Skip to Main Content
Purpose

Drawing on the technology affordance and affinity theories, this study proposes a framework explaining the antecedents and consequences of customers’ smart experiences (CSEs) in the artificial intelligence (AI) chatbot context.

Design/methodology/approach

The quantitative approach employing an online survey was adopted to obtain data from chatbot users (N = 761) and analyzed using structural equation modeling.

Findings

Results from a survey study show that chatbot affordances, including interactivity (two-way communication, active control and synchronicity), selectivity (customization and localization), information (argument quality and source credibility), association (connectivity and sense of safety) and navigation positively affect CSEs (hedonic and cognitive), leading to customer chatbot stickiness through affinity.

Originality/value

Our study provides evidence that supports and extends the affordances and affinity lens by highlighting the roles of specific chatbot affordances that contribute to a positive-smart experience and subsequently enhances customer chatbot stickiness through affinity.

Artificial intelligence (AI) chatbots have recently witnessed exponential growth in interactive marketing, particularly in online customer service (Chung et al., 2020; Wei et al., 2024; Xie et al., 2023). The development of new technologies, such as chatbots, has made interactive marketing one of the fastest-growing fields (Wang, 2022). Particularly, chatbots have significantly transformed the communication landscape, providing customers with superior shopping experiences and direct contact with brands (Wang, 2024). They are designed to mimic human conversations and enhance customers’ experiences by providing them round-the-clock services (Shahzad et al., 2024b; Zhu et al., 2022). Companies are heavily investing in developing AI programs to provide more personalized content and convenience, allowing customers to enjoy a seamless interactive experience (Ferreira et al., 2023; Wang, 2023). The global conversational AI market size is projected to increase from $13.2bn in 2024 to $49.9bn by 2029 (Research-and-Markets, 2024). However, chatbot-related studies often underline that chatbots generally struggle to handle complex and unusual customer queries, resulting in frustration and poor customer experiences (Gnewuch et al., 2023; Schuetzler et al., 2021).

Past research has highlighted that optimal customer experiences can be achieved when AI-based technologies (e.g. chatbots) provide customers with their requested information while offering support anytime and from anywhere (Japutra et al., 2021; Leung et al., 2022; Li et al., 2023). Service providers now largely use chatbots to deliver smart experiences to enhance sales performance and customer loyalty (Fan et al., 2023). While several studies have investigated what we call customers’ smart experiences (CSEs) in different contexts, such as AI technology (Gao et al., 2022), retailer mobile application (Japutra et al., 2021), smart services (Roy et al., 2019) and AR/VR services (Fan et al., 2020), limited studies have paid attention to investigating CSEs in chatbots setting (Fan et al., 2022, 2023; Upadhyay and Kamble, 2023). Fan et al. (2023) call for further research on “smart experiences” (representing cognitive and hedonic customer responses to chatbots) during online service encounters. A positive experience when using customer service chatbots can lead to an association with and perception of the importance of chatbots (chatbot affinity) as well as customer stickiness (continual use of chatbots) with them. Therefore, it is vital to examine factors that could influence CSEs, which subsequently lead to enhanced affinity and customer stickiness in this setting.

When it comes to studying factors that can influence CSEs in extant technology literature, some scholars discovered certain AI characteristics; for example, AI technology stimuli influence smart experiences (Gao et al., 2022) and chatbot anthropomorphism inspires smart experiences (Upadhyay and Kamble, 2023), while others unveiled that chatbot ambidexterity, such as service efficiency and flexibility (Fan et al., 2023) or chatbots sales/service ambidexterity (Fan et al., 2022) can augment smart experiences. Although past studies recognized that AI technologies (e.g. chatbots) can enhance CSEs (Fan et al., 2023; Gao et al., 2022; Upadhyay and Kamble, 2023) the literature has not acknowledged chatbot affordances, including interactivity (two-way communication; active control; synchronicity), selectivity (customization; localization), information (argument quality; source credibility), association (connectivity; sense of safety) and navigation as the potential antecedents of CSEs (hedonic; cognitive).

Furthermore, past research focused on the service quality or characteristic aspects of chatbots when it comes to customer satisfaction/loyalty or stickiness. For instance, Ashfaq et al. (2020) examined information/service quality to explain user satisfaction and usage intentions. Li and Zhang (2023) explored customers’ switching behaviors from human-mediated services to AI in frontline services (e.g. AI chatbots). Li et al. (2024) based their study on arousal and social response theories to examine the impact of AI stimuli on customer stickiness and the mediating effects of social presence. While these studies contribute to chatbot literature, research investigating the impact of CSEs on customer stickiness through perceived chatbot importance (chatbot affinity) remains lacking. This study develops a framework to better understand not only the antecedents (chatbot affordances) of CSEs but also its consequences (affinity and stickiness) in the AI chatbot context.

