Purpose

Drawing on parasocial interaction theory, this study examines how key attributes of artificial intelligence (AI)-enabled chatbot services, including functionality, trustworthiness, efficiency, human-likeness, responsiveness and reference to the service, affect consumers’ perceived interactivity and foster brand experience, thereby enhancing parasocial relationships between brands and customers.

Design/methodology/approach

Structural equation modeling was used to analyze the data. A total of 278 valid responses were obtained from an online questionnaire survey.

Findings

The attributes of AI-enabled chatbot services, namely trustworthiness, efficiency, human-likeness and reference to the service, positively affect their perceived interactivity. Furthermore, brand experience mediates the relationship between perceived interactivity and parasocial relationships.

Practical implications

AI-enabled chatbots serve as virtual frontline employees capable of simulating human-like interpersonal engagement, thereby enhancing brand experiences. When consumers perceive a high level of interactivity with an AI-enabled chatbot, they develop more favorable impressions of the brand, which in turn strengthens the parasocial bond between the brand and its customers. These findings provide practical guidance for designing AI-enabled chatbot services that foster deeper brand–customer relationships.

Originality/value

While previous research on AI-enabled chatbots has focused primarily on anthropomorphic features and service quality, limited attention has been paid to how specific service attributes contribute to building parasocial relationships in the context of online retail. This study addresses this gap by demonstrating how chatbot–customer interactions can strengthen brand–customer connections. Our findings provide valuable insight for managers across various industries and offer guidance on the application of AI technologies in digital transformation strategies to enhance customer service.

Artificial intelligence (AI)–enabled chatbots have become increasingly prominent in human–computer interactions, especially in enterprise applications that emphasize customer engagement (Liu, 2025). These chatbots, powered by technologies such as natural language understanding and machine learning, simulate human dialog to support interactive communication and improve service efficiency (Yang et al., 2023; Chen et al., 2025). In online retailing, AI-enabled service chatbots help firms automate customer interactions and deliver streamlined services that align with evolving consumer expectations. Unlike human agents or traditional self-service technologies, AI-enabled chatbots offer distinct advantages. They operate continuously, providing 24/7 support and responsiveness, and improve through self-learning (Truong and Chen, 2025). While human agents may offer deeper emotional engagement, chatbots simulate human-like interactions and outperform humans in areas such as computational accuracy and data processing. These capabilities make chatbots unique tools for service delivery.

Chatbot adoption spans various industries, including banking (Cheng et al., 2024), e-commerce (Song and Shin, 2024), tourism (Nazir et al., 2023; Zhang et al., 2024), home device (Shi et al., 2025) and health care (Schillaci et al., 2024; Ramya and Alur, 2025). Although some studies have examined chatbot features in online retail (Chen et al., 2022, 2023; Cheng et al., 2022b), most have focused on anthropomorphism or service quality. Research specifically addressing the distinctive features of AI-enabled service chatbots in online retail remains limited. With chatbots increasingly perceived by consumers as conversational partners, understanding how their attributes influence user responses has become critical (Kim and Hur, 2024).

Parasocial interaction (PSI) theory, introduced by Horton and Wohl (1956), explains how audiences form imagined interpersonal relationships with media figures. These one-sided connections create a sense of intimacy and reciprocity, despite the absence of real interactions (Feng et al., 2025). In the context of chatbot use, customers engage with preprogrammed agents to resolve issues or obtain information, often perceiving interactions as socially meaningful. With advances in natural language processing and context-aware responsiveness, such engagements increasingly resemble human communication and may foster parasocial relationships.

Although the computers are social actors (CASA) paradigm (Reeves and Nass, 1996) provides a foundation for understanding social responses to technology, PSI theory offers a more appropriate lens for capturing the deeper, ongoing bonds that can form between customers and AI-enabled chatbots. While CASA addresses surface-level social cues, PSI emphasizes the development of affective ties, which can extend to brand relationships. These parasocial relationships may enhance brand messaging and loyalty because brands use chatbots to mediate customer interactions (Nadroo et al., 2025). Drawing on PSI theory, this study examines whether customers’ perceived interactivity with AI-enabled chatbots influences the formation of parasocial relationships with brands. Accordingly, it addresses the following research question:

RQ1.

How does customers’ perceived interactivity influence their parasocial relationships with a brand when using AI-based chatbot services?

This study applies PSI theory to examine whether perceived interactivity with AI-enabled chatbots fosters parasocial relationships with brands. It focuses on six key chatbot attributes: functionality, trustworthiness, efficiency, human-likeness, responsiveness and reference to the service. Using Uniqlo’s customer service chatbot as the research context, this study introduces brand experience as a mediating variable to explore how perceived interactivity influences parasocial connections. The findings aim to extend PSI theory to AI-enabled service environments and provide practical guidance for organizations seeking to strengthen brand–customer relationships through chatbot technologies.

Originally developed to explain human–human mediated communication, PSI theory has increasingly been applied to human–computer interactions, including those involving AI-enabled agents, such as chatbots (Li and Wang, 2023). In online retail contexts, consumers frequently engage in conversational exchanges with AI-enabled chatbots to resolve problems. Similar to real-world relationships, parasocial bonds are formed through repeated interactions (Farivar et al., 2021). When chatbots respond effectively to customer needs, users may develop emotional attachments through increased reliance, resulting in one-sided yet asymmetrical emotional ties (Sheng et al., 2025).

As shown in Web Appendix 1, recent studies have applied PSI theory to examine customer interactions with chatbots. For example, Duong et al. (2025) found that travelers formed parasocial connections with ChatGPT during AI-assisted experiences. Li and Wang (2023) demonstrated that PSI mediates the relationship between a chatbot’s language style and users’ continued usage intentions. Ramya and Alur (2025) emphasized that this relationship can be strengthened through personalized service.

In the context of AI-enabled chatbots, interactions mimic two-way conversations with brands, shaping customers’ evaluations of those brands. Chatbots that serve to complement or substitute human frontline employees are increasingly perceived as brand representatives in online retail contexts (Li and Wang, 2023). Through ongoing dialog, retailers sustain brand–customer relationships via parasocial interactions with chatbots (Tsai et al., 2021). PSI theory provides a valuable framework for understanding the socioemotional outcomes of digital engagement, particularly in relation to AI-enabled interactions (Duong et al., 2025; Noor et al., 2022). The relationships formed, while asymmetrical, can generate authentic and lasting psychological bonds (Tyrväinen and Karjaluoto, 2025). In the context of online retail, chatbot interactions increasingly resemble human dialog and shape brand perceptions. Thus, this study adopts the perspective of brand–customer parasocial relationships to explore how AI-enabled chatbot services influence customer engagement.

Web Appendix 2 summarizes prior work identifying key chatbot attributes across service contexts. In customer service settings, Meyer-Waarden et al. (2020) highlighted tangibility, competence, reliability, responsiveness, empathy and credibility, while Daisy (2020) proposed functionality, trustworthiness, safety, efficiency, graphic appearance, humanity, empathy and responsiveness. Subsequent studies have refined these lists of attributes. For example, Johari and Nohuddin (2021) emphasized functionality, efficiency and humanity as the most valued characteristics of effective chatbots. Other researchers have introduced domain-specific models, such as the SERVBOT framework for social chatbots (Kharub et al., 2021). In mental health care, Kim et al. (2025) classified chatbot attributes into internal functions, such as counseling; external elements, such as image and emotion; and additional features, including facial expressions and empathetic communication.

These findings suggest that chatbot attributes can be grouped into two main categories. The first category centers on anthropomorphic features, including human-like expressions, language and behavior, which help make interactions feel natural and comfortable without misleading users into perceiving the chatbots as human (Chakraborty et al., 2024; Cheng et al., 2022b; Gkinko and Elbanna, 2022; Jeon and Kim, 2025; Liu et al., 2024; Sarraf et al., 2024). Although anthropomorphism is essential in designing engaging chatbot experiences, functionality is equally critical to ensure usefulness across service contexts (Chen et al., 2023; Dwivedi et al., 2023; Kim and Im, 2023).

The second category focuses on service quality. While studies have often evaluated chatbot performance using traditional service quality dimensions, such as reliability and responsiveness (Li et al., 2021; Kharub et al., 2021), recent research has argued that these measures reflect outcomes rather than intrinsic characteristics. Chen et al. (2023) suggested that chatbot service quality should be assessed based on its effectiveness in meeting user expectations. Moreover, Sonntag et al. (2023) further emphasized that chatbot designs vary by industry, reflecting differences in user needs and service environments.

This study focuses on the online retail industry for two main reasons. First, although physical retail remains dominant, online shopping continues to grow rapidly. Global e-commerce sales are projected to reach USD 9.4 trillion by 2026 (Sabanoglu, 2023), highlighting the economic impact of online shopping. Second, online retail can benefit significantly from chatbot services. AI-enabled chatbots support real-time customer inquiries and improve search visibility through keyword-based, search engine optimization (SEO)–friendly content. They also enable multilingual support and consistent communication across regions (Nadroo et al., 2025).

