between CS and shareholder value.

The aforementioned studies focused on the overall market, specific sectors, or the entire customer base of certain companies. Although fundamental, it is also pertinent to objectively analyze a company’s customer base regarding their satisfaction levels and evaluate its impact on business performance. There are two reasons for this to be true. First, there will always be variability in satisfaction levels. Second, the company will not be able to satisfy all of its customers.

Keiningham et al. [37] discovered a negative correlation between customer revenues and the profitability of unprofitable customers, while a positive correlation was observed for profitable customers. Before assigning resources to increase CS and share-of-wallet, the authors state that customers should first be segmented according to their profitability to the company. Similar and complementary results have also been discovered by Terpstra and Verbeeten [38] as well. They found that for the most profitable customer segments, higher customer satisfaction appears to lead to higher returns. Furthermore, CS is positively associated with future customer service costs and customer value. Last, the relationship between CS and customer value seems to be nonlinear. Helgesen [39]’s study is significant and provides further support for and integration with the previously described research. The study found that higher CS leads to higher CL, which in turn leads to higher customer profitability. It is worth noting that the relationships between these variables appear to be non-linear and only hold above certain levels or thresholds.

If previous research has established a link between CS and business performance, exploring the antecedents of CS becomes required. The relationship between CX and CS has been explored in a meaningful and growing body of literature. The implication is that CX is a predecessor of CS. Initially, Yuan and Wu [40] discovered that experiential marketing and experiential value have an impact on CX. Except CX takes precedence over CS, CL, and WOM, according to Maklan and Klaus [41]. O. Pappas et al. [42] see CX only as a moderator of CS and intent to repurchase, and Valdez-Juárez et al. [43] support this approach. However, some scholars posit that CX is the sole predecessor to CS, citing the works of Maklan and Klaus [41], Rose et al. [44], Rose et al. [45], Choi et al. [46], Klaus and Maklan [47], Fatma [48], Martin et al. [49], Parise et al. [50], Roy et al. [51], Pandey and Chawla [52], Roy [53], Terblanche [54], Rahman et al. [55], Banik and Gao [56], among others. Collectively, these studies outline a critical role for CX because, together, they indicate that CX is the sole predecessor to CS, and considering all this evidence, it appears that CX should be at the forefront of companies’ minds and should be a priority because it is the construct that is within their influence and control.

Considering the aforementioned, it is possible to discern the sequence of relationships between CX and CS, as well as between CS and CL. Nevertheless, it is important to acknowledge that there is no consensus among researchers on the constructs that precede CX. Furthermore, most studies tend to focus on a limited aspect of the decision-making process when shopping online, which presents significant challenges for companies trying to make the transition to and achieve success with e-commerce. First, they face difficulty allocating resources effectively since they are unsure of which constructs are relevant. Second, the lack of a holistic perspective results in poor decision making, limited understanding, inefficiency, conflict, and misunderstanding. Yet, as Rodríguez-Ardura et al. [57] emphasize, it is imperative to consider the application of classical theories and models to help explain the behavior of the online consumer. Moreover, Martínez-López et al. [58] assert that a comprehensive proposal should be evaluated holistically to ensure a unified approach. As a result, the conceptual framework of this research is based on theories-in-use (TIU). The fundamental tenets of employing a theories-in-use (TIU) methodology to construct marketing theory are centered on addressing significant marketing issues, offering strong explanatory and predictive capabilities, and cultivating indigenous marketing theories [59,60,61]. This approach emphasizes the necessity of theories that are applicable in both academic and practical marketing contexts, showcasing their ability to elucidate phenomena and accurately forecast consequences. Moreover, it underscores the importance of developing marketing theories that emerge within the discipline itself, rather than exclusively borrowing concepts from other fields, to enhance the intellectual vigor of marketing by nurturing the creation of indigenous theories [60,62].

Based on the above, this research aims to identify the relevant constructs in e-commerce and explore their relationships. Therefore, the research question is: “What are the relevant constructs in e-commerce and how do they relate to each other when consumers make online purchases?”.

2. Literature Review

The literature review followed a non-systematic process, based on the experience and intuition of the researchers. It was conducted through a combination of semi-systematic review and integrative review approaches [63]. The two approaches were integrated because the aim was to create a theoretical model or framework (an integrative review promotes the creation of new theoretical models; the semi-systematic review approach is intended for topics with different conceptualizations). At the outset of the literature review, the emphasis was on CX, CS, and CL. These constructs were used as keywords to identify relevant articles in various databases, and the articles with the highest number of citations were selected. Additionally, recent articles with a lower number of citations were also included to ensure that recent developments in the field were contemplated. The selected articles were analyzed to identify related constructs, and a new instance was then carried out by searching for articles using the identified constructs and selecting the most recent and highly cited articles. This process was repeated until all relevant constructs were identified, and analysis was performed on the resulting set of articles.

2.1. Customer Experience

Holbrook and Hirschman [64] established the notion of experience by stating that individuals seek pleasure, fantasy, and emotion through consumption experiences. They described consumer experience as an essentially subjective state of consciousness involving a variety of symbolic meanings, hedonic responses, and aesthetic criteria. The relevance of the concept of CX increased in the 1990s with the publication of Pine and Gilmore [65], in which the authors characterize this development as the progression of economic value. The concept of experiences has been introduced as the latest economic offering that surpasses commodities, products, and services. The authors call this evolution the progression of economic value, and later they updated their work [66]. Indeed, several authors see CX as a new and groundbreaking approach and orientation for the marketing discipline. Heinonen et al. [67] claim that CX is a pilar of customer-dominant marketing and business logic, despite the different names used, including service experience, consumer experience, and consumption experience. However, Tynan and McKechnie [68] explained that experience marketing under S-D logic requires a new approach.

According to Berry et al. [69], the purchase of a product is always an experience, and the critical aspect is the effectiveness of the company in managing that customer experience. Delivering a positive CX facilitates the creation of an emotional connection between a brand and its customers, thereby cultivating CS [13]. Companies are most competitive when they amalgamate functional and emotional advantages in their products [69]. But, as Gentile et al. [13] pointed out, the emotional connection does not mean that customers will ignore the importance of the functionalities.

In the interest of clarity, the definition of CX must be revised. Kacprzak and Hensel [70] revealed that the CX definitions of Meyer and Schwager [71], Verhoef et al. [72], and Rose et al. [45] are the most quoted. Bascur and Rusu [73] stated that CX has several definitions, depending on the specific context in which it is generated, though the influence of Verhoef et al. [72]’s work is significant. Therefore, CX is defined as the customer’s overall perception of the organization, influenced by their interactions at various touchpoints, with employees, and through technology [72]. An earlier but relevant definition is that of Meyer and Schwager [71], according to which CX is the internal and subjective reaction that customers have to every direct or indirect contact with a company. A more recent definition is the one given by Lemon and Verhoef [74], where CX is the customer’s journey with a company over a period of time during the purchasing cycle, across multiple touch points. As for online CX, it is defined as the cognitive, emotional, and behavioral responses of consumers to interactions with a company that take place through digital channels (e.g., websites, social media, mobile apps) [70].

Although there is widespread acknowledgment of the CX definitions, disagreement remains about other factors. And if there is also a clear consensus among researchers on the multidimensionality of CX, paradoxically, there is a clear dearth of consensus among them on the dimensions of CX and the relationship between CX and the other constructs, as demonstrated further ahead.

Mascarenhas et al. [75] claim that CX is defined by three elements: physical interactions, emotional engagement, and moments in the value chain. Gentile et al. [13] go a step further, stating that buying products involves customers’ senses, emotions, thoughts, acts, values, and relationships. For Yuan and Wu [40], experiential marketing comprises sensations, feelings, cognition, and SQ. Additionally, Tynan and McKechnie [68] confirmed that CX consists of sensory, emotional, functional, relational, social, informational, novelty, and utopian dimensions. One of the most significant proposals outlined by Verhoef et al. [72] states that the CX construct is a comprehensive framework that considers the customer’s cognitive, emotional, affective, social, and physical reactions to the retailer [72]. Lemon and Verhoef [74] add spiritual reactions. Rose et al. [44] also proposed that CX comprises both affective and cognitive states. In Teixeira et al. [76]‘s view, CX is inherently multidisciplinary, and Michaud and Stenger [77] proposed a conceptualization using four core dimensions: physical (place and senses, and time), ideological (shopping values, symbolism, and rituals), pragmatic (acts and gestures, appropriation of the environment), and social (socialization). A more recent work by Lemon and Verhoef [74] defines CX as a multidimensional construct that encompasses a customer’s cognitive, emotional, behavioral, sensory, and social responses to a company’s offerings throughout the customer’s purchase journey. In view of the different dimensions suggested in the bibliography, Lipkin [78] explained that the formation of the CX construct is multi-layered, complex, and intricate. Conceptually, similar work has also been in progress by other researchers and will be the subject of further discussion. For Jain et al. [79], CX is an integrated interactive process, enabled by cognitive and emotional cues, moderated by customer and contextual characteristics. This results in a unique level of pleasant memories. A complementary view is presented by Kranzbühler et al. [80], who contend that there are three lenses—cognition, affect, and sensation—that shape both static and dynamic CX. Bolton et al. [81] propose an alternative approach, indicating that CX is holistic by nature and encompasses cognitive, emotional, sensory, social (relational), and value (lifestyle, spiritual) components. McColl-Kennedy et al. [82] view CX as consisting of value creation components, including resources, activities, interactions, context, and the customer role. Additionally, CX can evoke a range of emotions, such as joy, love, surprise, anger, sadness, and fear, along with cognitive responses such as complaints, compliments, and suggestions that occur at touchpoints. Brun et al. [83] postulate that CX is multidimensional and contains cognitive, affective, sensory, behavioral, and social dimensions. The cognitive, affective, social, and physical dimensions are also recognized by Bustamante and Rubio [84]. De Keyser et al. [85] identified three primary building blocks of CX (touchpoints, context, and qualities), which in turn include twelve components. Waqas et al. [86] define CX in two ways: first, it is a multidimensional concept that is the result of sensory, affective, physical, and cognitive experiences and social identity; second, it is a holistic and multidimensional concept that includes the emotional, social, physical, affective, and cognitive responses of the customer to the retailer. Similarly, Keiningham et al. [87] contend that CX includes cognitive (what customers think), physical (how customers interact), sensory (what customers perceive through their senses), emotional (how customers feel), and social (how customers socialize). Williams et al. [88] contend that CX has components that are experience design and experience delivery. They refer to experience design as making CX memorable and experience delivery as making CX frictionless. As for Gahler et al. [89], CX has affective, cognitive, physical, relational, sensory, and symbolic dimensions. In summary, the preceding statements lead to the conclusion that there is a body of literature that outlines the components of CX, and based on an analysis of the various sources, it can be concluded that there is a lack of consensus.