To achieve our study objectives, this research set to solve three research questions: RQ1. Do chatbot affordances enhance CSEs? RQ2. What is the impact of CSEs on perceived chatbot importance (chatbot affinity)? RQ3. What is the impact of affinity on customer stickiness to chatbots? Our study is based on the technology affordance lens (Gibson, 1979) and affinity theory (Perse, 1986; Rubin, 1981) to understand how chatbot affordances facilitate customer behaviors, such as customer stickiness with the chatbot. The research empirically tests the proposed model by conducting an online survey with 761 chatbot users. We extend the lens of affordances and affinity theory by highlighting the roles of chatbot affordance factors in enhancing CSEs with chatbots, leading to customer stickiness through chatbot affinity. As a result, we provide insights and recommendations to enhance CSEs when using chatbots—a crucial factor in retaining customers with the technology and the brand more generally.

Conversational agents are often considered excellent illustrations of AI technologies that can be text- or voice-based, determined by the mode of communication (Li et al., 2023). Chatbots are text-based conversational agents that simulate human interactions mainly through text, while voice-based conversational agents are called AI voice assistants that interact with users via voice (Jan et al., 2023; Li et al., 2023). Recently, the use of chatbots is considered essential in improving customer service (Jan et al., 2023; Li and Zhang, 2023). They also enhance customer experiences by providing personalized content and enjoyable interaction (Aslam, 2023; Zhu et al., 2022) and guarantee efficiency by providing 24/7 support, thereby improving customer experiences (Loureiro et al., 2021; Shahzad et al., 2024b). Chatbots can also assist users with information seeking, booking and reservations and ordering food (Ashfaq et al., 2020; Jan et al., 2023).

Recent research has explored the multifaceted applications of chatbots across diverse domains, shedding light on their transformative potential in enhancing customer experiences, satisfaction, loyalty or usage intention (see Table 1). For example, Li et al. (2023) stressed that chatbot affordances are critical in enhancing customer experiences and continuance intention. Chung et al. (2020) explored the effectiveness of chatbot-based services in improving customer satisfaction. Ashfaq et al. (2020) reported that information and service quality, along with ease of use, are crucial predictors of satisfaction and chatbot usage intentions. Xie et al. (2023) studied the psychological dimensions of chatbot engagement, revealing the potential for psychological dependence during intense relationships formed with social chatbots.

Table 1

Literature review on AI-based chatbots

ArticleTheoryMethodResearch question/objectiveFinding
Chung et al. (2020) Five-dimension modelSurvey-basedAnalyze the effects of chatbots in luxury brands on customer perceptions based on interaction, trendiness, entertainment, problem-solving capabilities and customizationChatbots in luxury fashion retail brands offer interactive and engaging brand/customer service encounters
Ashfaq et al. (2020) ECM, ISS model and TAMSurvey-basedUnderstand the factors driving users’ satisfaction and their intention to continue using chatbot-based customer servicesInformation quality, service quality, enjoyment, ease of use and usefulness affect continuance intention directly or indirectly via satisfaction
Lee and Li (2023) Affordance actualization and social identitySurvey-basedExamine how chatbot affordances in the banking sector influence customer-brand identification and loyaltyChatbot affordances significantly impact customer-brand identification and loyalty in the banking sector
Li and Zhang (2023) Push-pull-mooring frameworkInterview/surveyExplore factors influencing consumers’ decision to switch from human agents to AI-based conversational agents in customer serviceFactors like low empathy and adaptability of human agents influence consumers’ switching behavior to AI-based conversational agents
Fan et al. (2023) AI chatbot ambidexteritySurvey-basedInvestigate the effects of different types of chatbot ambidexterity (service-sales, efficiency-flexibility and existing-new product selling) on smart experiences and customer patronageDifferent types of chatbot ambidexterity yield varying outcomes in terms of hedonic and cognitive smart experiences and customer patronage
Upadhyay and Kamble (2023) Anthropomorphism and smart experienceSurvey-basedInvestigate the relationship between consumers and conversational chatbots in mobile bankingExplore factors influencing consumers’ brand love for conversational AI chatbots in mobile banking
Xie et al. (2023) Loneliness, trust and personificationMixed-methods approachInvestigate the development of relationships between users and social chatbotsIntense relationships with chatbots can result in psychological dependence, characterized by salience, tolerance and withdrawal

Source(s): Authors’ own work

The above studies provide a comprehensive view of different chatbot applications, demonstrating their potential to enhance customer experiences and usage intention across diverse sectors. Although numerous studies on chatbots have been conducted recently, we are yet to understand how and whether chatbot affordances affect CSE and lead to stickiness through chatbot affinity. We extend the chatbot literature by proposing a model drawing on affordance and affinity theories to explain the antecedents and consequences of CSE in the present context.