Based on Daisy’s (2020) classification of chatbot attributes for customer relationship management, this study adopts five attributes aligned with online retailing, namely functionality, trustworthiness, efficiency, human-likeness and responsiveness, to represent AI-enabled service chatbots. These features are frequently cited as essential in the literature. For example, Johari and Nohuddin (2021) identified functionality, efficiency and humanity as the most valued chatbot traits. Li et al. (2021) emphasized the importance of responsiveness for timely interactions and trustworthiness for building user confidence. Other attributes, such as safety, graphic appearance and empathy, are excluded from this study. This is because Uniqlo’s chatbot does not require users to disclose personal data, making safety less relevant. In addition, its interface is text-based, so its graphic appearance is not applicable. Empathy, meanwhile, is treated in terms of human-likeness, since human-like interactions often convey empathy through emotional expression (Cai et al., 2022).

Reference to the service is also included as a key attribute. This refers to a chatbot’s ability to guide customers to use available resources, such as hyperlinks, options and on-screen elements (Borsci et al., 2022). Reference to the service not only provides relevant support but also helps foster customer relationships (Borsci et al., 2022), since chatbots also function as marketing tools that assist users in navigating service environments (McNeal and Newyear, 2013). Accordingly, this study includes six chatbot attributes: functionality, efficiency, human-likeness, trustworthiness, responsiveness and reference to the service. Figure 1 illustrates the research framework of this study.

Figure 1
A research framework diagram linking chatbot attributes to loyalty.The research framework diagram shows a dashed, rounded rectangle on the left labeled “Service chatbot’s attributes”. Inside it, six oval shapes are stacked vertically, labeled “Functionality”, “Trustworthiness”, “Efficiency”, “Human-likeness”, “Responsiveness”, and “Reference to the service”, from top to bottom. Arrows labeled “H 1”, “H 2”, “H 3”, “H 4”, “H 5”, and “H 6” extend from these six ovals toward a central oval labeled “Perceived interactivity”. From “Perceived interactivity”, an arrow labeled “H 7” points to an oval on the right labeled “Brand experience”. Another arrow labeled “H 8” points downward from “Perceived interactivity” to an oval labeled “Parasocial relationships”. From “Brand experience”, an arrow labeled “H 9” points downward to “Parasocial relationships”. To the right of “Brand experience” is the text “Validated by prior research” with a vertical arrow pointing downward to an oval labeled “Brand loyalty”. To the right of “Parasocial relationships” is the text “Validated by prior research” with a vertical arrow pointing upward to the same oval labeled “Brand loyalty”. At the bottom right, a dashed, rounded rectangle labeled “Control variables” contains four ovals labeled “Education”, “Gender”, “Age”, and “Chatbot’s use duration”, from left to right. Arrows extend upward from these four control variable ovals toward “Parasocial relationships”.

The research framework of this study

Figure 1
A research framework diagram linking chatbot attributes to loyalty.The research framework diagram shows a dashed, rounded rectangle on the left labeled “Service chatbot’s attributes”. Inside it, six oval shapes are stacked vertically, labeled “Functionality”, “Trustworthiness”, “Efficiency”, “Human-likeness”, “Responsiveness”, and “Reference to the service”, from top to bottom. Arrows labeled “H 1”, “H 2”, “H 3”, “H 4”, “H 5”, and “H 6” extend from these six ovals toward a central oval labeled “Perceived interactivity”. From “Perceived interactivity”, an arrow labeled “H 7” points to an oval on the right labeled “Brand experience”. Another arrow labeled “H 8” points downward from “Perceived interactivity” to an oval labeled “Parasocial relationships”. From “Brand experience”, an arrow labeled “H 9” points downward to “Parasocial relationships”. To the right of “Brand experience” is the text “Validated by prior research” with a vertical arrow pointing downward to an oval labeled “Brand loyalty”. To the right of “Parasocial relationships” is the text “Validated by prior research” with a vertical arrow pointing upward to the same oval labeled “Brand loyalty”. At the bottom right, a dashed, rounded rectangle labeled “Control variables” contains four ovals labeled “Education”, “Gender”, “Age”, and “Chatbot’s use duration”, from left to right. Arrows extend upward from these four control variable ovals toward “Parasocial relationships”.

The research framework of this study

Close modal

Perceived interactivity refers to an individual’s subjective experience of two-way communication during an interaction (Sundar et al., 2016). It reflects not only exchanges of messages but also the perceived responsiveness and relevance of system feedback (Labrecque, 2014). Factors such as prompt responses and the relevance of messages to previous user input are central to shaping interactivity perceptions (Song and Zinkhan, 2008). In this study, interactivity is conceptualized as a perceptual variable that captures consumers’ evaluations of bidirectional communications with chatbot’s services.

Meerschman and Verkeyn (2019) defined functionality as a chatbot’s ability to perform tasks that consumers expect, focusing solely on task execution without considering the quality of input commands or output responses. Prior research has shown that chatbots can handle most frequently asked questions (Suthar, 2020) and retrieve accurate responses (Suhaili et al., 2021). Sundar et al. (2016) emphasized that real-time functionality can enhance users’ perceptions of interactivity. When AI-enabled chatbots effectively fulfill their intended tasks, users are more likely to perceive interactions as interactive.

H1.

The functionality of AI-enabled chatbot services exerts a positive influence on perceived interactivity.

Trustworthiness refers to the perception of whether an entity is dependable and competent (Mayer et al., 1995). Prior studies have conceptualized trustworthiness as a multidimensional construct commonly comprising ability, benevolence and integrity (Cai et al., 2022; Torres-Moraga and Barra, 2023). In service technology contexts, trustworthiness is based on perceived competence (ability), reliability (benevolence) and value alignment (integrity) (Agnihotri and Bhattacharya, 2024). Among these, this study focuses on trustworthiness as a perceived aspect of ability, which refers to whether customers believe an AI-enabled chatbot can deliver accurate and helpful responses to inquiries (Cai et al., 2022).

Chatbots perceived as trustworthy increase users’ reliance on service interactions and decision-making processes (Cheng et al., 2022a). When chatbot responses are perceived as high in quality, users are more likely to experience a greater sense of interactivity (Meerschman and Verkeyn, 2019; Tsai, 2011). Trustworthiness is a crucial feature to facilitate perceived interactivity because reliable agents foster practical and desirable interactive experiences (Miller and Larson, 2005). In this study, trustworthiness captures the extent to which users believe AI-enabled chatbots are competent in solving problems.

H2.

The trustworthiness of AI-enabled chatbot services exerts a positive influence on perceived interactivity.

Firms are increasingly using chatbots to deliver real-time, around-the-clock customer service (Zhang, 2021). Efficiency, originally defined as a website’s ability to help users quickly retrieve relevant information with minimal effort (Teo et al., 2003), is extended here to refer to the ease with which customers can access useful information through AI-enabled chatbots at any time and from any location. Efficient chatbot services are critical for enhancing user experiences. Fan et al. (2022) highlighted the importance of providing fast and flexible services through AI chatbots. Xu et al. (2018) found that users’ ability to navigate messaging platforms effectively improved their experiences of social interactions. When chatbots provide assistance at any time and from any location, customers are more likely to view interactions as interactive.

H3.

The efficiency of AI-enabled chatbot services exerts a positive influence on perceived interactivity.

Human-likeness refers to the extent to which a chatbot displays behaviors that resemble human interaction, thereby simulating the interpersonal experiences typically provided by frontline employees (Adiwardana et al., 2020). Perceived humanness has been shown to enhance a chatbot’s conversational effectiveness (Ma and Huo, 2023). When technologies exhibit human-like cues, users are more likely to interpret interactions with them as socially meaningful (Rhim et al., 2022). Anthropomorphic conversational agents have been found to improve users’ perceptions of interactivity (Toader et al., 2019). When AI-enabled chatbots communicate in a friendly, human-like manner, they create more engaging experiences that strengthen users’ sense of interactivity.

H4.

The human-likeness of AI-enabled chatbot services exerts a positive influence on perceived interactivity.

Responsiveness refers to a chatbot’s ability to deliver timely and appropriate services, including prompt replies and immediate support (Meyer-Waarden et al., 2020). It reflects how effectively a chatbot meets user needs during interactions. Avidar (2013) emphasized that responsiveness is a key factor in shaping users’ perceptions of interactivity. Similarly, Go et al. (2020) found that consumers evaluated interactivity based on responsiveness and sense of control. When a chatbot provides synchronous responses to customer inquiries, it enhances the perception of interactivity through a series of back-and-forth exchanges.

H5.

The responsiveness of AI-enabled chatbot services exerts a positive influence on perceived interactivity.

Reference to the service involves directing users to relevant information through digital channels, including chat interfaces, web forms and automated responses (Borsci et al., 2023). Such reference features are commonly used to support real-time interactions (Bhawiyuga et al., 2017). Virtual reference systems often follow standardized procedures to ensure accuracy and completeness, similar to face-to-face communication (Hamer, 2021). In online settings, perceived interactivity is strengthened by interactive elements, such as clickable links, multimedia content and navigational tools (Bucy, 2004; McMillan, 2005). When an AI-enabled chatbot effectively provides relevant guidance, users are more likely to perceive the service as interactive.

H6.

Reference to the service exerts a positive influence on perceived interactivity.