While there is no consensus on the dimensions of CX, the lack of agreement is even more pronounced when it comes to the constructs to which CX is related. Let’s first remember that Puccinelli et al. [90] uncovered that goals, schema, information processing, memory, involvement, attitudes, affective processing, atmosphere, and consumer attributions and choices have a strong influence during CX. These factors fall into the dimensions mentioned above but are also directly associated with the idiosyncrasies of each consumer. Therefore, a priori, it would be challenging, if not impossible, to identify identical constructs for consumers’ transversality. Nevertheless, previous research proposing CX-related constructs will be discussed, starting by identifying the constructs and then establishing their place.

Novak et al. (2000) conducted a significant initial study on web applications, online shopping, and the evolution of the importance of constructs over time. They found that modeling the consumer experience requires the inclusion of the following constructs: web usage, interactivity, speed, importance, control and skills, challenges and excitement, focused attention, flow, telepresence, and temporal distortion. Among these constructs, flow primarily influences exploratory behavior. Verhoef et al. [72] developed a conceptual model of CX that is determined by various factors, including the social environment, service interface, retail atmosphere, product range, price, CX across different channels, brand, and past CX. Additionally, the context (situation) and the consumer act as moderators. The study by Lemke et al. [91] is noteworthy in the field. CX is defined as the customer’s subjective response to his direct and indirect interactions with the company, while CX quality refers to the perception of excellence or superiority. Their model consists of five constructs: encounter with communication, encounter with service, and encounter with use (all exogenous), as well as the mediating construct of the context of experience and the quality measurement construct called value-in-use, which includes the variables of utility, hedonism, relationship, and cost/sacrifice. Rose et al. [44] suggested that information processing, perceived ease of use, perceived usefulness, perceived benefits, perceived control, skill, trust propensity, perceived risk (PR), and enjoyment serve as antecedents of CX. CX affects CL and CS, and CS affects CL. Later, a new model was proposed [45], wherein CX comprises: cognitive experiential state (interactive speed, telepresence, challenge, skill), affective experiential state (perceived control (ease of use, customization, connectedness), aesthetics, and perceived benefits). Beyond CS and CL, a new construct has arisen that is affected by CX: trust in online shopping. In turn, Bilgihan et al. [92] found that website or app discoverability, ease of use, perceived usefulness, hedonic and utilitarian features, perceived enjoyment, personalization, social interactions, and multi-device compatibility are precursors to unified online CX. A proposal restricted to website elements was made by Bleier et al. [93], who researched and identified 13 design elements that form four domains of the online customer experience (informativeness, entertainment, social presence, and sensory appeal). For Roy [53], CX is impacted by product experience, outcome focus, moments of truth, and peace of mind. Along the same lines of research, Fernandes and Pinto [94] confirmed that CX is preceded by environment, frontline personnel, moments of truth, and product offering and influences relationship quality, which affects retention (loyalty) and WOM. Kemppainen et al. [95] are also proponents of the CX multidimensionality concept. They found that online store characteristics, including interface, assortment, pricing, and trustworthiness, affect CX. Singh and Söderlund [96] established that customer service, website experience, and brand experience are precursors to CX. Siqueira et al. [97] argued that customer perceptions of peer-to-peer interaction, SQ outcomes, and peace of mind positively influence CX. However, Williams et al. [88] only link customer experience to customer share of wallet. Yet another alternative by Yin and Xu [98], confirmed that CX is affected by website experience, product experience, service experience, brand experience, and emotional experience. Rahman et al. [55] proposed the omnichannel CX construct, which is preceded by social communication, value, personalization, customer service, consistency of product availability and pricing across channels, information security, delivery, product returns, and loyalty programs. The physical environment, the interior store environment and layout, the interaction with the staff, the interaction with other customers, the value and quality of the merchandise, and the variety of the merchandise are pointed out by Chatzoglou et al. [99] as being precursors of CX. The preceding sentences support the previous assertions by demonstrating the assortment of proposed constructs related to CX, yet no overarching pattern is discernible in the literature. Therefore, this remains a controversial territory due to inadequate evidence of the existence of a set of holistic constructs related to CX.

Another important source of information is the literature that has systematically reviewed CX. Bascur and Rusu [73] conducted a systematic literature review and found that the influence of CX was based on the work of Gentile et al. [13] and Verhoef et al. [72]. Kacprzak and Hensel [70] systematized that the precursors of CX are SQ, usefulness, social interactions, ease of use and convenience, visual and verbal cues, vividness and interactivity of the website, assortment variety, security and privacy (S&P), cost and price perception, risk perception, terms and conditions, and engagement with content, among others. Lastly, Koronaki et al. [100] found site convenience, site effectiveness, site informativeness, site security, site usability, site entertainment, site responsiveness, social site, and site interactivity. McKee et al. [101] and Zha et al. [4] did not specify CX-related constructs in their systematization.

Measuring CX is critical to this study. Therefore, it is important to identify existing scales. As shown before, the inherent difficulties make this task tough. Lemon and Verhoef [74] have the same line of thought in mind when they ask the question, “How can CX be measured while taking into account its rich, multidimensional nature?”. Becker and Jaakkola [102] have a different opinion when they say that companies cannot create the customer experience, but they can monitor it. There are several scales available to measure CX. Klaus and Maklan [103] developed and validated the EXQ scale for CX in services. The scale consists of 19 items within four dimensions: product experience, outcome focus, moments of truth, and peace of mind. The authors compared the predictive power of the EXQ scale to the predictive power of the CS construct. Ultimately, they concluded that the EXQ is a more accurate explanation and predictor of both CL and WOM than CS [47]. It is also worth noting that the authors also demonstrated that CX influences CS. These relationships have since been validated [104]. Kuppelwieser and Klaus [105] examined the existing CX scales, and among them, 28 were identified. The authors also revealed that EXQ offers a holistic assessment and does not differentiate between the importance of different stages or dimensions of the experience. Bagdare and Jain [106] proposed a scale to measure CX and found that it could be explained in four dimensions, namely leisure, pleasure, distinctiveness, and mood. Lastly, the Gahler et al. [89] scale should also be mentioned, which was developed to measure CX in the omnichannel space and had great success.

Despite some advancements, a unanimous agreement has yet to be reached concerning the dimensions of CX, the constructs associated with it, and the most dependable scales for evaluating CX. In terms of a comprehensive perspective that reflects shoppers’ engagement with e-commerce, there is a clear void in scholarly literature. From the analysis of the numerous references reviewed, the following sections contain all the main constructs related to CX, which, in their overall scope, holistically represent the purchasing process in online stores.

2.2. Customer Satisfaction

CS is the extent to which the perceived performance of an organization’s offering matches the customer’s expected performance [107,108,109]. The study of CS has had important developments with the works of Oliver [110], Bolton and Drew [111], and Gupta and Zeithaml [112], with emphasis on the evaluation of empirical relations between constructs to discover essential predictors, assess customer perceptions and attitudes about the experience, and identify key metrics to begin assessing the overall customer experience. Research consistently confirms a significant correlation between satisfaction and repeat purchases, greater brand loyalty, and the spread of positive opinions about the product [113]. But, as Oliver [114] pointed out, for some companies, CS is the only feasible goal that they should be striving to achieve because CL cannot be attained or attempted as a viable outcome due to the nature of the product category or consumer disinterest. As Rust and Zahorik [115] mentioned earlier, retention (CL) is seen as the most important component of market share, and it is driven by customer satisfaction. Therefore, CS is a key issue in the new perspective on market share. A different view is described by Taylor and Baker [116], who state that consumer satisfaction is best described as moderating the relationship between SQ and intention to purchase. One of the most influential works in the area of CS is the seminal paper of Fornell et al. [117], where they proposed a novel form of performance measurement, the American Customer Satisfaction Index (ACSI). In the model, CS depends on customer expectations, perceived quality, and perceived value (PV), and CS influences CL. The CS metric has become synonymous with the ACSI score [118]. Also of interest is Anderson [119]’s article, where he describes the relationship between CS and WOM. That statement was also backed by Maklan and Klaus [41] and Klaus and Maklan [47], who proved that CS explains WOM, and later on by Klaus [104].

There exists a substantial body of literature on the association between SQ and CS. Cronin Jr et al. [120] found that the SQ, the value of the service, and CS can all be linked in direct ways to behavioral intentions. They also found that SQ leads to CS. Recent research suggests that e-SQ is a precursor to e-CS, which subsequently impacts e-CL [121], or e-SQ precedes e-CS [122,123,124,125,126,127]. Al-Adwan and Al-Horani [128] contend that eTailQ and e-Trust precede e-CS, and Flavián et al. [129] indicated that perceived usability is an antecedent of CS, although it should be noted that this construct is integrated into the e-SQ. A better and more detailed approach is provided by the much more widely acclaimed work of Szymanski and Henard [130]. They suggested that CS is preceded by expectations, disconfirmation of expectations, performance, affect, and equity, and CS is succeeded by WOM and repurchase intention. Work that conceptually shares some common foundations was proposed by McKinney et al. [131], asserting that CS depends on expectations, performance, and SQ. Yet [132], asserted that SQ is prior to CS and CL, and that CS is in turn prior to CL.