Smart experiences can be defined as the emotional and cognitive reactions displayed by consumers toward smart and innovative technologies, such as chatbots (Fan et al., 2023). Scholars have used different taxonomies to measure smart experiences in several domains (Fan et al., 2023; Gao et al., 2022). Roy et al. (2019) proposed an analytical framework explaining smart experiences and recommended six first-order/sub-dimensions, including cognitive, hedonic, social, personal, economic and pragmatic. Previous studies have mainly investigated smart experiences as unidimensional perspectives focusing only on hedonic (Ameen et al., 2021; Shahzad et al., 2024b) or cognitive dimensions (Fan et al., 2020; Zhu et al., 2022). However, scholars acknowledging the hedonic and cognitive key dimensions of smart experience (Fan et al., 2023; Molinillo et al., 2020) describe such experiences as mental processes where hedonic experiences encompass emotions/mood (Molinillo et al., 2020) and cognitive experiences capture perception and problem-solving (Ameen et al., 2021). Chatbots can generate both experiences by providing flexible and tailored services (Fan et al., 2022) and offering quick, accurate and complete information (Gao et al., 2022).

Affordance theory—originating from ecological psychology—explains how individuals perform certain actions/behaviors when observing the opportunities that the environment provides (Gibson, 1979). In the information systems (IS) setting, affordance is defined as “the possibility of guiding the user to the target action through the interaction of a technological object with the user” (Zhou et al., 2023, p. 2). The theory, particularly in the technology setting, provides a robust framework for exploring the interaction between users and technology, offering a nuanced understanding of how technology attributes facilitate specific user actions and behavior (Fang, 2019; Liu, 2003; Ou et al., 2014; Shao et al., 2020). It has extensively been applied in several domains, predominantly in the IS/technology field. For example, Dong and Wang (2018) applied this theory to understand social tie formation in online social commerce. Specifically, interactivity in online social commerce between buyers and sellers fosters the formation of both strong and weak social ties (Dong and Wang, 2018). Similarly, Lee and Li (2023) highlighted how chatbot interactivity in the banking sector impact brand loyalty.

Furthermore, Fang (2019) emphasized how branded apps leverage selectivity affordance to enhance brand loyalty, while Shao et al. (2020) investigated information affordance in social networking sites (SNS), revealing how information quality, along with navigation affordances, influence user satisfaction, leading to SNS stickiness. Additionally, Lee and Li (2023) showed how association affordance enhances customer loyalty in the context of chatbots. In contrast, Zhu and Zhang (2007) research on web portals underscored the importance of navigation affordance (navigability), introducing metrics and guidelines for effective navigation design to enhance user experience, ensuring that users can efficiently find the information or features they need, which is crucial for user satisfaction and retention. Previous research highlighted that the features/characteristics of affordances vary depending on different research contexts (Lee and Li, 2023; Zhou et al., 2023). For example, several affordances can be associated with a single technology because of differences in users’ perceptions and usage intentions; consequently, identifying affordances from a multidimensional perspective is highly crucial (Fang, 2019; Lee and Li, 2023). Thus, we analyzed the extant technological affordances literature and proposed five affordance constructs that are specific to AI-based chatbots: interactivity, selectivity, information, association and navigation (Table 2).

Table 2

Affordances for AI chatbots, definition and sources

AffordanceSubcomponent/dimensionDefinitionExemplar studies
Interactivity affordanceTwo-way communicationAI chatbot facilitates two-way communication between the AI chatbot and the userFang (2019), Liu (2003), Ou et al. (2014), Shao et al. (2020) 
 Active controlAI chatbot allows users to control their experience with it freelyFang (2019), Liu (2003), Ou et al. (2014), Shao et al. (2020) 
 SynchronicityAI chatbot supports users’ instant access to requested information during communicationFang (2019),Liu (2003), Ou et al. (2014), Shao et al. (2020) 
Selectivity affordanceCustomizationAI chatbot can provide customized information to better meet users’ preferences and needsFang (2019), Shen et al. (2013) 
 LocalizationAI chatbot can provide location-specific information to users (e.g. nearby stores)Fang (2019), Shen et al. (2013) 
Information affordanceArgument qualityAI chatbot can offer complete and correct informationShao et al. (2020), Sussman and Siegal (2003) 
 Source credibilityAI chatbot can offer reliable and trustworthy informationShao et al. (2020), Sussman and Siegal (2003) 
Association affordanceConnectivityAI chatbot allows users to use it anytime, anywhereFang (2019), Lee and Li (2023) 
 Sense of safetyAI chatbot can calm users by helping them with specific information (e.g. the nearest store location)Fang (2019) 
Navigation affordance AI chatbot can provide a clear navigation bar/menu (e.g. links/hyperlinks) to find relevant informationZhu and Zhang (2007), Shao et al. (2020) 

Source(s): Authors’ own work

Media affinity theory (Perse, 1986; Rubin, 1981) suggesting “the importance of the medium in the lives of individuals” (Aldás-Manzano et al., 2009, p. 741) has applied in technology/IS literature, highlighting the significant role of technology in diverse aspects of individuals’ activities, enhancing their attachment to and association with technology (Aldás-Manzano et al., 2009; Niu and Mvondo, 2024; Xu and Du, 2018). Particularly, Niu and Mvondo (2024) define affinity in the AI chatbot setting as the perceived importance of chatbots in users’ lives. It is underlined that when a technology/system promptly offers users accurate and personalized information, it improves their affinity with them (Xu and Du, 2018). Technology researchers have examined how affinity influences various factors, such as customers’ satisfaction and usage intentions (Aldás-Manzano et al., 2009; Niu and Mvondo, 2024).