Brand experience refers to consumers’ internal perceptions and behavioral responses triggered by brand-related stimuli, including design, identity and communication (Bapat and Thanigan, 2016). In this study, brand experience is conceptualized as consumers’ internal perceptions and behavioral reactions elicited through interactions with AI-enabled chatbot services. These interactions help consumers evaluate product attributes based on their prior experiences with a brand (Saari et al., 2020). Brand experience may arise from direct interactions, such as product use, and indirect interactions, such as exposure to marketing messages (Moreau, 2020). In digital marketing contexts, brand experience is shaped by the reciprocal nature of brand–customer interactions (Wang et al., 2025).

Empirical studies have consistently shown a link between interactivity and brand experience. For instance, social media interactivity positively influences how consumers perceive and experience brands (Ye et al., 2019). Perceived interactivity engages users cognitively, emotionally and behaviorally, leading to deeper brand experiences (Yoon and Youn, 2016). Ischen et al. (2020) found that interacting with chatbots can generate enjoyable and memorable experiences. Therefore, when consumers perceive a high level of interactivity with AI-enabled chatbot services, they are more likely to develop positive brand experiences.

H7.

The perceived interactivity of AI-enabled chatbot services exerts a positive influence on brand experience.

Perceived interactivity has been shown to influence the development of parasocial relationships, particularly with social media influencers (Labrecque, 2014). As interactions accumulate over time, initial parasocial interactions may deepen to become more stable and enduring (Xu et al., 2024). Kim and Moon (2021) found that parasocial relationships are positively affected by perceived interactivity as well as by the endorsers’ credibility and the perceived quality of services and products offered online. When users experience interactive and dialogic communication with AI-enabled chatbots, they may begin to perceive the chatbot as a conversational partner. This sense of engagement can foster emotional closeness and strengthen the parasocial bond between the user and the brand.

H8.

The perceived interactivity of AI-enabled chatbot services exerts a positive influence on parasocial relationships.

Chatbots have transformed brand communication by enabling interactive engagement and immersive customer experiences (Li and Wang, 2023). Widely used across customer service functions, chatbots play an important role in shaping brand–consumer relationships (Wang et al., 2025). According to social exchange theory, the receipt of benefits creates a sense of reciprocity, which in turn strengthens relational ties (Blau, 1964). Interactions with a positive brand experience can foster an emotional connection and trust, leading to the development of a parasocial relationship. When consumers derive value from chatbot interactions, they are more likely to form parasocial bonds with brands.

H9.

The brand experience of AI-enabled chatbot services exerts a positive influence on parasocial relationships.

Uniqlo was selected as the research context for several reasons. First, the company provides AI-enabled chatbot services 24 h a day, along with LiveChat support from 9 a.m. to 9 p.m., to assist customers with product and service inquiries. Its AI-driven chatbot, Uniqlo IQ, serves as a digital concierge that offers personalized-style suggestions (Williams, 2018). Second, Uniqlo is recognized for its global presence and innovation in fabric technology. Founded in 1963, it has established a reputation for offering high-quality, functional apparel at affordable prices. As of the end of 2022, Uniqlo operated 809 stores in Japan and 1,585 internationally (Uniqlo, 2023), making it a prominent brand in the global retail market.

Consumers who had interacted with Uniqlo’s AI-enabled chatbot were targeted in this study. Data were collected using a self-administered online questionnaire, which was conducted between March 1 and April 5, 2022. Participants were recruited through convenience sampling and were required to have made at least one purchase at Uniqlo within the previous six months. To encourage participation, prizes were offered, with 20 participants who had provided valid responses randomly selected to receive convenience store vouchers.

All measurement items were adapted from established scales to fit the context of AI-enabled service chatbots. Perceived interactivity was assessed using a nine-point Likert scale ranging from 1 (strongly disagree) to 9 (strongly agree). The remaining constructs, including chatbot service attributes, brand experience and parasocial relationships, were measured using 7-point Likert scales ranging from 1 (strongly disagree) to 7 (strongly agree).

Functionality was measured using three items adapted from Bialkova (2024) and Meerschman and Verkeyn (2019). Trustworthiness was assessed using four items adapted from Agnihotri and Bhattacharya (2024), Liu et al. (2023) and Meerschman and Verkeyn (2019). Efficiency was measured using five items based on Chen et al. (2022) and Meerschman and Verkeyn (2019). Human-likeness was evaluated using three items from Cai et al. (2022) and Meerschman and Verkeyn (2019), while responsiveness was assessed using two items from Meyer-Waarden et al. (2020) and Meerschman and Verkeyn (2019). Reference to the service was measured using three items adapted from Borsci et al. (2022). Perceived interactivity was assessed using five items from Labrecque (2014) and Yuan et al. (2021). Brand experience was measured using six items from Yoon and Youn (2016), and parasocial relationships were evaluated using three items adapted from Labrecque (2014) and Yuan et al. (2021). A full list of questionnaire items is provided in Web Appendix 3.

A total of 302 questionnaires were collected. After data screening, 278 valid responses remained. Twelve cases were removed due to incomplete answers or completion times below the five-minute threshold established in the pilot test. Another 12 responses were excluded as multivariate outliers identified by Mahalanobis distance (Tabachnick and Fidell, 2007). Because of the small proportion of missing data, we followed Pepinsky’s (2018) recommendation for listwise deletion, which did not significantly affect the analysis results.

The sample characteristics revealed that 37.1% of the respondents were male. The majority of the participants (69.8%) were 20–29 years old, indicating that younger adults are the primary demographic purchasing clothing from Uniqlo. In terms of education, approximately 55.0% of the sample had a college or graduate degree. Most participants (71.2%) reported using Uniqlo’s customer service for 0–0.5 h, indicating that customers typically expect their inquiries to be resolved within 30 min. The highest purchase frequency was observed among those who made purchases every six months to one year (30.2%), followed by those who made purchases every 2–3 months (25.9%). As most participants were college students, 52.2% of participants typically spent an average of USD 35–70 per visit to Uniqlo. All participants resided in Taiwan.

Prior to estimating the structural model, multicollinearity was assessed. As shown in Table 1, the highest variance inflation factor was 2.84, which is well below the critical value of 5 (Hair et al., 2019). Internal consistency reliability was evaluated using Cronbach’s α and composite reliability (CR) of the latent variables. A Cronbach’s α greater than 0.7 is generally considered acceptable, with values of 0.5–0.7 deemed acceptable in some contexts, while CR should exceed 0.7. As shown in Table 1, the Cronbach’s α value for each construct ranged from 0.667 to 0.882, indicating acceptable reliability, and all CR values met the 0.7 threshold. Shrestha (2021) suggested that factor loadings should exceed 0.7, with an acceptable range of 0.6–0.7. Except for CH1 and PI5, the factor loadings for each dimension ranged from 0.691 to 0.915. However, due to the high correlation coefficients between responsiveness and efficiency and between responsiveness and reference to the service, responsiveness was removed to avoid construct overlap.

Table 1

Factor loadings and reliability

ConstructItemsFactor loadingVIFaCronbach’s αCRaAVEa
FunctionalityCF10.8862.4230.8740.9220.798
CF20.9152.672
CF30.8792.119
TrustworthinessCT10.8542.1940.8000.8700.627
CT20.8382.109
CT30.7651.531
CT40.7021.366
EfficiencyCE10.7291.5420.8340.8830.604
CE20.8292.166
CE30.8402.130
CE40.7861.823
CE50.6911.431
Human-likenessCH1DeletedDeleted0.6670.8570.750
CH20.8841.334
CH30.8481.334
ResponsivenessCR1Deletedb
CR2
Reference to the serviceRS10.8751.9240.8160.8910.731
RS20.8611.863
RS30.8281.685
Perceived interactivityPI10.8552.0930.8270.8860.660
PI20.8102.015
PI30.8251.883
PI40.7551.460
PI5deleteddeleted
Brand experienceBE10.6921.5760.8820.9110.631
BE20.7922.158
BE30.8272.307
BE40.8022.250
BE50.8142.161
BE60.8292.211
Parasocial relationshipsPR10.8651.9510.8300.8980.746
PR20.9022.259
PR30.8231.728

Note(s): CH1 and PI5 were deleted because their factor loadings were −0.659 and 0.494, respectively, which were below the threshold of 0.5. Before removing these items, the Cronbach’s α value for human likeness was 0.038 while that for interactivity was 0.799

a

VIF, Variance inflation factor; CR, composite reliability; AVE, average variance extracted

b

Responsiveness was removed to avoid construct overlap due to its high correlations with efficiency and service reference

Table 1 shows that the average variance extracted (AVE) values for each construct ranged from 0.604 to 0.798. All constructs exceeded the recommended threshold of 0.5 (Shrestha, 2021), thus demonstrating good convergent validity. Discriminant validity was assessed using the Fornell–Larcker criterion and the heterotrait-monotrait (HTMT) ratio of correlations. As shown in Table 2, the square root of AVE for each construct exceeded the correlations between the latent variables based on the Fornell–Larcker criterion, thus indicating good discriminant validity. The HTMT values in this study were all below the recommended threshold of 0.9, further confirming good discriminant validity.