As for Yuan and Wu [40], experiential marketing (sensations, feelings, cognition, and SQ) influences experiential value (emotional and functional), which in turn has an impact on CS. This is an indication that value does have an impact on CS. Similarly, according to Eklof et al. [33], PV impacts customer CS. Another perspective comes from Brakus et al. [133], who say that the brand experience has an impact on CS and that CS has an impact on CL. But Rose et al. [44] argue that CX affects CS and CL. Martin et al. [49] only indicate the successors of the CS as the CL. According to Audrain-Pontevia et al. [134], perceived purchase value and perceived transaction value are predictors of e-CS, which are precursors to e-CL. However, Turkyilmaz et al. [135] uncovered that perceived quality influences PV, and these two constructs influence CS. Oh [136], Hellier et al. [137], and Kuo et al. [138] pointed this out earlier. In contrast, McDougall and Levesque [139] consider value and quality as independent constructs that both precede CS, and Caruana [140] considers only SQ to be the predecessor of CS. Szymanski and Hise [141] determined that convenience, site design, and financial security are the most important drivers of e-CS, and their findings were validated by Evanschitzky et al. [142]. Morgeson et al. [143] used the American Client Satisfaction Index (ACSI) and found that CS is preceded by the constructs of PV and perceived quality, which in turn influence CL. A different view is presented by Sharma and Aggarwal [144], who indicate website SQ, WOM, personalization, and the customer support system as precedents for CS. Additionally, Eklof et al. [32] describe the European Customer Satisfaction Index (ECSI) as a derivation of the ASCI, containing the constructs perceived quality, expectations, image, PV, CS, complaints, and CL. The inclusion of corporate image, defined as an antecedent of satisfaction, is an important distinction.

There is a debate going on among scholars about the impact of CX on the CS, and it is now generally agreed about that impact, being the case of Choi et al. [46], Bustamante and Rubio [84], Kawaf and Tagg [145], Ali et al. [146], Roy [53], Singh and Söderlund [96], Chatzoglou et al. [99], Filieri et al. [147], Kacprzak and Hensel [70], Koronaki et al. [100], and Moliner-Tena et al. [148].

Several different scales have been proposed to measure CS. A general scale was proposed by Westbrook and Oliver [149], Crosby and Stephens [150], Oliver and Swan [151], Oliver and Swan [152], Ganesan [153], and Mägi [154]. Probably the most commonly used and well-known option is the one from Fornell et al. [117]. There are also scales that measure satisfaction with hypothetical experience [155], satisfaction with most recent experience [156], satisfaction with performance [157], satisfaction with purchase experience [158,159,160], satisfaction with retailers [161], satisfaction with service [162,163,164], and satisfaction with store [120,165,166].

2.3. Word-of-Mouth

WOM is the informal communication among consumers about a company, including email, blogs, and social media, and it can be about users’ experiences with products or services, their opinions or recommendations, or simply the provision of unbiased information [108]. The power of WOM in persuading consumers is well-documented and long-standing, as evidenced by the research of Engel et al. [167], Sheth [168], and Kumar et al. [169]. According to Litvin et al. [170], interpersonal influence and WOM rank as the top sources of information used by consumers to make purchasing decisions. Berger [171] complements the idea by saying that WOM is goal-oriented and is used in the acquisition of information and the persuasion of others. However, Lovett et al. [172] assert that social and functional factors are the primary drivers of online WOM, and consumers use WOM to save time and improve purchases, and that WOM significantly influences their behavior [173]. Mulyadi et al. [174] also claim that WOM has a positive and significant impact on purchasing decisions or may assume a significant role in the consumer decision-making process [175]. Lastly, positive customer experiences throughout the entire customer journey result in a positive WOM [74], repurchase intention, and WOM [176].

Once the importance of WOM has been established and verified, understanding its antecedents becomes required. There is a substantial body of literature examining the antecedents of WOM, and it is widely acknowledged that CX is a key predictor of WOM. The main publications are by Brown et al. [177], Maklan and Klaus [41], Klaus and Maklan [47], Klaus [104], Bilgihan et al. [92], Kawaf and Tagg [145], Cambra-Fierro et al. [178], Roy et al. [179], Siqueira et al. [97], Rahman et al. [55], Moliner-Tena et al. [148], Banik and Gao [56], and Kacprzak and Hensel [70]. One exception comes from Singh and Söderlund [96]’s study, where they suggest CX predicts CS, which influences WOM.

Regarding the WOM construct, it remains to be seen how to measure it. WOM measurement scales can be: WOM negative [180,181,182,183,184,185,186], WOM hypothetical [187,188], WOM support seeking [189], WOM intention positive [177,188,190,191,192], WOM intention [193], WOM pleasure [194], WOM social benefits [195], WOM intensity [196,197], WOM skepticism [198], and WOM general [188]. Additionally, Goyette et al. [199]’s WOM scale consists of four dimensions: WOM intensity, positive WOM valence, negative WOM valence, and WOM content.

2.4. Customer Loyalty

CL occurs when a person regularly patronizes a particular retailer whom they know, like, and trust [109] and is supported by a positive experience throughout the entire customer journey [74]. Jain et al. [79] argue that to attract, satisfy, and retain customers, it is critical to create an experiential perspective, provide excellent service, and prioritize CX. CL is important because the cost of marketing to loyal customers is lower than the cost of marketing to new customers (up to five times more expensive) [108], even though loyal consumers tend to be the most satisfied, though being satisfied does not always translate into staying loyal [114]. Then there is the loyalty ripple effect, which is the result of customers creating awareness by encouraging new customer acquisition and value-creating actions or behaviors [200]. However, it is vital to know that CL is a process that consists of four distinct and sequential stages [201]. Cognitive loyalty refers to believing that a particular brand is preferred over other brands. Affective loyalty represents a positive attitude or preference resulting from a satisfying experience. Conative loyalty involves the development of behavioral intentions that reflect a deep level of commitment. Action loyalty is the willingness to overcome obstacles to convert these intentions into action. However, Chitturi et al. [202] suggest that customer loyalty is formed by WOM and repurchase intention and is preceded by satisfaction, which in turn is preceded by hedonic and utilitarian benefits, whereas Chiu et al. [203] only mention the last two constructs. Additionally, Bandyopadhyay and Martell [204] claim that attitudinal loyalty plays a pivotal role in fostering behavioral loyalty. Still, CL is indeed a much more complex construct, as Dick and Basu [205]’s relevant and seminal work revealed.

As for CL’s predecessors and successors, the bibliography is extensive and controversial. Several studies have been published examining the impact of CS (as the sole antecedent) on CL [33,42,44,96,121,123,134,146,206,207,208]. A minority of authors report an association between the brand and the CL, in particular that the CL is preceded by CS and brand reputation [209], or that the CL is preceded by the brand experience [133]. PV has also been explored. CL is preceded by PV, affective commitment, and brand equity [210], or by PV, trust, and CS [166], or by CS and PV [211,212], or by PV, CS, and SQ [213]. Within the literature, there has been extensive research on the relationship between SQ and CL. CL has its precursors in CS and SQ [140,214], or in SQ [215]. As Giao et al. [124] claim, website quality has a positive effect on e-loyalty, which is mediated partially through consumer e-trust and e-satisfaction. But most important, Xu et al. [216] found that SQ increases CL, and it is a critical factor in fostering loyal customers. A broad level of interest is also aroused by trust, as validated by the works of Ribbink et al. [217], Flavián and Guinalíu [218], Kim et al. [125], Al-Adwan and Al-Horani [128], Giao et al. [124], and Mofokeng [219].

Also, CX is widely recognized in the literature as a precursor to CL, as a standalone or combined with other constructs [44,46,47,51,53,55,70,75,83,84,88,92,94,98,99,100,104,144,145,146,220,221,222,223,224,225]. Two additional constructs are also mentioned: commitment [226] and performance perception [227].

To measure and manage CL, several metrics have been widely used [228]: recency, frequency, and monetary value (RFM), past customer value (PCV), and share of wallet (SOW). As for the scales, they are as follows: Loyalty general [229], loyalty action [167], loyalty active [230,231], loyalty affective [166], loyalty brand [232,233], loyalty cognitive [166,234], loyalty conative [166], loyalty passive [231], loyalty proneness to the product [235], loyalty to the e-retailer [231], loyalty to the store [236,237,238,239], and loyalty altruistic (willingness to pay more) [240].

2.5. Service Quality

Within the literature, there has been much research surrounding the SQ and e-SQ concepts. Work in this field was heavily influenced by Parasuraman et al. [241], who developed and refined the SERVQUAL model [242]. There is widespread agreement on the five dimensions that determine service quality: reliability, responsiveness, assurance, empathy, and tangibles. There is also a divergence between SERVQUAL [241] and SERVPERF [243,244], which is whether SQ should be assessed with perceptions minus expectations or with perceptions alone. There is, however, no consensus on the dimensions of online service quality.

E-service refers to providing services in cyberspace [245]. Traditional SQ refers to service quality based on human interactions and experiences during service encounters, while e-SQ involves interactions between people and technology [246]. Zeithaml et al. [247] were the first to prepare a formal definition of website service quality. They state that e-SQ is the degree to which a website enables efficient and effective shopping, purchasing, and delivery. The authors further assert that the provision of high-quality service through websites is critical to the achievement of strategic objectives and that SQ is likely to be more important than low prices or web presence [248]. Similarly, service quality sets the stage for experiences, influences journey mapping, and values CX [74], and it is key to providing excellent service and prioritizing the customer experience [79].

After several decades of research, the successors of the SQ continue to be the topic of debate among researchers. Three guidelines are presently available. The first proposition holds that SQ influences CX [46,70]; the second posits that SQ has an impact on CS [122,138,143,174,214,249,250,251], while the third proposes that SQ has an impact on CL [128,138,214,227,249,251,252].

The research on the e-SQ has been productive and continuous and has resulted in a considerable number of models, theories, and constructs, namely by Liu and Arnett [253], Zeithaml et al. [247], van Riel et al. [254], Yang et al. [255], SiteQual by Yoo and Donthu [256], WebQual by Barnes and Vidgen [257], WebQual by Loiacono et al. [258], Aladwani and Palvia [259], Madu and Madu [260], eTailQ by Wolfinbarger and Gilly [261], Santos [245], Trocchia and Janda [262], Kim and Stoel [263], Lee and Lin [249], Bell et al. [264], E-S-QUAL by Parasuraman et al. [265], Kim et al. [246], eTransQual by Bauer et al. [266], Gupta et al. [267], SERVCON by Seiders et al. [122,251,268,269,270,271], e-SERVAR by Chen et al. [272].