Behavioral reasoning theory (BRT; Westaby, 2005) is more suitable for studying adoption behaviors toward innovations (Ashfaq et al., 2021) because it incorporates reasoning factors that can influence customers’ choices to either adopt or avoid using a particular technology (Shahzad et al., 2024a), such as augmented reality (Nadeem et al., 2024), chatbots (Pillai et al., 2023) or AI-powered conversational agents (Jan et al., 2023). Specifically, it enables scholars to explore positive (facilitators) and negative (inhibitors) factors that affect customer decision-making (Westaby, 2005). In technology settings, positive and motivational factors such as interactivity and personalization (Pillai et al., 2023), customization and augmentation quality (Nadeem et al., 2024) and informativeness and usefulness (Jan et al., 2023) are often considered important factors that inspire customers’ technology adoption intentions. Accordingly, BRT offers a deeper understanding of the influence of chatbot affordances (i.e. interactivity, selectivity and information) on customer stickiness to chatbots through CSEs and affinity.

Demers and Lev (2001) proposed the concept of “stickiness” to websites. They defined it as the ability of a site to retain a user once they have arrived there. It also represents how long a user uses a particular technology and the likeliness/willingness or frequent use intention of the technology (Lin, 2007; Thakur and AlSaleh, 2018). In our case, stickiness refers to the degree to which customers repeatedly use chatbots to perform certain activities. Scholars have studied customer stickiness in different domains and settings, such as social commerce (Zhou et al., 2024), SNS (Shao et al., 2020) and brand love (Mostafa and Temerak, 2024).

Building upon the affordance and affinity lens, we propose a model in the chatbot context (see Figure 1). In this study, we examine five chatbot affordances as antecedents of CSE. Next, we investigate the impact of CSE on chatbot affinity, which is expected to influence customers’ chatbot stickiness.

Figure 1
A conceptual model shows affordances shaping smart chatbot experience, affinity, and stickiness with controls.The conceptual model shows three types of boxes distinguished by their borders. Double-bordered rectangular boxes represent second-order constructs, rounded rectangular boxes represent first-order dimensions of second-order constructs, and single-bordered rectangular boxes represent first-order latent variables. On the left side of the model, nine vertically arranged rounded rectangles appear, labeled from top to bottom as “Two-way communication”, “Active control”, “Synchronicity”, “Customization”, “Localization”, “Argument quality”, “Source credibility”, “Connectivity”, and “Sense of safety”. To the right of these rounded rectangles, four vertically arranged double-bordered rectangles are shown, labeled from top to bottom as “Interactivity affordance”, “Selectivity affordance”, “Information affordance”, and “Association affordance”. Rightward arrows emerge from “Two-way communication”, “Active control”, and “Synchronicity” and point to “Interactivity affordance”. Rightward arrows emerge from “Customization” and “Localization” and point to “Selectivity affordance”. Rightward arrows emerge from “Argument quality” and “Source credibility” and point to “Information affordance”. Rightward arrows emerge from “Connectivity” and “Sense of safety” and point to “Association affordance”. In the center of the model, to the right of these four affordance constructs, a double-bordered rectangle labeled “Customers’ smart experience with A I chatbot” appears. Diagonal arrows labeled “H 1”, “H 2”, “H 3”, and “H 4” emerge respectively from “Interactivity affordance”, “Selectivity affordance”, “Information affordance”, and “Association affordance” and point to “Customers’ smart experience with A I chatbot”. Below this central box, a single-bordered rectangular box labeled “Navigation affordance” appears, with an upward arrow labeled “H 5” pointing to “Customers’ smart experience with A I chatbot”. Above the central box, two rounded rectangular boxes are present side by side, labeled “Hedonic smart experience” on the left and “Cognitive smart experience” on the right. “Customers’ smart experience with A I chatbot” is connected to “Hedonic smart experience” and “Cognitive smart experience” by upward arrows. From “Customers’ smart experience with A I chatbot”, a rightward arrow labeled “H 6” points to a single-bordered rectangular box labeled “A I chatbot affinity”, which is followed by another rightward arrow labeled “H 7” pointing to a single-bordered rectangular box on the far right labeled “Stickiness to A I chatbot”. Above “Stickiness to A I chatbot”, a single-bordered rectangular box labeled “Control variables Age, Gender, Education, Frequency, Type, Income” appears, with a downward arrow pointing to “Stickiness to A I chatbot”. A legend at the bottom right states that double-bordered rectangular boxes represent “second-order constructs”, rounded rectangular boxes represent “first-order dimensions of second-order constructs”, and single-bordered rectangular boxes represent “first-order latent variables”.