Table 2

Correlations among major constructs

VariableCFCTCECHRSPIBEPR
Functionality (CF)0.8940.8500.6170.5620.5990.5910.5220.719
Trustworthiness (CT)0.7020.7920.7760.5670.7820.7760.6040.775
Efficiency (CE)0.5250.6370.7770.4640.8520.6050.2960.436
Human-likeness (CH)0.7310.7870.6110.8660.6130.7630.7090.841
Reference to the service (RS)0.5080.6350.7020.4560.8550.5970.3170.490
Perceive interactivity (PI)0.5060.6330.7240.5730.7270.8120.4700.643
Brand experience (BE)0.4630.5030.3480.5390.3790.5420.7940.804
Parasocial relationships (PR)0.6140.6250.5250.6230.5970.5370.6940.864
Mean2.662.482.292.852.293.473.112.79
Standard deviation1.030.870.931.070.840.871.181.00

Note(s): Diagonal elements are the square root of average variance extracted (AVE) of the reflective scales. Off-diagonal elements are correlations between construct. Above the diagonal element is the HTMT value

The results of the analysis are presented in Figure 2. Education, gender, age and AI chatbot use duration were included as control variables. The results showed that the trustworthiness (ß = 0.250, t = 3.924, p < 0.001), efficiency (ß = 0.204, t = 3.177, p < 0.01), human-likeness (ß = 0.262, t = 4.364, p < 0.001) and reference to the service (ß = 0.184, t = 2.245, p < 0.05) of the AI-enabled chatbot service influenced the perceived interactivity. Therefore, H2, H3, H4 and H6 are supported. However, the functionality of the AI-enabled chatbot service did not significantly affect the perceived interactivity (ß = −0.017, t = 0.278, p > 0.05). Thus, H1 is not supported.

Figure 2
A research framework diagram with path coefficients and R squared values.The research framework diagram shows a dashed, rounded rectangle on the left labeled “Service chatbot’s attributes”. Inside it, five vertically arranged ovals read “Functionality”, “Trustworthiness”, “Efficiency”, “Human-likeness”, and “Reference to the service”, from top to bottom. Arrows from these ovals point to a central oval labeled “Perceived interactivity”. The path coefficients shown are negative 0.017 from “Functionality”, 0.250 triple asterisk from “Trustworthiness”, 0.204 double asterisk from “Efficiency”, 0.262 triple asterisk from “Human-likeness”, and 0.184 single asterisk from “Reference to the service”. Above, “Perceived interactivity” appears “R-squared equals 0.533”. An arrow labeled 0.470 triple asterisk extends from “Perceived interactivity” to an oval labeled “Brand experience”, with “R-squared equals 0.221” shown near it. Another arrow labeled 0.267 triple asterisk extends from “Perceived interactivity” to an oval labeled “Parasocial relationships”. A vertical arrow labeled 0.567 triple asterisk points downwards from “Brand experience” to “Parasocial relationships”, with “R-squared equals 0.538” shown near “Parasocial relationships”. At the bottom right, a dashed, rounded rectangle labeled “Control variables” contains four ovals labeled “Education”, “Gender”, “Age”, and “Chatbot’s use duration”, from left to right. Dashed arrows from these ovals point to “Parasocial relationships”, labeled 0.015 from “Education”, 0.001 from “Gender”, negative 0.041 from “Age”, and 0.037 from “Chatbot’s use duration”.

PLS results for the proposed model. Note: *p < 0.05, **p < 0.01, ***p < 0.001. Responsiveness was removed to avoid construct overlap due to its high correlations with efficiency and service reference

Figure 2
A research framework diagram with path coefficients and R squared values.The research framework diagram shows a dashed, rounded rectangle on the left labeled “Service chatbot’s attributes”. Inside it, five vertically arranged ovals read “Functionality”, “Trustworthiness”, “Efficiency”, “Human-likeness”, and “Reference to the service”, from top to bottom. Arrows from these ovals point to a central oval labeled “Perceived interactivity”. The path coefficients shown are negative 0.017 from “Functionality”, 0.250 triple asterisk from “Trustworthiness”, 0.204 double asterisk from “Efficiency”, 0.262 triple asterisk from “Human-likeness”, and 0.184 single asterisk from “Reference to the service”. Above, “Perceived interactivity” appears “R-squared equals 0.533”. An arrow labeled 0.470 triple asterisk extends from “Perceived interactivity” to an oval labeled “Brand experience”, with “R-squared equals 0.221” shown near it. Another arrow labeled 0.267 triple asterisk extends from “Perceived interactivity” to an oval labeled “Parasocial relationships”. A vertical arrow labeled 0.567 triple asterisk points downwards from “Brand experience” to “Parasocial relationships”, with “R-squared equals 0.538” shown near “Parasocial relationships”. At the bottom right, a dashed, rounded rectangle labeled “Control variables” contains four ovals labeled “Education”, “Gender”, “Age”, and “Chatbot’s use duration”, from left to right. Dashed arrows from these ovals point to “Parasocial relationships”, labeled 0.015 from “Education”, 0.001 from “Gender”, negative 0.041 from “Age”, and 0.037 from “Chatbot’s use duration”.

PLS results for the proposed model. Note: *p < 0.05, **p < 0.01, ***p < 0.001. Responsiveness was removed to avoid construct overlap due to its high correlations with efficiency and service reference

Close modal

Perceived interactivity had a significant effect on brand experience (ß = 0.470, t = 8.729, p < 0.001) and parasocial relationships (ß = 0.267, t = 5.430, p < 0.001). Brand experience had a significant effect on parasocial relationships (ß = 0.567, t = 13.478, p < 0.001). Thus, H7, H8 and H9 are supported. As shown in Figure 2, the R2 was 0.533 for perceived interaction, 0.221 for brand experience and 0.538 for parasocial relationships.

To examine the mediating role of brand experience, a mediation analysis was conducted following the procedures outlined by Baron and Kenny (1986) and Hair et al. (2019). Indirect effects were tested using a bootstrapping approach with 5,000 resamples, and the percentile and bias-corrected confidence intervals (CIs) were set at 95%. Statistical significance was determined by examining whether the 95% CI for the indirect effect excluded zero. Table 3 reports the direct, indirect and total effects. The total effect was calculated by summing the direct and indirect effects. The indirect effect of perceived interactivity on parasocial relationships via brand experience was positive and statistically significant (β = 0.264, p < 0.001). The bootstrapped standard error was 0.034 and the 95% CI was (0.198, 0.336), which did not include zero, thus confirming the presence of a significant indirect effect (Wang et al., 2019). In addition, perceived interactivity exhibited a significant direct effect on brand experience (β = 0.470, p < 0.001), suggesting that brand experience partially mediates the relationship between perceived interactivity and parasocial relationships. A summary of the hypothesis-testing results is presented in Table 4.

Table 3

Mediation effect of brand experience

PathDirect effectIndirect effectTotal effect
Perceived interactivity → brand experience0.470*** (0.365, 0.571) 0.470*** (0.365, 0.571)
Brand experience → parasocial relationships0.562*** (0.468, 0.646) 0.562*** (0.468, 0.646)
Perceived interactivity → brand experience → parasocial relationships0.267*** (0.148, 0.377)0.264*** (0.198, 0.336)0.531*** (0.402, 0.645)

Note(s): ***p < 0.001. Values in parentheses indicate 95% bias-corrected confidence intervals (CIs) based on bootstrapping

Table 4

Results of hypotheses tests

HypothesisSupported?
H1 The functionality of AI-enabled chatbot services exerts a positive influence on perceived interactivityNot supported
H2 The trustworthiness of AI-enabled chatbot services exerts a positive influence on perceived interactivityYes, ß = 0.250***
H3 The efficiency of AI-enabled chatbot services exerts a positive influence on perceived interactivityYes, ß = 0.204**
H4 The human-likeness of AI-enabled chatbot services exerts a positive influence on perceived interactivityYes, ß = 0.262***
H5 The responsiveness of AI-enabled chatbot services exerts a positive influence on perceived interactivityRemoved
H6 Reference to the service exerts a positive influence on perceived interactivityYes, ß = 0.184*
H7 The perceived interactivity of AI-enabled chatbot services exerts a positive influence on brand experienceYes, ß = 0.470***
H8 The perceived interactivity of AI-enabled chatbot services exerts a positive influence on parasocial relationshipsYes, ß = 0.267***
H9 The brand experience of AI-enabled chatbot services exerts a positive influence on parasocial relationshipsYes, ß = 0.567***

Note(s): *p < 0.05, **p < 0.01, ***p < 0.001; H5 was removed to avoid construct overlap, because responsiveness had high correlations with efficiency and service reference

Several key findings emerged from this study. First, the attributes of AI-enabled chatbot services significantly influence consumer responses (Cheng et al., 2022b). In interactions with Uniqlo’s chatbot, consumers who perceived the information as trustworthy reported higher levels of perceived interactivity. This aligns with Sonntag et al. (2023), who found that trust-enhancing design elements improve users’ perceptions of interactivity. In addition, efficiency positively affected perceived interactivity, supporting the findings of Xu et al. (2018), who demonstrated a link between service efficiency and interactive engagement. When consumers experience prompt service, they are more likely to perceive the chatbot as interactive.

Second, human-like features enhance interaction quality. When Uniqlo’s chatbot mimicked human conversational behavior, it increased consumers’ perceptions of interactivity. This finding supports research by Chandra et al. (2022), who observed that human-like traits in conversational AI foster user engagement. In addition, reference to the service contributed to higher perceived interactivity, echoing McMillan’s (2005) view that embedded reference features improve user interaction with digital systems.