2.6. Web Content

The purchasing process comprises two stages: consideration and evaluation, which rely on web content (WebC). To manage the complexity of consumer decisions, CX requires a rational information processing approach to consumer decision making [80]. Kawaf and Tagg [145] also agree with this perspective. Bleier et al. [93] research recognizes 13 design elements that create four domains of CX, including informativeness, that influence purchasing. According to Suh and Moradi [273], the clarity of the information is an essential element of the website that shows a key function in the positive experience that users have with the site and positively affects the CX. Also, Koronaki et al. [100] identified site informativeness, which is related to the quality of information issues, as a support for cues with high task relevance, but Pandey and Chawla [52] found that website informativeness was negatively and insignificantly related to satisfaction but positively and significantly related to loyalty. Several other benefits have been discussed in the literature. Starting with Chen and Dubinsky [274], they suggest including the relevant information construct, stating that consumers only perceive relevant information as useful and valuable. Park and Kim [275] identified information satisfaction as a construct, defined as the level of satisfaction that customers feel with the general information in the service encounter, and Geraldo and Mainardes [276] stated that product information is an antecedent of purchase intention. Pavlou and Fygenson [277] showed that information is a pertinent construct in the purchase process as it is related to intention and purchase behavior, and in Thongpapanl and Ashraf [278]’s experience, personalized information has an influence on customer satisfaction and purchase intention. Moreover, Orús et al. [279] identified that the type of WebC also exerts an influence on behavioral intention. Therefore, a company needs to provide attractive, high-quality content to satisfy customer needs [278,280]. Furthermore, several authors have added additional contributions, maintaining that WebC can induce consumers’ attitudes and purchases [281,282,283,284], experiences [274,285] and PR [286,287].

However, it should be noted that several authors include WebC as a dimension of service quality, as in the cases of Yang et al. [255], Barnes and Vidgen [257], Loiacono et al. [258], Aladwani and Palvia [259], Santos [245], Trocchia and Janda [262], Kim and Stoel [263], Kim et al. [246], Ha and Stoel [270], Orehovački et al. [271], Bressolles et al. [122], and Chen et al. [272]. Others include WebC in web design [288].

Although limited in number, there are some scales for evaluating WebC: attitude toward the website content [289,290], attitude toward the website information value [291], website design information quality [258,292], attitude toward the webpage informativeness [93,293], and informedness pre-purchase [294,295]. It is also conceivable to use the SQ scales mentioned above, the part that includes the WebC as well.

2.7. Security and Privacy

S&P is defined by Wolfinbarger and Gilly [261] as the protection of credit card payments and the confidentiality of shared information, and Zeithaml et al. [247] defined it as the extent to which customers consider that the site is safe from unauthorized access and that personal information is properly safeguarded. Kacprzak and Hensel [70] pointed out that in the purchasing phase, two of the most relevant factors for CX are privacy and security. Koronaki et al. [100] support the concept that S&P are inextricably linked and relevant to the CX experience. But Kawaf and Tagg [145] argue that, in some contexts, S&P may not be the deciding factor. A slightly different approach was followed by Valdez-Juárez et al. [43], who found that consumer satisfaction is positively and significantly affected by higher levels of website security. Rahman et al. [55], however, return to the more conventional approach with the statement that CX is preceded by information safety (S&P). The same is not supported by Chen and Barnes [296], who only suggest the security dimension.

S&P is also included in SQ, in various models. Liu and Arnett [253], Zeithamlet al. [247], Yoo and Donthu [256], Wolfinbarger and Gilly [261], Parasuraman et al. [265], Gupta et al. [267], Kim et al. [246], Kim et al. [125], Bressolles et al. [122], Giao et al. [124], and Rodríguez et al. [121] mention S&P; Aladwani and Palvia [259], Madu and Madu [260], Park and Kim [275], Santos [245], Trocchia and Janda [262], Orehovački et al. [271], Orel and Kara [214], and Chen et al. [272] point out security.

Several studies and analyses have been undertaken to explore the placement of S&P, with inconsistent results. Flavián and Guinalíu [218] showed that S&P needs to come before trust, and CL. Chen and Barnes [296]’s view is that trust is a function of perceived S&P. Al-Adwan and Al-Horani [128], and Mofokeng [219] have the same line of thought. Differently, Ha and Stoel [270] have shown that S&P comes before purchasing intention, and CS, Anshu et al. [297] have shown that they are the precursors of CX, and Sharma and Aggarwal [144] have shown that they are essential to the successful operation of e-commerce. Some authors also claim that S&P precedes PR [298,299].

To measure S&P, one can refer to the SQ bibliography, which includes these two constructs. Alternatively, additional scales can be used: security importance [300], security of Internet financial transactions [301], privacy concerns [302,303,304,305,306,307,308].

2.8. Perceived Price

Sheth and Malhotra [228] emphasize that price is a complex and important construct. As the authors highlight, price requires processing and is a challenging and complex numerical cognitive process. It is also influenced by remembering ability, reference prices, perceptions of value, the relationship between price and quality, the relationship between price and the perception of monetary sacrifice, and these components are integrated. A large body of evidence about online shopping reveals that price is one of the most relevant factors for consumers [309,310,311]. It has become even more important in online shopping [312,313]. Price can be categorized as either objective or perceived [314] and while objective price refers to the actual price paid by the customer, PP refers to the price as determined by the customer.

Zeithaml et al. [247] suggested that PP is comprised of e-coupons, shipping costs, and comparison costs. But for Dubrovski [113], PP is composed of monetary and non-monetary prices. Price negatively impacts value as a financial sacrifice [276], and Zeithaml et al. [247] point out that PP is a precursor to PV. For Kacprzak and Hensel [70], PP is a precursor of CX and is of primary significance in the purchase phase, but according to Wang [315], price moderates the relationship between CX and CS. Some researchers, however, have identified price as the successor of CX [46]. Also of interest is the paper of Lin et al. [316], claiming that PP positively influences online CS, also confirmed by Alzoubi et al. [317] and by Herrmann et al. [318]. Martín-Consuegra et al. [319] and Kaura et al. [320] add CL in addition to CS. However, some studies do not confirm that PP influences CS [321]. Kim et al. [322]’s takeaway, however, is that PP is a precursor to PV.

As far as scales are concerned, they are very diverse and can be very different: price perception [323]; price perception—internal reference [324], price perception—store comparison [325,326,327], price-consciousness [328,329,330,331,332], and price [333].

2.9. Perceived Risk

The Internet has the potential to exacerbate certain fears inherent to the purchasing process [334]. PR can be difficult to measure, depending on how the construct is conceptualized. Baker [107] detailed that the purchase of a product has two elements that, in combination, have a certain degree of PR: the first is the consequences of making the wrong choice, while the second is the certainty or uncertainty about the likelihood of that product choice being a bad choice. Consequently, PR can be conceptualized as the degree of unpredictability that consumers face when they cannot accurately predict the outcomes of their purchasing choices [109] or the belief that the implications of a buying experience may be more negative or adverse than the buyer believes them to be [5]. When shopping online, the main risks are financial, performance, and privacy [274]. But for Chen and Barnes [296], PR is made up of perceived security and perceived privacy, and it increases due to a lack of privacy and security [297]. Additionally, online shopping involves higher PR due to payment, information delivery, and information exchange [128]. It is also worth mentioning that, according to Maklan and Klaus [41] and Klaus and Maklan [103], PR occurs at the moment of truth.

PR complexity can be best grasped by considering the diversity of factors in a highly intricate context and highlighting the brand, information, and channels. PR usually decreases as trust increases [217], and as individuals perceive less risk, they tend to buy their preferred brands [210]. Therefore, brands can significantly reduce PR and reassure customers [217]. Moving towards information, Kim et al. [246] claim that more information can reduce PR. Furthermore, customers rely on informal communication to reduce the PR of a product before they buy it [222]. As for the channels, Lemon and Verhoef [74] said that PR can be reduced by integrating online and offline channels and Shi et al. [335] go on to say that omnichannel purchase intent is negatively related to PR. Yet, Liu and Forsythe [336] posit that PR is contingent upon the risk associated with the channel and the product in question, or only the product [337].

As for the relationship and placement of PR with other constructs, there is also a puzzling context. Rose et al. [44] suggest that PR precedes CX, which is also backed up by Kawaf and Tagg [145] and by Waqas et al. [86]. In contrast, Roy et al. [51] think that CX precedes PR and, similarly, a high level of PR can be caused by inconsistencies in the CX [273]. However, Kacprzak and Hensel [70] reverted to the most widely accepted assumption, claiming that risk perception occurs at the purchase stage and precedes CX.

There are also studies linking PR to CS, CL, trust, and WOM. Lovett et al. [172] found that PR does affect WOM in online contexts. Martin et al. [49] and Valdez-Juárez et al. [43] confirmed that PR influences CS. But PR is also related to CL [223] and repurchase intention [49,203,338]. But perhaps more controversial is the relationship between PR and trust. What comes first, trust or PR? The PR associated with transaction processes can be reduced through trust [296]. Therefore, according to the authors, trust precedes PR. Martin et al. [49] support the previous statement, saying that trust is an antecedent of PR and that the latter is influenced only by the affective component of experience. Then, in contrast, Dhaigude and Mohan [339], argue that trust is a precursor to risk and vice versa, so it is preferable to address them simultaneously. And, as seen above, there are a number of relevant studies that point out that trust is influenced by PR.

To assess risk, there is a substantial and comprehensive set of risk measurement scales: risk—financial [340,341,342], risk—general [287,343,344,345], risk—performance [341,342,346,347], risk—personal [348], riskiness of providing information online [305,308], risk—service [349], and risk general [334].

2.10. Trust

Berry and Parasuraman [350] define trust as the customer’s belief that the service will be delivered as agreed and that the company can be regarded as truthful and fair. They said trust is critical in developing customer relationships and is built by providing coherent and dependable service, communicating with clarity, and satisfying customer needs and concerns. The most historically and scientifically accepted definition of trust comes from Morgan and Hunt [351], who define trust as the readiness to rely on an interlocutor who trusts you. Trust is at the foundation of relationship marketing [352].