Proposed framework

Figure 1
A conceptual model shows affordances shaping smart chatbot experience, affinity, and stickiness with controls.The conceptual model shows three types of boxes distinguished by their borders. Double-bordered rectangular boxes represent second-order constructs, rounded rectangular boxes represent first-order dimensions of second-order constructs, and single-bordered rectangular boxes represent first-order latent variables. On the left side of the model, nine vertically arranged rounded rectangles appear, labeled from top to bottom as “Two-way communication”, “Active control”, “Synchronicity”, “Customization”, “Localization”, “Argument quality”, “Source credibility”, “Connectivity”, and “Sense of safety”. To the right of these rounded rectangles, four vertically arranged double-bordered rectangles are shown, labeled from top to bottom as “Interactivity affordance”, “Selectivity affordance”, “Information affordance”, and “Association affordance”. Rightward arrows emerge from “Two-way communication”, “Active control”, and “Synchronicity” and point to “Interactivity affordance”. Rightward arrows emerge from “Customization” and “Localization” and point to “Selectivity affordance”. Rightward arrows emerge from “Argument quality” and “Source credibility” and point to “Information affordance”. Rightward arrows emerge from “Connectivity” and “Sense of safety” and point to “Association affordance”. In the center of the model, to the right of these four affordance constructs, a double-bordered rectangle labeled “Customers’ smart experience with A I chatbot” appears. Diagonal arrows labeled “H 1”, “H 2”, “H 3”, and “H 4” emerge respectively from “Interactivity affordance”, “Selectivity affordance”, “Information affordance”, and “Association affordance” and point to “Customers’ smart experience with A I chatbot”. Below this central box, a single-bordered rectangular box labeled “Navigation affordance” appears, with an upward arrow labeled “H 5” pointing to “Customers’ smart experience with A I chatbot”. Above the central box, two rounded rectangular boxes are present side by side, labeled “Hedonic smart experience” on the left and “Cognitive smart experience” on the right. “Customers’ smart experience with A I chatbot” is connected to “Hedonic smart experience” and “Cognitive smart experience” by upward arrows. From “Customers’ smart experience with A I chatbot”, a rightward arrow labeled “H 6” points to a single-bordered rectangular box labeled “A I chatbot affinity”, which is followed by another rightward arrow labeled “H 7” pointing to a single-bordered rectangular box on the far right labeled “Stickiness to A I chatbot”. Above “Stickiness to A I chatbot”, a single-bordered rectangular box labeled “Control variables Age, Gender, Education, Frequency, Type, Income” appears, with a downward arrow pointing to “Stickiness to A I chatbot”. A legend at the bottom right states that double-bordered rectangular boxes represent “second-order constructs”, rounded rectangular boxes represent “first-order dimensions of second-order constructs”, and single-bordered rectangular boxes represent “first-order latent variables”.

Proposed framework

Close modal

Interactivity—underlining the significance of online interaction/communication (Shao et al., 2020)—is a multidimensional factor consisting of three dimensions: two-way communication, active control and synchronicity (Fang, 2019; Ou et al., 2014). Two-way communication refers to the bi-directional information flow between users and the technology (Leung et al., 2022). Chatbots can efficiently communicate with customers in real-time and provide accurate responses by appropriately understanding their queries, thus enhancing real-time information for meaningful conversations (Lee and Li, 2023; Li et al., 2023). Active control indicates customers’ perceptions of having control over objects, while synchronicity involves customers’ beliefs that the technology will respond quickly to their needs (Liu, 2003). Chatbots allow customers to ask questions by providing input and starting conversations in their desired direction (Ashfaq et al., 2020; Jan et al., 2023). Prior literature highlighted that customers often use chatbots for their interactivity (Lee and Li, 2023), and a chatbot with high levels of interactivity can develop strong relationships with customers, enhancing their satisfaction and positive experiences (Fang, 2019; Li et al., 2023). Therefore, chatbot interactivity can be crucial in creating an effective and smart experience. Interactive attributes enable chatbots to communicate with customers at any time and from anywhere (Lee and Li, 2023; Li and Zhang, 2023), allowing them to control their interactions by asking a set of questions and assisting them by providing relevant information to their queries in real-time, thus contributing to enhancing a smart experience. This leads us to propose the following hypothesis:

H1.

Interactivity affordance is positively related to smart experiences.

Selectivity affordance includes two key dimensions: customization and localization (Fang, 2019; Shen et al., 2013). Customization indicates that customers believe the system provides information according to their preferences, whereas localization specifies belief that the system provides information based on their location (Fang, 2019). Chatbots can offer customized/personalized information to customers by asking about their preferences, analyzing input and behavior and remembering past interactions (Chung et al., 2020; Li and Zhang, 2023; Zhu et al., 2022). Selectivity affordances can better meet customers’ needs, improving relationships and customer experiences (Fang, 2019). Chatbots can suggest to customers nearby points (e.g. restaurants and services), locally relevant news and events and current weather updates (Ashfaq et al., 2020; Chung et al., 2020). Moreover, they are frequently being used by customers to get updates about local information, and a high selectivity chatbot can provide customers with optimal information and services tailored to their specific locations and preferences (Fang, 2019; Shen et al., 2013). Selectivity can also be an important factor in forming a positive experience with the chatbot, leading us to propose the following hypothesis:

H2.