Third, contrary to expectations, functionality does not significantly influence perceived interactivity. To explore this result, an experiment was conducted using two chatbots with different levels of functionality. Thirty college students interacted with both chatbots for at least 10 min before completing a questionnaire. The results showed that high functionality significantly influenced perceived interactivity (b = 0.284, p < 0.05) but that low functionality did not (b = 0.209, p > 0.05). This suggests that users may not view limited-function chatbots as interactive. Web Appendix 4 shows a screenshot of Uniqlo’s chatbot services. Uniqlo’s chatbot, which cannot manage personal order statuses, identify personal accounts or process images, offers only a limited service to customers. It cannot assist a customer with canceling orders but instead directs them to log into their online account and follow the suggested steps. As all the participants in this study interacted with the same chatbot features, the variability in functionality perception was likely constrained. Therefore, the likelihood that users’ perceptions of interactivity were influenced by the limited functionality is low. These findings are consistent with Grimes et al.’s (2021) argument that the functionality of AI systems is not always apparent to users, as AI capabilities can vary widely.

Fourth, perceived interactivity enhances brand experience. This is consistent with Nguyen et al. (2023), who found that interactive service encounters improve brand-related outcomes. In the current study, the consumers’ interactions with the chatbot enriched their overall experiences with the Uniqlo brand. Moreover, perceived interactivity also positively influenced parasocial relationships, reinforcing Kim and Moon’s (2021) argument that interactive communication fosters emotional connections. Engaging with the chatbot contributed to the consumers’ feelings of relational closeness to the brand.

Finally, PSI theory offers a relevant lens for interpreting these findings. Traditionally applied to real individuals, such as celebrities and media hosts, PSI theory has been extended to anthropomorphized digital agents, including AI assistants (Lee and Lee, 2023). The results showed that brand experience enhanced parasocial relationships, supporting Tsai et al. (2021), who emphasized the role of brand experience in forming relational bonds. As the consumers engaged with Uniqlo’s chatbot, they may have begun to perceive the brand in interpersonal terms, imagining a social connection. These findings highlight the strategic potential of AI-enabled chatbots in building consumer engagement and strengthening brand relationships.

This study offers several novel contributions. First, it extends the application of PSI theory to the context of AI-enabled chatbot services. Although PSI theory has been applied to studies of customer support involving chatbots across various domains, anthropomorphism and social presence have been emphasized as key antecedents to parasocial interactions or relationships, with satisfaction and behavioral intention commonly examined as outcomes. By adopting the PSI perspective, this study offers new insights into how retailers can cultivate such relationships through chatbot-mediated interactions. This approach also addresses the concerns raised by Shi et al. (2025), who questioned whether theories grounded in human communication are applicable to human–computer interaction or if entirely new theoretical frameworks are necessary.

Second, perceived interactivity was identified as a core characteristic of AI-enabled chatbot services. Rather than being an inherent feature of the technology, perceived interactivity was conceptualized as a consumer-centered perception, similar to constructs such as perceived value or perceived risk. In this study, perceived interactivity was treated as a unidimensional construct that reflects consumers’ evaluative responses following interactions with AI-enabled chatbots. Specifically, this study investigated how key service attributes, in terms of trustworthiness, efficiency, human-likeness and reference to the service, shape consumers’ perceptions of interactivity. These perceptions, in turn, enhance brand experiences and foster parasocial relationships between consumers and brands. The findings reinforce the perspective proposed by Yang and Shen (2018), which views interactivity as a subjective evaluation shaped by users’ interpretations of service attributes.

Finally, while previous studies have often treated brand experience as an outcome of consumer interactions (Zhang, 2021), few have explored its role in building parasocial relationships. This study posits that brand experience can result from parasocial, rather than face-to-face, interactions. When consumers perceive a high degree of interactivity with an AI-enabled chatbot, they form a favorable impression of the brand, thereby strengthening the parasocial bond between the brand and its customers. These findings provide implications for both researchers and practitioners by offering guidance on how to design AI-enabled chatbot services that deepen brand–customer relationships.

Grounded in PSI theory, this study confirms that the trustworthiness, efficiency, human-likeness and reference to the service of AI-enabled chatbots shape perceived interactivity and brand experience, which in turn strengthen customer–brand parasocial bonds. The findings yield several actionable recommendations for retailers. First, drawing on social exchange theory (Blau, 1964), which posits that people engage in reciprocal relationships with the expectation of information and rewards such as emotional support, this study offers several practical recommendations. To enhance trust in AI-powered chatbot services, retailers should ensure that brand information is consistently up to date and provide incentives for customers to use AI-driven customer service, such as free shipping coupons or shopping vouchers. The use of AI-driven services enables retailers to gather valuable insights from customer interactions, leading to more accurate responses and, ultimately, stronger trust. A key challenge for brands is motivating customers to disclose personal information or provide feedback for service improvement, as perceived interactivity is heightened through two-way communication. In addition, implementing robust identity verification processes assures customers that their data will remain secure and will not be shared with third parties, further enhancing security and trustworthiness.

Second, many brands have established official accounts on social media to announce their latest news and events. By linking their AI-enabled chatbot services to these official accounts, retailers can allow consumers to access customer service through their official websites, apps or commonly used communication platforms. Making AI-enabled chatbots easily accessible significantly enhances customer service efficiency. Moreover, retailers should focus on improving the reference to the service. Although AI-enabled chatbots guide consumers to relevant services or information, issues with hyperlink accuracy can sometimes direct users to incorrect URLs, causing frustration. To address this issue, companies can include keywords or brief descriptions alongside hyperlinks to improve clarity and readability.

Third, AI-enabled chatbot services should closely mimic human personalities. Such human-likeness can be achieved through anthropomorphism and by incorporating an identity, small talk or expressions of sympathy (Adam et al., 2021). AI-enabled chatbots with a high degree of human-likeness make it difficult for consumers to distinguish whether they are communicating with chatbots or human representatives. For example, when addressing an inquiry related to a sale return, an AI-enabled chatbot can use a soothing tone to alleviate customer dissatisfaction. Retailers can also enhance the internal and external human-like features of chatbots by creating distinct personalities and names for their AI-enabled services.

Finally, retailers should consider adding hedonic elements to AI-enabled chatbot services. For example, a firm could integrate augmented reality technology with an AI-enabled chatbot to create a customer service experience that aligns with its brand image and personality. Consumers can scan product labels to activate virtual customer service through their mobile devices, thereby seamlessly merging digital services with the real world. This integration of visual and auditory elements that resonate with the brand extends AI-enabled chatbot services beyond traditional web pages, making interactions more engaging and enriching consumers’ brand experiences.

This study has several limitations. First, the data were collected from an online survey distributed via social media platforms in Taiwan. Consequently, the majority of respondents were between 20 and 29 years old, which may constrain the generalizability of the findings to broader consumer groups or international markets. Given that Uniqlo operates globally, future research should incorporate cross-cultural comparisons to examine potential variations across countries. Second, this study focused exclusively on Uniqlo and examined six chatbot attributes: functionality, trustworthiness, efficiency, human-likeness, responsiveness and reference to the service. As AI-enabled chatbots are still evolving, the insignificant relationship between functionality and perceived interactivity may reflect limitations specific to early-stage chatbot capabilities. Additionally, other relevant attributes, such as personalization, were not considered. Future studies should explore a broader range of service features and test these constructs in diverse brand contexts.

Third, this study used a single screening item to confirm the participants’ prior use of Uniqlo’s AI-enabled chatbot. It did not verify whether their evaluations were based solely on chatbot interactions or whether they had also interacted with human service representatives. Such a crossover may have confounded the relationship between perceived interactivity and brand experience. Future research should implement stricter screening protocols to exclude respondents who have engaged with human agents during service encounters. Finally, while this study confirms that perceived interactivity and brand experience foster parasocial relationships, the downstream effects of these relationships remain unexplored. Future research could examine the behavioral and commercial outcomes of parasocial bonds, such as their influence on purchase intentions, loyalty and word-of-mouth advocacy.

An earlier version of this paper was presented at the 2023 International Conference on Innovation and Management in Osaka, Japan.

The supplementary material for this article can be found online.