As a result of the controversy discussed in the previous section, trust is also a construct that lacks consensus in terms of its associations and place in the e-commerce framework. It can also be explored from a multidimensional perspective that includes both cognitive and affective responses, such as consumer trust, website trust, brand trust, and retailer trust, among others [100]. Some authors have included trust in the SQ [144,247,257,258,260,263]. However, there are additional claims: the quality of the website can impose trustworthiness [253], trust is a function of SQ [166], or trust has an impact on SQ [249]. Additionally, it can be argued that CL are preceded by commitment and trust, which in turn are preceded by trust [353].

The literature presents varying perspectives on the relationship between trust and CX, with some arguing for its role as an antecedent and others as a consequence [44]. The authors also stated that trust is a recurring construct in models of online service or customer behavior. Suggestions regarding the relationship between trust and CX are: CX influences trust [49,55], CX is a moderator between trust and CS [42], CX depends on trust propensity [145], and trust affects CX [95].

Concerning the CS and CL constructs, divergences persist: trust is a precursor to CL and CS [125,128,166], trust is preceded by CS and precedes CL [217], trust has an impact on CS [52,249,354], trust affects CL [129,218,219]. Other relationships found in the bibliography are: trust precedes perceived control [247], perceived security affects trust [218,219], trust is a function of usability [129], trust depends on perceived usefulness, perceived security, perceived privacy, perceived good reputation of the company, and personal trust disposition [296], trust depends on fulfillment/reliability, and security/privacy [125], trust precedes WOM [199], justice precedes trust [250], trust is dependent on CS [45], trust precedes perceived usefulness [354], trust precedes PR [49], trust is influenced by reputation, social commerce components, fulfillment and reliability, S&P, and customer service [128], social experience has an impact on trust [355], and trust mediates partially the relationship between website quality and CL [124].

The following trust scales are available: trust [351,356,357,358], trust in the company [239,359,360,361], trust in the service provider [162,166], trust in the brand [362], trust in the company’s website transaction skills [363].

2.11. Perceived Value

Perceived value is the difference between the customer’s evaluation of the benefits and all costs of the offer and the perceived alternatives [109], while Dacko [108] defined it as the consumer’s assessment of the value of the offer in terms of money, quality, or some other measure of its worth. On the other hand, Sheth and Malhotra [228] argue that perceived value is the sum of the reference value and the differentiation value. The reference value is the net value of the nearest alternative, whereas the differentiation value refers to any value that the focal product provides beyond the reference product. Perceived value is composed of various elements, including the consumer’s perception of product performance, channel fulfillment, warranty quality, and customer support, as well as more intangible elements such as the supplier’s reputation, trustworthiness, and esteem [109]. It is the result of a trade-off between consumers’ perception of quality and the sacrifices they make [228]. But Fornell et al. [117] define PV as the perceived level of product quality relative to the paid price. In the view of Audrain-Pontevia et al. [134], perceived acquisition value and perceived transaction value are de dimensions of perceived value.

Fornell et al. [117], in the ACSI model, PV is preceded by customer expectations and perceived quality and drives CS. Chen and Dubinsky [274] explained that PP is influenced by the retailer’s reputation, the price of the product, and the perceived quality of the product. This, in turn, affects PV. As a matter of fact, a substantial body of literature states that PV is preceded by SQ, or product quality. Consider the evidence that follows: PV drives CL, and it is preceded by price and SQ [247,248] or SQ drives PV [265,266]; perceived quality and perceived equity precede PV, which drives CS and brand preference [137]; PV is preceded by SQ and it drives CS and post-purchase intentions [138]; SQ drives PV, which impacts trust and CL [166]; functional quality and hedonic quality have a positive impact on PV [269]. Literature also highlights the importance of the relationship between CX and PV. PV is related to CX [79]; CX is positively associated with PV, and PV drives CL (behavioral and attitudinal) [221]; CX and in-store emotions affect PV, which in turn affects CS and CL [99]; PV is a dimension of CX [100]; for Yuan and Wu [40], experiential marketing (sensations, feelings, cognition, and service quality) influences experiential value (emotional and functional), which in turn has an impact on CS. CS and CL are often mentioned. PV impacts CS and CL [212]; Audrain-Pontevia et al. [134] showed that PV is a precursor of CS; according to Eklof et al. [33], PV impacts CS, which in turn impacts CL. But, according to Harris and Goode [166], value is a function of perceived benefits and PR, and, for Johnson et al. [210], PV drives affective commitment, brand equity, and intentions.

There are several relevant scales for measuring value, including the following: value consciousness [364]; value of the offer [324,365,366,367,368,369,370]; value of the product [166,230,371,372,373,374,375]; value of the transaction [238]; value—social [376]; value in e-commerce [377]. It is also worthy of note that a more comprehensive PERVAL scale has been developed [376].

2.12. Literature Review Critique and Summary

As Rodríguez-Ardura et al. [57] note, incorporating classical theories and models when studying online consumer behavior has significant advantages. Martínez-López et al. [58] further complement this approach by claiming that a holistic view is needed to provide a unified approach. The literature review identified TIUs that are relevant constructs in e-commerce, complying with the recommendations and suggestions of Rodríguez-Ardura et al. [57] and Martínez-López et al. [58].

The following TIUs were used as macro foundational: service quality by Parasuraman et al. [378], Cronin and Taylor [243], and Zeithaml et al. [186]; consumer perceptions of price, quality, and value from Zeithaml [379]; the commitment-trust theory of relationship marketing from Morgan and Hunt [351]; customer experience from Gentile et al. [13], Verhoefet al. [72], and Lemon and Verhoef [74]; S-D logic from Vargo and Lusch [380]; word-of-mouth from Buttle [381]; satisfaction from Fornell et al. [117]. Academic research underscores the significance of theories in anticipating consumer decision making when data is scarce [61]. Although practitioners have voiced criticisms about theory development in academic journals, theories remain indispensable in marketing practice by furnishing valuable insights and decision-making frameworks [382]. It is thus imperative to comprehend the interrelationship between academics and practitioners. For Cornelissen [383], the relationship between the marketing academy and practice has sparked debates on how academics and practitioners theorize, evaluate, and utilize marketing theories, highlighting the divergent orientations between the two groups. The author continues by stating that research has also focused on understanding how practitioners value and use academic theories, urging further exploration into this area to enhance the understanding of how marketing practitioners engage with scientific theories [384]. It is therefore of paramount importance to ascertain whether the theories in use have been integrated with unified and holistic models that enable practitioners to devise and implement customer experience-focused strategies. Conversely, it is vital to consolidate existing models and frameworks. This is demonstrated in the

Table 1. Revising CX models based on constructs, study type, and theoretical background.

Antecedent Constructs	Consequent Constructs	Study Type	Theoretical Background	Holistic View	Authors
Past customer experience, Social environment, Service interface, Atmosphere, Price, Assortment, Channel, Moderators (situation and consumer)	-	Conceptual framework	Integration of theories based on the literature review	No	[72]
Information, Processing, Perceived Ease of-Use, Perceived, Usefulness, Perceived Benefits, Perceived Control, Skill, Trust Propensity, Perceived Risk, Enjoyment	CS and Repurchase Intention	Conceptual framework	Acceptance and adoption models	Yes	[44]
Peace of mind, Outcome focus, Moments of truth, and Product experience	CS, CL, and WOM	Conceptual framework and empirical work	Integration of theories	Yes	[41]
Interactive Speed, Telepresence, Challenge, Skill, Ease-of-use, Customization, Connectedness, Aesthetics, and Perceived Benefits	Online Shopping Satisfaction, Trust in Online Shopping, and Online Repurchase Intention	Conceptual framework and empirical work	Literature review	Yes	[45]
Brand Performance, Multichannel Interaction, Service Interface, Physical Environment, Social Environment, Price and Promotions	CL, CS, and Customer Equity	Conceptual framework	Integration of theories based on the literature review	Yes	[48]
Easiness to locate, Ease of use, Hedonic and Utilitarian Features, Usability, Perceived Usefulness, Perceived Easy of Use, Perceived Enjoyment, and Social Interaction	-	Conceptual framework	Integration of theories based on the literature review	No	[92]
Perception of the environment, emotion, risk and trust, behavior, and situation		Conceptual framework and empirical work	Personal construct theory	Yes	[145]
Webpage Design Elements (Verbal, Visual, and Verbal/Visual)	Behavior	Conceptual framework and empirical work	Integration of theories based on the literature review	No	[93]
Stimuli within firm-controlled touchpoints, Stimuli outside firm-controlled touchpoints, and Dynamic interplay stimuli	SQ, CS, and Value in use	Conceptual framework	Integration of theories based on the literature review	Yes	[102]
Customer service, Website experience, Product experience, Delivery experience, Brand experience	CS, WOM, and Repurchase Intention	Conceptual framework and empirical work	Integration of theories based on the literature review	Yes	[96]
Self-identity, Social bonding, Humor, and Aesthetics	Customer Engagement	Conceptual framework	Consumer culture theory	No	[86]
Pre-purchase, purchase, and post-purchase stages (dozens of constructs)	CS, CL, Behavioral Intention, Engagement, Attitude, and Other	Literature review	-	Yes	[70]
Web site elements (site atmospheric cues and site interface features)	Consumer response (cognitive, affective, behavioral, and multidimensional)	Literature review	-	Yes	[100]

.

Overall, in consideration of the findings presented in the preceding table, it can be concluded that none of the research presented here proposes a holistic perspective grounded in TIU and a conceptual framework that has been empirically validated. Another significant constraint in all the aforementioned research is the absence of constructs that are indispensable for a more comprehensive or complete perspective of CX. Additionally, concerns exist regarding the reliability of some of this research due to a lack of empirical confirmation. A closer look at the literature reveals a number of gaps and shortcomings. Research requires studies that comply with the requirements stated by Rodríguez-Ardura et al. [57]—incorporating classical theories and models—and by Martínez-López et al. [58] a holistic view that provides a unified approach. In view of heightened interest and existing constraints, and in light of the rising prevalence of CX, it becomes imperative to conduct research that addresses the gap that has been identified.

3. Conceptual Framework and Development of Hypotheses

The main objective of this research is to identify the constructs that are at the root of e-commerce and to determine how they relate to each other. There have been 11 constructs identified in the literature review, and several issues have come to light. The first of these is found in the very definition of the constructs themselves. In their definition, there is a certain amount of vagueness. If a construct is not well defined, there will be limitations to its operationalization and measurement. Secondly, there is an overlap between constructs. The most paradigmatic case of this is the different proposals for the online SQ. As a result, these different views may lead to somewhat confusing conclusions in their explanation of the e-commerce experienced by consumers. Based on the aforementioned review, the conceptual framework was developed, proposing the constructs and placing them and their relationships.