Selectivity affordance is positively related to smart experiences.

Based on adoption theories, Sussman and Siegal (2003) categorized information affordance into two distinct dimensions: argument quality and source credibility, where the first category involves customers’ perceptions that the obtained information is accurate and complete, while the second indicates perceptions of trustworthiness and reliability of sources of information (Shao et al., 2020; Sussman and Siegal, 2003). People mostly use the internet and communication technology to get valuable information in real-time. Like other communication technologies, one of the key motives for customers to utilize chatbots is obtaining valuable real-time information (Li et al., 2023; Loureiro et al., 2021). Accordingly, when individuals perceive that the information offered by the chatbot is complete and reliable, they are more likely to perceive them as valuable (Ashfaq et al., 2020). Previous research pointed out that customer experience can be enhanced by providing high-quality and trustworthy information in chatbot settings (Kushwaha et al., 2021). Therefore, we propose:

H3.

Information affordance is positively related to smart experiences.

Wagner et al. (2014) defined association as relationships between individuals or between people and content/information. Association affordance has two dimensions: connectivity and sense of safety (Fang, 2019). Connectivity “denotes a physical association between users and technology that allows users to access services anytime and anywhere” (Fang, 2019, p. 380). One of the key attributes of chatbots is their always availability on websites and apps, and customers can use them anytime and from any location to get immediate information or support (Lee and Li, 2023; Li et al., 2023). Sense of safety “represents a psychological aspect of association wherein users feel safe, easiness, and calmness when using technology in an unfamiliar place” (Fang, 2019, p. 380). For example, a branded app has a great ability to calm customers by offering information about the nearest store location or brand information (Fang, 2019). Using specific technology also creates an association in the form of conversations on social media platforms (Li and Zhang, 2023). In the chatbot context, associations occur as relationships between users and information such that they utilize chatbots to connect with content/information and pertinent products/services in real-time (Lee and Li, 2023). The timely response to inquiries contributes to the formation of a positive customer experience, and when customers receive answers from the chatbot at any moment and from anywhere, their positive experience may be enhanced (Fang, 2019; Li et al., 2023). Therefore, we propose:

H4.

Association affordance is positively related to smart experiences.

Navigation in the context of a website refers to the process by which users achieve their objectives, such as finding relevant information to perform certain tasks (Zhu and Zhang, 2007). Similarly, navigation in chatbots can refer to the process by which customers using the chatbot’s functionalities attain their goals (e.g. getting pertinent information). Navigation has been considered one of the most effective ways to assist users in addressing their inquiries (Zhang and Von Dran, 2001). Thus, navigation affordance can play a crucial role in enhancing CSE with chatbots. For example, chatbots, through conversation, can enhance CSE by suggesting clear options and links/hyperlinks, thus enabling customers to find relevant information quickly to complete tasks. The more customers believe the chatbot can provide pertinent information to complete tasks by suggesting understandable paths/links, the more positive their experience will be. Therefore, we propose:

H5.

Navigation affordance is positively related to smart experiences.

As defined previously, smart experiences are related to consumers’ emotional and cognitive reactions toward chatbots (Fan et al., 2023). Researchers agree that chatbots can generate smart experiences by providing comprehensive information and tailored services (Fan et al., 2022). Service providers offering chatbot-based services put considerable effort into developing perceived importance (affinity) of chatbots (Niu and Mvondo, 2024) by enhancing customers’ hedonic and cognitive experiences through high-quality service delivery (Fang, 2019). Consequently, CSEs with chatbots might develop chatbots’ affinity (perceived chatbot importance). For example, when interacting with a chatbot, customers with favorable hedonic and cognitive experiences will likely consider them more important to perform their activities. Thus, smart experiences can contribute to enhancing customers’ perceptions of the importance of chatbots. Moreover, there is evidence in previous studies that demonstrates customer experience influences affinity in different settings (e.g. Wolf et al., 2023). Aligning to prior research, we propose the following hypothesis in the present context:

H6.

Smart experiences are positively related to affinity.

As discussed earlier, chatbot affinity refers to customers’ perceived importance of chatbots in their lives (Niu and Mvondo, 2024). Several scholars have investigated how affinity affects users’ attitudes, satisfaction and intentions toward using specific technology (Aldás-Manzano et al., 2009; Niu and Mvondo, 2024; Xu and Du, 2018). Although affinity has been widely studied in several domains and has significantly influenced customer satisfaction and intention toward the technology (Aldás-Manzano et al., 2009; Niu and Mvondo, 2024), no research has been conducted to identify its role in chatbot stickiness. The present study proposes that chatbot affinity may influence customer stickiness to chatbots. For example, when interacting with the chatbot, if customers develop a positive affinity toward them based on their interaction and the content of the information they deliver (Niu and Mvondo, 2024), customers may be more likely to expand their chatbot usage (stickiness). Accordingly, chatbot affinity can significantly enhance customer chatbot stickiness, leading us to propose the following hypothesis:

H7.