Adam
,
M.
,
Wessel
,
M.
and
Benlian
,
A.
(
2021
), “
AI-based chatbots in customer service and their effects on user compliance
”,
Electronic Markets
, Vol. 
31
No. 
2
, pp. 
427
-
445
, doi: .
Adiwardana
,
D.
,
Luong
,
M.T.
,
So
,
D.R.
,
Hall
,
J.
,
Fiedel
,
N.
,
Thoppilan
,
R.
,
Yang
,
Z.
,
Kulshreshtha
,
A.
,
Nemade
,
G.
,
Lu
,
Y.
and
Le
,
Q.V.
(
2020
), “
Towards a human-like open-domain chatbot
”,
available at:
 https://arxiv.org/abs/2001.09977/ (
accessed
 24 September 2025).
Agnihotri
,
A.
and
Bhattacharya
,
S.
(
2024
), “
Chatbots' effectiveness in service recovery
”,
International Journal of Information Management
, Vol. 
76
, 102679, doi: .
Avidar
,
R.
(
2013
), “
The responsiveness pyramid: embedding responsiveness and interactivity into public relations theory
”,
Public Relations Review
, Vol. 
39
No. 
5
, pp. 
440
-
450
, doi: .
Bapat
,
D.
and
Thanigan
,
J.
(
2016
), “
Exploring relationship among brand experience dimensions, brand evaluation and brand loyalty
”,
Global Business Review
, Vol. 
17
No. 
6
, pp. 
1357
-
1372
, doi: .
Baron
,
R.M.
and
Kenny
,
D.A.
(
1986
), “
The moderator–mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations
”,
Journal of Personality and Social Psychology
, Vol. 
51
No. 
6
, pp.
1173
-
1182
, doi: .
Bhawiyuga
,
A.
,
Fauzi
,
M.A.
,
Pramukantoro
,
E.S.
and
Yahya
,
W.
(
2017
), “
Design of e-commerce chat robot for automatically answering customer questions
”,
Proceedings of the 2017 International Conference on Sustainable Information Engineering and Technology
,
IEEE
,
Malang, Indonesia
, pp. 
159
-
162
.
Bialkova
,
S.
(
2024
), “
How to optimise interaction with chatbots? Key parameters emerging from actual application
”,
International Journal of Human-Computer Interaction
, Vol. 
40
No. 
17
, pp. 
4688
-
4697
, doi: .
Blau
,
P.M.
(
1964
),
Exchange and Power in Social Life
,
Wiley
,
New York
.
Borsci
,
S.
,
Malizia
,
A.
,
Schmettow
,
M.
,
Van Der Velde
,
F.
,
Tariverdiyeva
,
G.
,
Balaji
,
D.
and
Chamberlain
,
A.
(
2022
), “
The chatbot usability scale: the design and pilot of a usability scale for interaction with AI-based conversational agents
”,
Personal and Ubiquitous Computing
, Vol. 
26
No. 
1
, pp. 
95
-
119
, doi: .
Borsci
,
S.
,
Schmettow
,
M.
,
Malizia
,
A.
,
Chamberlain
,
A.
and
Van Der Velde
,
F.
(
2023
), “
A confirmatory factorial analysis of the chatbot usability scale: a multilanguage validation
”,
Personal and Ubiquitous Computing
, Vol. 
27
No. 
2
, pp. 
317
-
330
, doi: .
Bucy
,
E.P.
(
2004
), “
Interactivity in society: locating an elusive concept
”,
The Information Society
, Vol. 
20
No. 
5
, pp. 
373
-
383
, doi: .
Cai
,
D.
,
Li
,
H.
and
Law
,
R.
(
2022
), “
Anthropomorphism and OTA chatbot adoption: a mixed-methods study
”,
Journal of Travel and Tourism Marketing
, Vol. 
39
No. 
2
, pp. 
228
-
255
, doi: .
Chakraborty
,
D.
,
Kar
,
A.K.
,
Patre
,
S.
and
Gupta
,
S.
(
2024
), “
Enhancing trust in online grocery shopping through generative AI chatbots
”,
Journal of Business Research
, Vol. 
180
, 114737, doi: .
Chandra
,
S.
,
Shirish
,
A.
and
Srivastava
,
S.C.
(
2022
), “
To be or not to be… human? Theorizing the role of human-like competencies in conversational artificial intelligence agents
”,
Journal of Management Information Systems
, Vol. 
39
No. 
4
, pp. 
969
-
1005
, doi: .
Chen
,
Q.
,
Gong
,
Y.
,
Lu
,
Y.
and
Tang
,
J.
(
2022
), “
Classifying and measuring the service quality of AI chatbot in frontline service
”,
Journal of Business Research
, Vol. 
145
, pp. 
552
-
568
, doi: .
Chen
,
Q.
,
Lu
,
Y.
,
Gong
,
Y.
and
Xiong
,
J.
(
2023
), “
Can AI chatbots help retain customers? Impact of AI service quality on customer loyalty
”,
Internet Research
, Vol. 
33
No. 
6
, pp. 
2205
-
2243
, doi: .
Chen
,
Q.
,
Yin
,
C.
and
Gong
,
Y.
(
2025
), “
Would an AI chatbot persuade you: an empirical answer from the elaboration likelihood model
”,
Information Technology and People
, Vol. 
38
No. 
2
, pp. 
937
-
962
, doi: .
Cheng
,
X.
,
Bao
,
Y.
,
Zarifis
,
A.
,
Gong
,
W.
and
Mou
,
J.
(
2022a
), “
Exploring consumers' response to text-based chatbots in e-commerce: the moderating role of task complexity and chatbot disclosure
”,
Internet Research
, Vol. 
32
No. 
2
, pp. 
496
-
517
, doi: .
Cheng
,
X.
,
Zhang
,
X.
,
Cohen
,
J.
and
Mou
,
J.
(
2022b
), “
Human vs. AI: understanding the impact of anthropomorphism on consumer response to chatbots from the perspective of trust and relationship norms
”,
Information Processing and Management
, Vol. 
59
No. 
3
, 102940, doi: .
Cheng
,
X.
,
Qiao
,
L.
,
Yang
,
B.
and
Li
,
Z.
(
2024
), “
An investigation on the influencing factors of elderly people's intention to use financial AI customer service
”,
Internet Research
, Vol. 
34
No. 
3
, pp. 
690
-
717
, doi: .
Daisy
,
A.
(
2020
), “
Stimulus of chatbot with customer relationship pipeline
”,
Journal of Advanced Research in Dynamical and Control Systems
, Vol. 
12
No. 
5
, pp. 
885
-
889
, doi: .
Duong
,
C.D.
,
Nguyen
,
T.H.
,
Ngo
,
T.V.N.
,
Pham
,
T.T.P.
,
Vu
,
A.T.
and
Dang
,
N.S.
(
2025
), “
Using generative artificial intelligence (ChatGPT) for travel purposes: parasocial interaction and tourists' continuance intention
”,
Tourism Review
, Vol. 
80
No. 
4
, pp. 
813
-
827
, doi: .
Dwivedi
,
Y.K.
,
Balakrishnan
,
J.
,
Baabdullah
,
A.M.
and
Das
,
R.
(
2023
), “
Do chatbots establish ‘humanness’ in the customer purchase journey? An investigation through explanatory sequential design
”,
Psychology and Marketing
, Vol. 
40
No. 
11
, pp. 
2244
-
2271
, doi: .
Fan
,
H.
,
Han
,
B.
and
Gao
,
W.
(
2022
), “
(Im) balanced customer-oriented behaviors and AI chatbots' efficiency-flexibility performance: the moderating role of customers' rational choices
”,
Journal of Retailing and Consumer Services
, Vol. 
66
, 102937, doi: .
Farivar
,
S.
,
Wang
,
F.
and
Yuan
,
Y.
(
2021
), “
Opinion leadership vs. parasocial relationship: key factors in influencer marketing
”,
Journal of Retailing and Consumer Services
, Vol. 
59
, 102371, doi: .
Feng
,
Y.
,
Meng
,
J.
and
Cheah
,
J.H.
(
2025
), “
From virtual trainers to companions? Examining how digital agency types, anthropomorphism, and support shape para‐social relationships in online fitness
”,
Psychology and Marketing
, Vol. 
42
No. 
3
, pp. 
842
-
865
, doi: .
Gkinko
,
L.
and
Elbanna
,
A.
(
2022
), “
Hope, tolerance and empathy: employees' emotions when using an AI-enabled chatbot in a digitalised workplace
”,
Information Technology and People
, Vol. 
35
No. 
6
, pp. 
1714
-
1743
, doi: .
Go
,
H.
,
Kang
,
M.
and
Suh
,
S.C.
(
2020
), “
Machine learning of robots in tourism and hospitality: interactive technology acceptance model (iTAM)–cutting edge
”,
Tourism Review
, Vol. 
75
No. 
4
, pp. 
625
-
636
, doi: .
Grimes
,
G.M.
,
Schuetzler
,
R.M.
and
Giboney
,
J.S.
(
2021
), “
Mental models and expectation violations in conversational AI interactions
”,
Decision Support Systems
, Vol. 
144
, 113515, doi: .
Hair
,
J.F.
,
Black
,
W.C.
,
Babin
,
B.J.
and
Anderson
,
R.E.
(
2019
),
Multivariate Data Analysis
,
Cengage Learning EMEA
,
Hampshire, United Kingdom
.
Hamer
,
S.
(
2021
), “
Colour blind: investigating the racial bias of virtual reference services in English academic libraries
”,
The Journal of Academic Librarianship
, Vol. 
47
No. 
5
, 102416, doi: .
Horton
,
D.
and
Wohl
,
R.R.
(
1956
), “
Mass communication and parasocial interaction: observations on intimacy at a distance
”,
Psychiatry
, Vol. 
19
No. 
3
, pp. 
215
-
229
, doi: .
Ischen
,
C.
,
Araujo
,
T.
,
van Noort
,
G.
,
Voorveld
,
H.
and
Smit
,
E.
(
2020
), “