Table 2. E-commerce constructs.

Construct	Research Hypotheses	Main Studies
Web Content (WebC)	H1: WebC reduces PR	[286,287]
Service Quality (SQ)	H2.1: SQ reduces PR H2.2: SQ positively influences CX	[46,70]
S&P	H3.1: S&P positively influences Trust H3.2: S&P reduce PR	[218,296,298,299]
Perceived Price (PP)	H4: PP positively influences PV	[247,322]
Perceived Risk (PR)	H5: PR positively influences Trust	[49,296]
Perceived Value (PV)	H6: PV positively influences CX	[79,100]
Trust	H7: Trust positively influences CX	[44,145,385]
Customer Experience (CX)	H8: CX positively influences CS	[41,44,45,46,47,48,49,50,51,52,53,54,55,56], among many others
Customer Satisfaction (CS)	H9: CS positively influences WOM	[41,47,104,119,130]
Customer Satisfaction (CS)	H10: CS positively influences CL	[33,42,44,96,121,123,134,146,206,207,208], among many others

contains the constructs, their research hypotheses, and their support.

Based on the constructs and research hypotheses presented in the previous table and supported by the literature, the conceptual model shown in Figure 1 was developed.

4. Methodology

This study is a type of exploratory research, which is classified as basic research. It is a cross-sectional study that falls under the category of quantitative research, using survey research as the data collection technique. Primary sources were used to collect the data. To evaluate the research hypotheses presented in the research, a questionnaire was developed, which was distributed in a self-directed form. A pre-test was conducted, and suggestions and recommendations were obtained from the participants. The operational model of the constructs to be measured is presented in

Table 3. Research Model Constructs.

Constructs	Items of the Construct	Adapted from
WebC	1. The online store contains up-to-date information. 2. The online store provides complete information. 3. The online store has information that is easy to comprehend. 4. The information in the online store reflects your competence.	[275]
SQ	1. The online store performed the services I requested correctly the first time I used it. 2. The online store provides the services that I requested at the time that was promised. 3. It is easy for me to find a product and carry out a transaction in this online store. 4. The online store is well-designed and has a professional look and feel.	[258,261,265]
S&P	1. The online store has privacy protections in place for me. 2. The online store has appropriate security features.	[261]
PP	1. In my opinion, I have received more benefits than what I paid for my last purchase in the online store. 2. In my opinion, the value of my purchases was worth the price I paid for them. 3. In my opinion, the prices in the online store are reasonable. 4. In my opinion, the prices in the online store are clear and easy to understand.	[333]
PR	1. I did not doubt that the product I purchased would arrive undamaged. 2. I did not feel worried about the payment process. 3. I did not feel worried about shipping, delivery, returns, and exchanges.	[274,349]
PV	1. Comparing what I paid with what I received, my last purchase at the store I frequent most was a good buy. 2. Compared to other stores, the products purchased from the online store were reasonably priced. 3. The products I bought were considered to be a good value purchase. 4. The effort I put in to make this online purchase was well worth it for what I received in return.	[274,377,386]
Trust	1. This online store is honest. 2. This online store has a good reputation. 3. This online store does not act in a way that harms me. 4. This online store protects my interests.	[362]
CX	1. This online store gave me what I wanted. 2. This online store was a pleasure to shop at. 3. All interactions with the online store were efficient. 4. The online store knew or would know how to handle something going wrong with buying.	[47,103,106]
CS	1. If I had to make a choice, I would choose to buy from this online store again. 2. I made a good decision to buy from this online store. 3. My expectations were met by my purchase from this online store. 4. I feel comfortable making purchases from this online store.	[120,230,261]
CL	1. I am not going to switch to a different online store to make the same kind of purchases. 2. This online store is my first choice when I need to make a purchase. 3. This is the online store that I prefer. 4. It is my intention to shop at this online store for many years.	[47,186,200,231]
WOM	1. I am going to recommend this online store to my friends and acquaintances. 2. I am going to give a good recommendation about this online store to other people. 3. I am going to encourage my friends and family to visit this online store. 4. I will recommend this online store to anyone who asks me to do so.	[186,231]

, along with the corresponding scales and the authors of these scales. All constructs were measured using a five-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). The following is particularly noteworthy. The primary challenge was the application of common scales, which would result in an unworkable questionnaire with excessive dimensions and hundreds of questions. Therefore, obtaining sufficient responses for construct validation and evaluating the relationship between them would not be feasible. For this reason, the use of constructs and scales adapted from other authors is proposed in this research.

As shown by Rigdon et al. [387] and Hair Jr. et al. [388], reducing the number of indicators per construct increases measurement uncertainty, but it also increases model fit. The use of scales with a reduced number of indicators was only due to the reasons mentioned previously and was not intended to obtain an adjusted model. Furthermore, uncertainty can be addressed by repeating the study.

The study targeted Portuguese consumers who shop online. Non-probability sampling was used, specifically convenience or snowball sampling, to increase the likelihood of reaching respondents and for the convenience of researchers and respondents. The dissemination of the questionnaire was via social media and e-mail. The model was tested using two independent samples collected on different dates. The first fieldwork took place between May and June 2021, and the second fieldwork took place in May 2022. The use of two independent samples sought to mitigate the occurrence of bias, non-response bias, common method variance, and common method bias that can affect the reliability and validity of empirical results. Partial Least Squares Structural Equation Modeling (PLS-SEM) was used to assess the research hypothesis because the study is testing a conceptual model and is suitable for examining new constructs when the theoretical foundation is not well established.

5. Results

5.1. Sociodemographic Data

The socio-demographic data of the sample are presented in

Table 4. Sociodemographic and last purchase data.

Gender		Female			Male
	S1 * S2 **	60% 58%			40% 42%
Age group		18–24	25–34		35–44	45–54		>54
	S1 S2	42% 63%	27% 23%		9% 5%	12% 5%		10% 5%
Profession		Employee	Freelancer	Student-worker	Student	Unemployed		Retired
	S1 S2	52% 54%	13% 5%	8% 13%	24% 25%	2% 2%		1% 1%
Amount spent on the last order		EUR 1–19	EUR 20–39	EUR 40–59	EUR 60–79	EUR 80–99		EUR ≥100
	S1 S2	9% 7%	24% 25%	21% 22%	15% 31%	5% 9%		26% 24%
Units ordered		1	2	3	4	5		≥ 6
	S1 S2	42% 30%	22% 22%	13% 17%	6% 8%	5% 7%		12% 16%
Online store type		Fashion, Footwear and Accessories	Marketplace	Food	Beauty and Health	Technology and Entertainment	Others	Sports
	S1 S2	39% 37%	13% 33%	9% 3%	10% 5%	10% 8%	10% 3%	9% 11%

* S1 is the first sample. ** S2 is the second sample.

. The data refer to and characterize the last purchase made in the online store where the respondents make their most frequent purchases. The sample consists of 441 for the second sample and 407 for the first sample of duly validated respondents.

Characterizing the sample shows that the majority of those surveyed are female (58% or 60%), while males make up 42% or 40%. More than half, 54% or 52%, work for others (employees), 25% or 24% are students, and 13% or 8% are student workers. The remaining 8%, or 16%, are freelancers, unemployed, and retired. In terms of how much they spent on their last shopping trip, there is no consistent pattern across the six groups, although they can be divided into two subsets: one with four groups and one with two groups. As for the number of products bought in the last order, a decreasing pattern was observed, with 30%, or 42%, buying one product and 7%, or 5%, buying five products. Nevertheless, there is the exception of the last group, which represents 16% or 12% of the respondents. When it comes to the type of online store, 37% or 39% prefer fashion, footwear, and accessories stores; 33% or 13% favor marketplaces; and the rest go to food, beauty and health, technology and entertainment, sports, and other stores.

5.2. Results from the PLS-SEM

The results presented in this section refer to the second sample, in the interest of simplicity. Nonetheless, the two samples de facto underwent the same procedure, and consequently, the outcomes are comparable.

No multivariate outliers were identified using Mahalanobis distance [389], and there were no missing values. PLS-SEM does not enforce distribution assumptions, but a key issue is to confirm that data do not exhibit significant deviations from the normal distribution, as severe non-normal data may affect the determination of parameter significance [390]. Therefore, variables were assessed for skewness and kurtosis, and no variables with asymmetric distributions or highly peaked distributions were found. The systematic assessment procedure for the outcomes of PLS-SEM comprises two phases and conforms to the guidance set forth by Hair et al. [390]: (I phase) evaluation of the measurement model; (II phase) evaluation of the structural model. The evaluation of the measurement model comprises the evaluation of both reflective and formative measurement models. The reflective measurement model assesses the reliability of individual indicators, internal consistency reliability, convergent validity, and discriminant validity. The formative measurement models consist of assessing the convergent validity, the collinearity between the indicators, and the importance of the external weights. The structural model evaluates collinearity through VIF, the significance, and relevance of the relationships in the model through path coefficients, and the explanatory power through coefficients of determination (R²). It is worth noting that in the conceptual model, all constructs are reflective except for the S&P construct, which is formative.

5.2.1. Measurement Model-Reflective

Table 5. Outer loadings, internal consistency, and convergent validity–measurement model.

Constructs	Indicators	Outer Loadings	Cronbach’s Alpha	rho_A	Composite Reliability	AVE
WebC	1	0.847	0.885	0.892	0.921	0.744
	2	0.856
	3	0.887
	4	0.858
SQ	1	0.835	0.833	0.833	0.889	0.666
	2	0.831
	3	0.805
	4	0.793
PP	1	0.654	0.695	0.748	0.812	0.528
	2	0.851
	3	0.817
	4	0.539
PR	1	0.826	0.840	0.846	0.904	0.759
	2	0.904
	3	0.881
PV	1	0.862	0.809	0.832	0.876	0.644
	2	0.610
	3	0.874
	4	0.834
Trust	1	0.918	0.911	0.913	0.938	0.790
	2	0.890
	3	0.893
	4	0.853
CX	1	0.874	0.855	0.873	0.903	0.700
	2	0.896
	3	0.715
	4	0.851
CS	1	0.923	0.947	0.947	0.962	0.862
	2	0.942
	3	0.923
	4	0.927
CL	1	0.743	0.851	0.870	0.899	0.691
	2	0.890
	3	0.852
	4	0.833
WOM	1	0.910	0.932	0.950	0.951	0.829
	2	0.927
	3	0.885
	4	0.919

contains the values used to assess the inclusion of outer loadings, internal consistency, and convergent validity.