Affinity is positively related to stickiness.

The survey items were adapted from well-established academic sources and measured using a 7-point scale (Likert) ranging from “strongly disagree” (score = 1) to “strongly agree” (score = 7). The present study measured interactivity affordance (Fang, 2019; Shao et al., 2020), selectivity affordance (Fang, 2019; Shen et al., 2013), information affordance (Shao et al., 2020; Sussman and Siegal, 2003), association affordance (Fang, 2019) and smart experience (Fan et al., 2023) as second-order constructs, while navigation affordance (Shao et al., 2020), affinity (Niu and Mvondo, 2024) and stickiness (Lin, 2007; Shao et al., 2020) were evaluated as first-order reflective constructs. Stickiness was measured using a four-item scale, whereas all other constructs were measured based on a three-item scale.

We collected data from chatbot users using the Prolific platform to test the hypothesized model. Following Guha et al. (2023), a screening study was initially conducted to identify participants who had previously interacted with chatbots by requesting 1,000 respondents (e.g. whether they used the chatbot before, how often, and their willingness to take part in a follow-up study). As a result of the screening study, 889 participants who specified they had interacted with chatbots were identified. Of them, 792 participated in the subsequent study. There were, however, 31 respondents who did not pass the attention check question, resulting in 761 responses (Table 3). We also asked respondents about their purpose for using chatbots (e.g. For what purpose do you mostly use the chatbot?), and Figure 2 shows that users mostly use chatbots to ask questions, get help, solve problems and receive quick information.

Figure 2
A word cloud highlights common uses of chatbots with prominent terms like “help”, “questions”, and “information”.The word cloud shows prominent terms related to chatbot use displayed in varying sizes, with the largest words reading “help”, “questions”, “information”, and “use” at the center. Surrounding these central terms are frequently appearing words such as “ask”, “answers”, “website”, “time”, “get”, “chat”, “service”, “support”, and “search”. Additional smaller words appear throughout the cloud, including “personal”, “problem”, “shopping”, “banking”, “directions”, “weather”, “email”, “contact”, “tasks”, “simple”, “quick”, “issues”, “content”, “recommendations”, “accounts”, “resume”, “general”, “specific”, “usually”, “sometimes”, and “online”. The arrangement places the largest and most frequent words near the center, with less frequent terms distributed around them, visually emphasizing common themes associated with chatbot assistance and information seeking.

Word cloud analysis

Figure 2
A word cloud highlights common uses of chatbots with prominent terms like “help”, “questions”, and “information”.The word cloud shows prominent terms related to chatbot use displayed in varying sizes, with the largest words reading “help”, “questions”, “information”, and “use” at the center. Surrounding these central terms are frequently appearing words such as “ask”, “answers”, “website”, “time”, “get”, “chat”, “service”, “support”, and “search”. Additional smaller words appear throughout the cloud, including “personal”, “problem”, “shopping”, “banking”, “directions”, “weather”, “email”, “contact”, “tasks”, “simple”, “quick”, “issues”, “content”, “recommendations”, “accounts”, “resume”, “general”, “specific”, “usually”, “sometimes”, and “online”. The arrangement places the largest and most frequent words near the center, with less frequent terms distributed around them, visually emphasizing common themes associated with chatbot assistance and information seeking.

Word cloud analysis

Close modal
Table 3

Profile of the respondents

CharacteristicsDistributionFrequency%
GenderMale37148.75
Female39051.25
FrequencyOnce a week30740.34
2–5 times a month31441.26
6–10 times a month547.10
>10 times a month8611.30
AgeBelow 20141.84
20–3040252.83
31–4020827.33
41–508611.30
Above 50516.70
EducationLess than high school40.53
High school11414.98
Some college13918.27
Bachelor’s degree38049.93
Master’s degree10413.67
Doctoral degree121.58
Professional degree81.05
Income (household income)$0–$5,00013417.61
$5,001–$10,00010613.93
$10,001–$20,00013117.21
$20,001–$30,00012616.56
$30,001–$40,0008611.30
Above $40,00017823.39

Source(s): Authors’ own work

This study employed SmartPLS4 and the PLS-SEM approach to analyze the collected data (Ringle et al., 2022; Sarstedt et al., 2017). The PLS-SEM is commonly preferred to test complex models (Sarstedt et al., 2017), which also accomplishes high levels of statistical power for testing the hypotheses (Guha et al., 2023). Notably, earlier technology studies, including research on chatbots, have employed this approach (e.g. Ashfaq et al., 2023; Li and Zhang, 2023).