I am here to assist you today’: the role of entity, interactivity and experiential perceptions in chatbot persuasion
”,
Journal of Broadcasting and Electronic Media
, Vol. 
64
No. 
4
, pp. 
615
-
639
, doi: .
Jeon
,
E.
and
Kim
,
J.
(
2025
), “
When time is crucial: effects of compensation type and anthropomorphic chatbots on chatbot service recovery satisfaction
”,
Journal of Research in Interactive Marketing
,
EarlyCite
, doi: .
Johari
,
N.M.
and
Nohuddin
,
P.N.
(
2021
), “
Quality attributes for a good chatbot: a literature review
”,
International Journal of Electrical Engineering and Technology
, Vol. 
12
No. 
7
, pp. 
109
-
119
.
Kharub
,
I.
,
Lwin
,
M.
,
Khan
,
A.
and
Mubin
,
O.
(
2021
), “
Perceived service quality in HRI: applying the SERVBOT framework
”,
Frontiers in Robotics and AI
, Vol. 
8
, 746674, doi: .
Kim
,
W.B.
and
Hur
,
H.J.
(
2024
), “
What makes people feel empathy for AI chatbots? Assessing the role of competence and warmth
”,
International Journal of Human-Computer Interaction
, Vol. 
40
No. 
17
, pp. 
4674
-
4687
, doi: .
Kim
,
J.
and
Im
,
I.
(
2023
), “
Anthropomorphic response: understanding interactions between humans and artificial intelligence agents
”,
Computers in Human Behavior
, Vol. 
139
, 107512, doi: .
Kim
,
K.H.
and
Moon
,
H.
(
2021
), “
Innovative digital marketing management in B2B markets
”,
Industrial Marketing Management
, Vol. 
95
, pp. 
1
-
4
, doi: .
Kim
,
M.
,
Oh
,
J.
,
Kim
,
D.
,
Shin
,
J.
and
Lee
,
D.
(
2025
), “
Understanding user preferences in developing a mental healthcare AI chatbot: a conjoint analysis approach
”,
International Journal of Human-Computer Interaction
, Vol. 
41
No. 
8
, pp. 
4813
-
4821
, doi: .
Labrecque
,
L.I.
(
2014
), “
Fostering consumer-brand relationships in social media environments: the role of parasocial interaction
”,
Journal of Interactive Marketing
, Vol. 
28
No. 
2
, pp. 
134
-
148
, doi: .
Lee
,
J.
and
Lee
,
D.
(
2023
), “
User perception and self-disclosure towards an AI psychotherapy chatbot according to the anthropomorphism of its profile picture
”,
Telematics and Informatics
, Vol. 
85
, 102052, doi: .
Li
,
M.
and
Wang
,
R.
(
2023
), “
Chatbots in e-commerce: the effect of chatbot language style on customers' continuance usage intention and attitude toward brand
”,
Journal of Retailing and Consumer Services
, Vol. 
71
, 103209, doi: .
Li
,
L.
,
Lee
,
K.Y.
,
Emokpae
,
E.
and
Yang
,
S.B.
(
2021
), “
What makes you continuously use chatbot services? Evidence from Chinese online travel agencies
”,
Electronic Markets
, Vol. 
31
No. 
3
, pp. 
575
-
599
, doi: .
Liu
,
D.
(
2025
), “
Is interaction between human and artificial intelligence–driven agents (para) social? A scoping review
”,
Cyberpsychology, Behavior, and Social Networking
, Vol. 
28
No. 
8
, pp. 
551
-
558
, doi: .
Liu
,
J.
,
Marriott
,
K.
,
Dwyer
,
T.
and
Tack
,
G.
(
2023
), “
Increasing user trust in optimisation through feedback and interaction
”,
ACM Transactions on Computer-Human Interaction
, Vol. 
29
No. 
5
, pp. 
1
-
34
, doi: .
Liu
,
W.
,
Jiang
,
M.
,
Li
,
W.
and
Mou
,
J.
(
2024
), “
How does the anthropomorphism of AI chatbots facilitate users' reuse intention in online health consultation services? The moderating role of disease severity
”,
Technological Forecasting and Social Change
, Vol. 
203
, 123407, doi: .
Ma
,
X.
and
Huo
,
Y.
(
2023
), “
Are users willing to embrace ChatGPT? Exploring the factors on the acceptance of chatbots from the perspective of AIDUA framework
”,
Technology in Society
, Vol. 
75
, 102362, doi: .
Mayer
,
R.C.
,
Davis
,
J.H.
and
Schoorman
,
F.D.
(
1995
), “
An integrative model of organizational trust
”,
Academy of Management Review
, Vol. 
20
No. 
3
, pp. 
709
-
734
, doi: .
McMillan
,
S.J.
(
2005
), “
The researchers and the concept: moving beyond a blind examination of interactivity
”,
Journal of Interactive Advertising
, Vol. 
5
No. 
2
, pp. 
1
-
4
, doi: .
Meerschman
,
H.
and
Verkeyn
,
J.
(
2019
), “
Towards a better understanding of service quality attributes of a chatbot
”,
Master’s thesis, Ghent University
.
Meyer-Waarden
,
L.
,
Pavone
,
G.
,
Poocharoentou
,
T.
,
Prayatsup
,
P.
,
Ratinaud
,
M.
,
Tison
,
A.
and
Torné
,
S.
(
2020
), “
How service quality influences customer acceptance and usage of chatbots
”,
Journal of Service Management Research
, Vol. 
4
No. 
1
, pp. 
35
-
51
, doi: .
McNeal
,
M.L.
and
Newyear
,
D.
(
2013
), “
Introducing chatbots in libraries
”,
Library Technology Reports
, Vol. 
49
No. 
8
, pp. 
5
-
10
.
Miller
,
K.
and
Larson
,
D.
(
2005
), “
Angels and artifacts: moral agents in the age of computers and networks
”,
Journal of Information, Communication and Ethics in Society
, Vol. 
3
No. 
3
, pp. 
151
-
157
, doi: .
Moreau
,
C.P.
(
2020
), “
Brand building on the doorstep: the importance of the first (physical) impression
”,
Journal of Retailing
, Vol. 
96
No. 
1
, pp. 
155
-
167
, doi: .
Nadroo
,
Z.M.
,
Islam
,
J.U.
and
Naqshbandi
,
M.A.
(
2025
), “
Parasocial Interaction in marketing domain: offering insights through a systematic literature review
”,
International Journal of Consumer Studies
, Vol. 
49
No. 
4
, e70079, doi: .
Nazir
,
S.
,
Khadim
,
S.
,
Asadullah
,
M.A.
and
Syed
,
N.
(
2023
), “
Exploring the influence of artificial intelligence technology on consumer repurchase intention: the mediation and moderation approach
”,
Technology in Society
, Vol. 
72
, 102190, doi: .
Nguyen
,
T.
,
Sun
,
Q.
and
Ganesh
,
G.
(
2023
), “
Customer service brand engagement among international students: the roles of preferred university brand attainment
”,
Journal of Applied Marketing Theory
, Vol. 
10
No. 
2
, pp. 
44
-
62
, doi: .
Noor
,
N.
,
Rao Hill
,
S.
and
Troshani
,
I.
(
2022
), “
Artificial intelligence service agents: role of parasocial relationship
”,
Journal of Computer Information Systems
, Vol. 
62
No. 
5
, pp. 
1009
-
1023
, doi: .
Pepinsky
,
T.B.
(
2018
), “
A note on listwise deletion versus multiple imputation
”,
Political Analysis
, Vol. 
26
No. 
4
, pp. 
480
-
488
, doi: .
Ramya
,
J.B.
and
Alur
,
S.
(
2025
), “
The mediating role of parasocial relationship in customer services chatbots among millennials and gen Z population
”,
International Journal of Human-Computer Interaction
, Vol. 
41
No. 
2
, pp. 
919
-
931
, doi: .
Reeves
,
B.
and
Nass
,
C.
(
1996
),
The Media Equation: How People Treat Computers, Television, and New Media like Real People
,
Cambridge University Press
,
New York
.
Rhim
,
J.
,
Kwak
,
M.
,
Gong
,
Y.
and
Gweon
,
G.
(
2022
), “
Application of humanization to survey chatbots: change in chatbot perception, interaction experience, and survey data quality
”,
Computers in Human Behavior
, Vol. 
126
, 107034, doi: .
Saari
,
U.A.
,
Mäkinen
,
S.J.
,
Baumgartner
,
R.J.
,
Hillebrand
,
B.
and
Driessen
,
P.H.
(
2020
), “
How consumers' respect for nature and environmental self-assets influence their car brand experiences
”,
Journal of Cleaner Production
, Vol. 
261
, 121023, doi: .
Sabanoglu
,
T.
(
2023
), “
Retail sales by channel worldwide 2019
”,
available at:
 https://www.statista.com/statistics/1095969/retail-sales-by-channel-worldwide/ (
accessed
 24 September 2025).
Sarraf
,
S.
,
Kar
,
A.K.
and
Janssen
,
M.
(
2024
), “
How do system and user characteristics, along with anthropomorphism, impact cognitive absorption of chatbots–introducing SUCCAST through a mixed method study
”,
Decision Support Systems
, Vol. 
178
, 114132, doi: .
Schillaci
,
C.E.
,
de Cosmo
,
L.M.
,
Piper
,
L.
,
Nicotra
,
M.
and
Guido
,
G.
(
2024
), “