The first step in the examination of a reflective measurement model is the examination of the external loadings of the indicators (see Table 5). Outer loadings less than 0.40 are excluded, while those greater than or equal to 0.7 are kept. Within the above values, the reliability of internal consistency and the convergent validity of the construct are analyzed. In these conditions, there are two constructs: PP, which has two indicators, and PV, which has one indicator. The indicators were kept because their removal caused a major impact on AVE.

Internal consistency and reliability are the second criteria for evaluation. The standard measure of internal consistency and reliability in formal settings is Cronbach’s alpha (see Table 5). Hair et al. [390] suggest that values between 0.60 and 0.70 are adequate for exploratory research, whereas values ranging from 0.70 to 0.90 are considered satisfactory for more mature research. Values show that all constructs are dependable, apart from the salience PP, yet still allow for further analysis of the model.

Convergent validity describes the degree of positive correlation between an indicator and other indicators that assess the same construct. For the average variance extracted (AVE), all the values are greater than 0.5, therefore above the required minimum.

Discriminant validity assesses whether a construct can be distinguished from other constructs through empirical evidence. The use of the heterotrait-monotrait ratio (HTMT) is necessary. A threshold of 0.85 is recommended for path models with conceptually distinct constructs, as an HTMT value greater than 0.90 suggests inadequate discriminant validity. As all values are less than 0.785, discriminant validity is ensured.

5.2.2. Measurement Model-Formative

The presence of severe collinearity between formative indicators has a major impact on the calculation of weights and the determination of statistical significance. Confirming that each indicator’s VIF (variance inflation factor) value is within an acceptable range is critical to evaluating a model’s performance. A VIF value greater than 3 indicates potential multicollinearity, but, more importantly, a value greater than 5 requires corrective action. The VIF value for the S&P is 3.78, indicating that no corrective measures are necessary.

To carry out the significance test for the outer weights, a bootstrap sampling algorithm with 10,000 subsamples was used. Results showed that the two indicators’ weights from S&P were significant, providing empirical support for retaining them.

5.2.3. Structural Model

To evaluate collinearity, it is necessary to analyze each construct indicator separately for each subset of the structural model (

Table 6. Inner VIF Values—structural model.

	CL	CS	CX	PP	PR	PV	S&P	SQ	Trust	WOM	WebC
CL
CS	1.000									1.000
CX		1.000
PP						1.000
PR									1.238
PV			1.199
S&P					1.185				1.238
SQ			1.477		1.612
Trust			1.428
WOM
WebC					1.559

).

From the results, it can be confidently concluded that collinearity is not a significant problem in the structural model, as the VIF values for all predictor constructs are significantly lower than the threshold of 5. The above results are encouraging, given that the absence of collinearity between the constructs means that they are measuring different entities and do not overlap. Therefore, the analysis of the structural model can proceed.

Standardized path coefficients in the structural model usually fall between −1 and +1, representing theorized relationships among constructs. A coefficient near 1 indicates a strong positive correlation, while a coefficient near −1 indicates a significant negative correlation. Constructs with coefficients close to zero have weak relationships and are less meaningful in the prediction of other constructs (see

Table 7. Path coefficients–structural model.

	Sample Mean		T Statistics		p-Values
	S1	S2	S1	S2	S1	S2
CS -> CL	0.516	0.516	0.036	0.040	0.000	0.000
CS -> WOM	0.686	0.686	0.035	0.045	0.000	0.000
CX -> CS	0.802	0.802	0.026	0.050	0.000	0.000
PP -> PV	0.502	0.502	0.042	0.041	0.000	0.000
PR -> Trust	0.394	0.394	0.048	0.047	0.000	0.000
PV -> CX	0.434	0.434	0.049	0.038	0.000	0.000
S&P -> PR	0.196	0.196	0.047	0.053	0.000	0.000
S&P -> Trust	0.291	0.291	0.053	0.049	0.000	0.000
SQ -> CX	0.148	0.148	0.047	0.046	0.002	0.000
SQ -> PR	0.285	0.285	0.064	0.056	0.000	0.000
Trust -> CX	0.330	0.330	0.05	0.047	0.000	0.000
WebC -> PR	0.184	0.184	0.061	0.058	0.003	0.000

).

The values in the table confirm the existence and significance of the relationships between the constructs, as the path coefficients are all statistically significant for S1 and S2, as confirmed by the bootstrapping procedure. It must be emphasized that the path coefficients express the research hypotheses. Consequently, all the research hypotheses are supported. It should be noted that the path coefficients for the two samples are similar, proving the model’s reliability and consistency.

By looking at the magnitude of the path coefficients, it is also possible to identify the strongest relationships in the model. This is of great importance, as it helps to identify the constructs and relationships on which the resources should be focused. Starting with the key constructs identified in the introductory section–CX, CS, and CL–the relationships between CS and CL, CS and WOM, CX and CS, PP and PV, PR, and trust are strong. The remaining relationships are less strong. Since CS is the most important construct, CX is the single predictive factor to focus on. With a value of 0.690 or 0.802, this relationship is quite robust. In turn, CX is preceded by SQ, Trust, and PV. Accordingly, trust (0.379 or 0.330), SQ (0.291 or 0.148), and PV (0.274 or 0.434) are what should be looked at. In other words, more resources should be devoted to the building of trust than to the provision of quality service. Conversely, a comparison between SQ and PV does not permit the differentiation of which is of greater importance to customers.

The assessment of the structural model requires a further step, namely the evaluation of its explanatory power (see

Table 8. Coefficient of determination–structural model.

		CL	CS	CX	PR	PV	Trust	WOM
R Square	S1	0.267	0.644	0.542	0.290	0.252	0.328	0.470
	S2	0.286	0.476	0.548	0.323	0.361	0.360	0.297
R Square Adjusted	S1	0.286	0.476	0.548	0.323	0.361	0.360	0.297
	S2	0.285	0.475	0.545	0.318	0.360	0.355	0.295

). It is a way to measure how well the model fits the data, quantifying the strength of the relationships. The coefficient of determination, or R², is the metric that is frequently used. R² expresses the percentage of variance in the dependent construct that is explainable by the independent constructs associated with it.

As can be observed, the CX and CL constructs have moderate values, and the other constructs have weaker values. Accordingly, CX is explained by almost 55%, or 54%, and CS by almost 48%, or 64%. The other constructs are explained by roughly 30% (the exception is WOM).

These results are far from discouraging. On the contrary, the most important observation of this study is the explanation of CX (0.542 or 548) and CS (0.476 or 0.644). As already mentioned, they are among the most important. Effect sizes f² were also used as a measure of the strength of the relationships in the structural model. The previous findings were confirmed by the results of this test. The model’s predictive power was also assessed using the Q² statistic, and results showed that the model has higher predictive power than the naïve LM benchmark model.

After conducting all necessary tests according to the guidelines for PLS-SEM, the next step will be the examination of the research hypotheses (

Table 9. Assessment of research hypotheses.

Research Hypotheses	Value				Result
	S1	S2	S1	S2
H1: WebC reduces the PR	0.184	0.214	0.000	0.003	Supported
H2.1: SQ reduces PR	0.285	0.227	0.000	0.000	Supported
H2.2: SQ positively influences CX	0.148	0.291	0.000	0.002	Supported
H3.1: S&P positively influences Trust	0.291	0.250	0.000	0.000	Supported
H3.2: S&P reduce PR	0.196	0.284	0.000	0.000	Supported
H4: PP positively influences PV	0.502	0.601	0.000	0.000	Supported
H5: PR positively influences Trust	0.394	0.444	0.000	0.000	Supported
H6: PV positively influences CX	0.434	0.274	0.000	0.000	Supported
H7: Trust positively influences CX	0.330	0.379	0.000	0.000	Supported
H8: CX positively influences CS	0.802	0.690	0.000	0.000	Supported
H9: CS positively influences WOM	0.686	0.545	0.000	0.000	Supported
H10: CS positively influences CL	0.516	0.535	0.000	0.000	Supported

).

Given the results obtained and after all the validations, the following observations can be made: all the proposed research hypotheses have been confirmed, and all the relationships in the model have been verified in the two samples. As a result, it is possible to state that the identified constructs and the relationships between them form a holistic e-commerce model.

6. Discussion

This research provides an in-depth and better understanding of a holistic conceptual model, that is representative of the repeated buying decisions in e-commerce. New findings arose, and the improvements are unmistakable. The implications of the results are discussed and confronted with the literature review.

The results are an indication that CX may be the only antecedent construct of CS, as has been suggested by Maklan and Klaus [41], Rose et al. [44], Rose et al. [45], Choi et al. [46], Klaus and Maklan [47], Fatma [48], Martin et al. [49], Parise et al. [50], Roy et al. [51], Pandey and Chawla [52], Roy [53], Terblanche [54], Rahman et al. [55], Banik and Gao [56], and others. Notably, the strength of the relationship between CX and CS has the highest value in the model. The preceding statement contradicts a significant number of studies that link CX to CL or WOM. In a tentative calculation, it was found that the relationships between CX and CL and CX and WOM are weak, with values of 0.13 and 0.19, respectively. This suggests that CX should only precede CS.

Two other relationships are worth noting: CS and CL, as well as between CS and WOM. Both showed strong values. Beginning with the relationship between CS and CL, the proposed model confirms the key contribution of Fornell et al. [118] to the ACSI. In addition, the relationship is in accordance with the numerous research studies that point to his existence, including those of Martin et al. [49], Morgeson et al. [143], and Eklof et al. [33]. More significant is the observation that the only precursor to CL is CS, which is supported by the many aforementioned research studies [33,42,44,96,121,123,134,146,206,207,208]. Several authors have also reported a relationship between SQ and CL that was not considered in the conceptual model. Tentative calculations have determined that this relationship does not occur. Concerning the second relationship between CS and WOM, the work of Singh and Söderlund [96] is noteworthy, as they have proposed and confirmed it. The most striking aspect is the large number of authors who promote the relationship between CX and CL in favor of CS and CL, among them Brown [177], Maklan and Klaus [41], Klaus and Maklan [47], Klaus [104], Bilgihan, et al. [92], Kawaf and Tagg [145], Cambra-Fierro et al. [178], Roy et al. [179], Siqueira et al. [97], Rahman et al. [55], Moliner-Tena et al. [148], Banik and Gao [56], and Kacprzak and Hensel [70], for example. The conceptual model does not include the relationship between CX and CL, and tentative results did not support this relationship.

A significant and strong relationship was also established between PP and PV, acknowledging Zeithaml et al. [247], Chen and Dubinsky [274], and Kim et al. [322]’s indications. Both PP and PV were reported as preceding anteceded CS, or CL, in the bibliography. The conceptual model did not include these relationships, and additional testing proved that they failed to exist. However, the relationship between PV and CX has been considered, and it has been found to have empirical support but is weak. When considering the precedents of the CX, namely the PV, the Trust, and the SQ, it is important to interpret the results with some caution. Although there is support in the literature for these relationships and the results are statistically significant, the relationships are weak. The same was true for the relationships between S&P and PR, S&P and Trust, WebC and PR, and SQ and PR.

It should be stressed that the results found for weak relationships are not entirely discouraging or unsatisfactory. First and foremost, these findings do not challenge the holistic conceptual model or the validity of the relationships; rather, what is found are some weak relationships resulting from the underlying data and the selected scales. Therefore, we cannot exclude the possibility that these two factors may have contributed to some weak relationships. Regarding the data and the premise of the existence of the model, there is always a possibility that the sample may not be a true reflection of the population. Furthermore, the occurrence of population heterogeneity for these constructs could also lead to these outcomes. As far as the scales are concerned, the reflections are different. First and foremost, it should be noted that the scales meet the compulsory requirements. But the development of a scale is a complex and time-consuming process, consisting of four phases [41,84,103]: (1) scale generation and initial purification (item generation); (2) scale purification through EFA (purification and refinement); (3) reliability and validity assessment (refinement and final validation); (4) scale validation. As most of the scales have been adjusted due to the reasons previously outlined, it is reasonable to assume that some degree of uncertainty has been embedded in the relationships. There is one other possible explanation for this occurrence as well. This is an attempt to model a complex reality, i.e., consumer behavior in online shopping. Since a model is a simplified representation of reality, it is important to consider whether the proposed holistic conceptual model may be too simplistic. It is crucial to avoid oversimplification and ensure that the model accurately reflects the complexity of the subject matter. If this is the case, then the omission of certain constructs may have had an impact on the results obtained. As a final point, it is also noteworthy to mention the occurrence of endogeneity and that the model may account for the intrinsic variations in the preferences or behavior of consumers. In addition, Table 4 also illustrates the diversity of online store types and products. Stores can be specialized or generic, such as marketplaces or single brands; products can have a high or low level of involvement. This diversity could potentially have shaped the weak relationships.

7. Conclusions

After over two decades of researching e-commerce, there is still a lack of a holistic model that incorporates and connects the major marketing constructs. Additionally, it is crucial to acknowledge that these constructs and relationships must be substantiated by classic theoretical frameworks and models. It is equally essential to have models that can be applied in both academic and practical contexts. Based on these premises and the theory-in-use, this research identified those constructs and developed a comprehensive conceptual model that represents the process of purchasing online. The identified constructs are customer experience, customer satisfaction, customer loyalty, word-of-mouth, trust, perceived risk, security and privacy, web content, perceived price, perceived value, and service quality.

An equally significant aspect is the identification and quantification of the relationship between the constructs. The conceptual model, illustrated in Figure 1, includes 12 relationships and was validated using PLS-SEM. Given the length of the available scales for assessing the constructs and their overlap, the scales were adjusted to develop a comprehensive model. The findings are thought-provoking and striking. The first relationship postulates that the information made available and formalized as the web content construct influences consumers’ perceptions of risk. Therefore, the more up-to-date, complete, understandable, and appropriate the information, the lower the perceived risk. Concerning service quality, two relationships may be identified: an inverse correlation between service quality and perceived risk and a positive correlation between service quality and customer experience. Furthermore, the security and privacy constructs are associated with two distinct relationships. The higher the security and privacy levels are, the greater the consumer trust in the online store and the lower the perceived risk of the purchase. The construct of perceived price is only related to that of perceived value. When there is greater clarity and understanding of both the price in question and the benefits, clarity, and simplicity of the transaction, there is a greater perception of value. In essence, perceived risk is solely a function of trust, and the lower the perception of risk throughout the entire purchase process, the greater the consumer’s trust in the online store. Returning to the customer experience construct, two relationships warrant particular attention. The first to be examined is the relationship between perceived value and customer experience. A consumer’s experience can be positively influenced by a perception of high value, as indicated by their expectations, the value received, the value they perceive to have been paid for, the value offered by the competition, and the effort spent. Secondly, the relationship with trust must be considered. A positive reputation and activities that are beneficial to the customer will result in a more positive experience. Conversely, if there is a lack of honesty in perception, reputation, or activities that are detrimental to the customer, the experience will be less positive. The three most significant relationships identified in this study align with the existing literature. These include the relationship between customer experience and satisfaction and the relationship between satisfaction with word-of-mouth and loyalty. The findings imply that an enhanced customer experience leads to increased satisfaction, which in turn fosters greater loyalty to the online store and increases the likelihood of positive opinions being expressed about the online store.

The findings presented in this study can be applied to a wide range of online stores. The added results offer somewhat surprising insights, providing a novel perspective on the relationship between various constructs. Firstly, customer experience is only related to satisfaction, with no strong relationship being confirmed between customer experience and loyalty or word-of-mouth, as indicated in the literature. Furthermore, no relationships were found between service quality, perceived price, perceived value, and loyalty or satisfaction.

7.1. Theoretical Contributions

This research proposes a new and complete model with eleven marketing constructs and twelve relationships. It is a new way of looking at the purchasing process in online stores. Two other major contributions are noteworthy. Firstly, the identification of unique and distinct constructs. The results of the PLS-SEM analysis indicate discriminant validity, which implies that the constructs do not overlap and represent unique and distinct phenomena. This allows for the redefinition of constructs that in the literature have been overlapping, partially measuring the same phenomena, and necessarily leading to correlations between them. Secondly, the scales were adapted to be feasibly operational, enabling consumers to respond to eleven constructs simultaneously within a reasonable amount of time and with moderate effort. This is verified by the results of the PLS-SEM, which revealed the adequacy of the scales through composite reliability. Thirdly, it reveals the pivotal role of the customer experience construct and the relationships between the customer experience and satisfaction, satisfaction and loyalty, and satisfaction and word-of-mouth. Furthermore, this research identified and validated a number of antecedents to the customer experience, including service quality, web content, security and privacy, perceived price, perceived risk, trust, and perceived value. Additionally, the relationship between these antecedents was also identified and validated. This represents an exciting theoretical advance, as the model is based on the application of classic theories and frameworks, using theories-in-use, and its holistic nature ensures a unified perspective on online purchasing behavior, with customer experience at its core.

7.2. Managerial Implications

This research is relevant for marketing professionals and managers as well because it will enable them to direct their resources toward the constructs that have been identified and to devise strategies in line with the holistic conceptual model being proposed. Furthermore, as both the constructs and the relationships between them can be ranked by their importance to consumers, it is feasible to allocate resources and concentrate efforts by prioritizing the most relevant ones. As a result, companies can concentrate on operational activities and develop and set strategies that are valued by consumers and backed by empirical evidence. Two significant findings emerged from the literature review. The first is evidence that there is a correlation between customer satisfaction and business performance. The second is that the marketing cost to loyal customers is less than the cost to market to new customers, and word-of-mouth is one of the most persuasive sources of information. Following the first evidence, marketing managers must ensure that activities that support the relationships preceding satisfaction are aligned with customer requirements. In descending order of importance to customers, the following relationships stand out: perceived price and perceived value; perceived risk and trust; perceived value and customer experience; security and privacy with perceived risk and with trust; quality of service with customer experience and with perceived risk; trust with customer experience; and web content with perceived risk. It is also crucial to acknowledge that the customer experience represents the sole construct that precedes satisfaction. Consequently, customers who have a positive experience will move on to a state of satisfaction. In turn, the customer experience is contingent upon the quality of the service provided, the perceived value of the purchase, and the level of trust that the customer has in the online store. Marketing managers have to guarantee this sequence right up to the satisfaction construct. Moving onto the second piece of evidence, and after ensuring satisfaction, marketing actions must be taken to promote loyalty and word-of-mouth.

7.3. Limitations and Future Research Scope

Like all research, this one also has a few strengths and limitations. The research’s first limitation pertains to the data. Specifically, the non-random sampling process, the time constraint, and the restricted geographic region. A second limitation stems from the lack of inclusion of all potential constructs. Although it is not feasible to incorporate them all, the absence of some constructs may have reduced the predictive power of the model. The third limitation arises from the adaptation of the scales that needed to be made so that consumers would be able to complete the questionnaire. As mentioned above, developing scales is a complex and time-consuming process. Therefore, it is reasonable to assume that there is some variability in the outcome as a result of this limitation. Undoubtedly, the research limitations provide an opportunity for future studies to build upon and expand upon the findings of this research.

While these limitations may not be considered critical, they do serve to highlight areas for improvement and further exploration. Therefore, it is imperative to extend the application of the model to additional datasets to establish its validity and generalize, and this should be done across diverse geographic regions and over extended periods. To achieve the objective of developing a set of concise scales to measure the online shopping process, it is essential to replicate the process of scale creation. The goal is to establish a set of scales that can capture the whole purchasing experience in online stores while minimizing bias and variability. This is a crucial step in ensuring that the construct is accurately measured. Lastly, the inclusion of more constructs in the model serves to enhance its comprehensiveness and accuracy. Specifically, the inclusion of the Digital Channels and Brand Equity constructs will improve its comprehensive coverage, while the addition of the Terms and Conditions, Customer Support, Catalog, and Promotions constructs will increase its accuracy.