Anthropomorphic chatbots' for future healthcare services: effects of personality, gender, and roles on source credibility, user satisfaction, and intention to use
”,
Technological Forecasting and Social Change
, Vol. 
199
, 123025, doi: .
Sheng
,
J.
,
Kostyk
,
A.
and
Chatzipanagiotou
,
K.
(
2025
), “
From parasocial interaction to parasocial relationship: a review and research agenda
”,
International Journal of Consumer Studies
, Vol. 
49
No. 
2
, e70038, doi: .
Shi
,
J.
,
Lin
,
C.
,
Wang
,
H.
,
Jung
,
S.
,
Ta
,
N.
,
Liu
,
R.
,
Gao
,
Y.
and
Cao
,
H.
(
2025
), “
Sociable robots or focused speakers? Transforming customer experience with communication style and embodiment type in smart home devices adoption
”,
Behaviour and Information Technology
, Vol. 
44
No. 
14
, pp. 
3559
-
3574
, doi: .
Shrestha
,
N.
(
2021
), “
Factor analysis as a tool for survey analysis
”,
American Journal of Applied Mathematics and Statistics
, Vol. 
9
No. 
1
, pp. 
4
-
11
, doi: .
Song
,
S.W.
and
Shin
,
M.
(
2024
), “
Uncanny valley effects on chatbot trust, purchase intention, and adoption intention in the context of e-commerce: the moderating role of avatar familiarity
”,
International Journal of Human-Computer Interaction
, Vol. 
40
No. 
2
, pp. 
441
-
456
, doi: .
Song
,
J.H.
and
Zinkhan
,
G.M.
(
2008
), “
Determinants of perceived web site interactivity
”,
Journal of Marketing
, Vol. 
72
No. 
2
, pp. 
99
-
113
, doi: .
Sonntag
,
M.
,
Mehmann
,
J.
and
Teuteberg
,
F.
(
2023
), “
Deriving trust-supporting design knowledge for AI-based chatbots in customer service: a use case from the automotive industry
”,
Journal of Organizational Computing and Electronic Commerce
, Vol. 
33
Nos
3-4
, pp. 
178
-
210
, doi: .
Suhaili
,
S.M.
,
Salim
,
N.
and
Jambli
,
M.N.
(
2021
), “
Service chatbots: a systematic review
”,
Expert Systems with Applications
, Vol. 
184
, 115461, doi: .
Sundar
,
S.S.
,
Bellur
,
S.
,
Oh
,
J.
,
Jia
,
H.
and
Kim
,
H.S.
(
2016
), “
Theoretical importance of contingency in human-computer interaction: effects of message interactivity on user engagement
”,
Communication Research
, Vol. 
43
No. 
5
, pp. 
595
-
625
, doi: .
Suthar
,
S.
(
2020
), “
Top 10 AI-powered chatbots to skyrocket your sales
”,
available at:
 https://thenextscoop.com/ai-powered-chatbots-to-skyrocket-sales/ (
accessed
 24 September 2025).
Tabachnick
,
B.G.
and
Fidell
,
L.S.
(
2007
),
Using Multivariate Statistics
,
Pearson
,
Boston
.
Teo
,
H.
,
Oh
,
L.
,
Liu
,
C.
and
Wei
,
K.
(
2003
), “
An empirical study of the effects of interactivity on web user attitude
”,
International Journal of Human-Computer Studies
, Vol. 
58
No. 
3
, pp. 
281
-
305
, doi: .
Toader
,
D.C.
,
Boca
,
G.
,
Toader
,
R.
,
Măcelaru
,
M.
,
Toader
,
C.
,
Ighian
,
D.
and
Rădulescu
,
A.T.
(
2019
), “
The effect of social presence and chatbot errors on trust
”,
Sustainability
, Vol. 
12
No. 
1
, pp. 
256
-
280
, doi: .
Torres-Moraga
,
E.
and
Barra
,
C.
(
2023
), “
Does destination brand experience help build trust? Disentangling the effects on trust and trustworthiness
”,
Journal of Destination Marketing and Management
, Vol. 
27
, 100767, doi: .
Truong
,
T.T.H.
and
Chen
,
J.S.
(
2025
), “
When empathy is enhanced by human-AI interaction: an investigation of anthropomorphism and responsiveness on customer experience with AI chatbots
”,
Asia Pacific Journal of Marketing and Logistics
, Vol. 
37
No. 
12
, pp. 
3908
-
3925
, doi: .
Tsai
,
C.H.
(
2011
), “
The e-commerce model of health websites: an integration of web quality, perceived interactivity, and web outcomes
”,
Journal of Networks
, Vol. 
6
No. 
7
, pp. 
1017
-
1024
, doi: .
Tsai
,
W.H.S.
,
Liu
,
Y.
and
Chuan
,
C.H.
(
2021
), “
How chatbots' social presence communication enhances consumer engagement: the mediating role of parasocial interaction and dialogue
”,
The Journal of Research in Indian Medicine
, Vol. 
15
No. 
3
, pp. 
460
-
482
, doi: .
Tyrväinen
,
O.
and
Karjaluoto
,
H.
(
2025
), “
Unrequited love? A mixed-methods study of parasocial engagement with social media influencers
”,
International Journal of Information Management
, Vol. 
80
, 102845, doi: .
Uniqlo
(
2023
), “
Uniqlo business model
”,
available at:
 https://www.fastretailing.com/eng/group/strategy/uniqlobusiness.html (
accessed
 24 September 2025).
Wang
,
P.
,
Liu
,
S.
,
Zhao
,
M.
,
Yang
,
X.
,
Zhang
,
G.
,
Chu
,
X.
,
Wang
,
X.
,
Zeng
,
P.
and
Lei
,
L.
(
2019
), “
How is problematic smartphone use related to adolescent depression? A moderated mediation analysis
”,
Children and Youth Services Review
, Vol. 
104
, 104384, doi: .
Wang
,
S.
,
Yuan
,
C.
,
Moon
,
H.
and
Kim
,
K.H.
(
2025
), “
Enhancing parasocial brand experience and brand equity: unleashing the advantages of ChatGPT service
”,
International Journal of Consumer Studies
, Vol. 
49
No. 
2
, e70029, doi: .
Williams
,
R.
(
2018
), “
Uniqlo offers personalized style tips within in-app chatbot
”,
available at:
 https://www.marketingdive.com/news/uniqlo-offers-personalized-style-tips-with-in-app-chatbot/529211/ (
accessed
 24 September 2025).
Xu
,
X.
,
Yao
,
Z.
and
Sun
,
Q.
(
2018
), “
Social media environments effect on perceived interactivity: an empirical investigation from WeChat moments
”,
Online Information Review
, Vol. 
43
No. 
2
, pp. 
239
-
255
, doi: .
Xu
,
H.
,
Law
,
R.
,
Lovett
,
J.
,
Luo
,
J.M.
and
Liu
,
L.
(
2024
), “
Tourist acceptance of ChatGPT in travel services: the mediating role of parasocial interaction
”,
Journal of Travel and Tourism Marketing
, Vol. 
41
No. 
7
, pp. 
955
-
972
, doi: .
Yang
,
F.
and
Shen
,
F.
(
2018
), “
Effects of web interactivity: a meta-analysis
”,
Communication Research
, Vol. 
45
No. 
5
, pp. 
635
-
658
, doi: .
Yang
,
B.
,
Sun
,
Y.
and
Shen
,
X.L.
(
2023
), “
Understanding AI-based customer service resistance: a perspective of defective AI features and tri-dimensional distrusting beliefs
”,
Information Processing and Management
, Vol. 
60
No. 
3
, 103257, doi: .
Ye
,
B.H.
,
Barreda
,
A.A.
,
Okumus
,
F.
and
Nusair
,
K.
(
2019
), “
Website interactivity and brand development of online travel agencies in China: the moderating role of age
”,
Journal of Business Research
, Vol. 
99
, pp. 
382
-
389
, doi: .
Yoon
,
D.
and
Youn
,
S.
(
2016
), “
Brand experience on the website: its mediating role between perceived interactivity and relationship quality
”,
Journal of Interactive Advertising
, Vol. 
16
No. 
1
, pp. 
1
-
15
, doi: .
Yuan
,
C.L.
,
Moon
,
H.
,
Kim
,
K.H.
and
Wang
,
S.
(
2021
), “
The influence of parasocial relationship in fashion web on customer equity
”,
Journal of Business Research
, Vol. 
130
, pp. 
610
-
617
, doi: .
Zhang
,
H.
(
2021
), “
Augmenting brand experience in native advertising on social medias using fuzzy-set qualitative comparative analysis
”,
Microprocessors and Microsystems
, Vol. 
82
, 103991, doi: .
Zhang
,
J.
,
Chen
,
Q.
,
Lu
,
J.
,
Wang
,
X.
,
Liu
,
L.
and
Feng
,
Y.
(
2024
), “
Emotional expression by artificial intelligence chatbots to improve customer satisfaction: underlying mechanism and boundary conditions
”,
Tourism Management
, Vol. 
100
, 104835, doi: .
Published by Emerald Publishing Limited. This article is published under the Creative Commons Attribution (CC BY 4.0) licence. Anyone may reproduce, distribute, translate and create derivative works of this article (for both commercial and non-commercial purposes), subject to full attribution to the original publication and authors. The full terms of this licence may be seen at Link to the terms of the CC BY 4.0 licence.

Supplementary data

or Create an Account

Close Modal
Close Modal