The CMB was confirmed using two commonly used methods for PLS-SEM studies. First, the “variance inflation factor” (VIF) was below 5 (Hair et al., 2019). Second, the “marker variable” method proposed by Lindell and Whitney (2001) was performed in our study following a recent chatbot study (Chandra et al., 2022). This study considered “attitude toward energy drink consumption” (3 items: e.g. “I like energy drinks”) as a marker variable (Swani, 2021) because this variable is not theoretically related and might display lower correlations with other variables suggested in our model (Lindell and Whitney, 2001). The analysis exhibited that the correlations between the marker and other study variables are notably low, where the highest correlation was 0.109 (between customization and marker variable), which is greatly below 0.300 (Chandra et al., 2022). Furthermore, when correlations are squared, we observed that the largest shared variance with the marker variable is less than 1.2% (0.109 squared), and all other correlations are far below the threshold of 0.900, signifying unsubstantial CMB in our study (Chandra et al., 2022).

Hair et al. (2019) suggest that indicator loading should be >0.70 and our study findings fulfilled the suggested criteria (see Table 4). Composite reliability (CR) along with Cronbach’s alpha (CA) are often used in the MM to assess the reliability of the constructs, which should also be >0.70 (Hair et al., 2019). The results confirmed that CR and CA values of all constructs are above the implied level (Table 4). The convergent validity based on the average variance extracted (AVE) metric was addressed next, and AVE values in Table 4 are above the recommended 0.50 level (Hair et al., 2019). Further, the discriminant validity (DV) was examined using the HTMT approach, and our results met the criteria, for example, HTMT <0.85 (Figure 3) (Henseler et al., 2015).

Figure 3
A vertical bar chart shows H T M T ratios for multiple construct pairs along a labeled horizontal axis.The vertical bar chart titled “Heterotrait-monotrait ratio (H T M T)” shows the vertical axis labeled “Heterotrait-monotrait ratio (H T M T)” with numeric markings ranging from 0.00 to 0.84 in increments of 0.06 units, and the horizontal axis lists multiple construct labels written. Each construct is represented by a narrow vertical bar extending upward from the baseline to its corresponding H T M T value. The data for the bars is shown as follows: Affinity points to Active control: 0.411 Argument quality points to Affinity: 0.426 Cognitive smart experience points to Affinity: 0.63 Connectivity points to Active control: 0.383 Connectivity points to Argument quality: 0.731 Customization points to Active control: 0.654 Customization points to Connectivity: 0.393 Hedonic smart experience points to Affinity: 0.536 Hedonic smart experience points to Cognitive smart experience: 0.776 Hedonic smart experience points to Customization: 0.697 Localization points to Affinity: 0.362 Localization points to Cognitive smart experience: 0.392 Localization points to Customization: 0.49 Navigation affordance points to Active control: 0.539 Navigation affordance points to Argument quality: 0.676 Navigation affordance points to Connectivity: 0.53 Navigation affordance points to Hedonic smart experience: 0.673 Sense of safety points to Active control: 0.636 Sense of safety points to Argument quality: 0.758 Sense of safety points to Connectivity: 0.481 Sense of safety points to Hedonic smart experience: 0.813 Sense of safety points to Navigation affordance: 0.703 Source credibility points to Affinity: 0.393 Source credibility points to Cognitive smart experience: 0.533 Source credibility points to Customization: 0.621 Source credibility points to Localization: 0.478 Source credibility points to Sense of safety: 0.733 Stickiness points to Affinity: 0.761 Stickiness points to Cognitive smart experience: 0.74 Stickiness points to Customization: 0.615 Stickiness points to Localization: 0.377 Stickiness points to Sense of safety: 0.733 Synchronicity points to Active control: 0.639 Synchronicity points to Argument quality: 0.584 Synchronicity points to Connectivity: 0.52 Synchronicity points to Hedonic smart experience: 0.67 Synchronicity points to Navigation affordance: 0.67 Synchronicity points to Source credibility: 0.569 Two-way communication points to Active control: 0.56 Two-way communication points to Argument quality: 0.493 Two-way communication points to Connectivity: 0.387 Two-way communication points to Hedonic smart experience: 0.514 Two-way communication points to Navigation affordance: 0.414 Two-way communication points to Source credibility: 0.478 Two-way communication points to Synchronicity: 0.484 Note: All numerical data values are approximated.

HTMT ratio

Figure 3
A vertical bar chart shows H T M T ratios for multiple construct pairs along a labeled horizontal axis.The vertical bar chart titled “Heterotrait-monotrait ratio (H T M T)” shows the vertical axis labeled “Heterotrait-monotrait ratio (H T M T)” with numeric markings ranging from 0.00 to 0.84 in increments of 0.06 units, and the horizontal axis lists multiple construct labels written. Each construct is represented by a narrow vertical bar extending upward from the baseline to its corresponding H T M T value. The data for the bars is shown as follows: Affinity points to Active control: 0.411 Argument quality points to Affinity: 0.426 Cognitive smart experience points to Affinity: 0.63 Connectivity points to Active control: 0.383 Connectivity points to Argument quality: 0.731 Customization points to Active control: 0.654 Customization points to Connectivity: 0.393 Hedonic smart experience points to Affinity: 0.536 Hedonic smart experience points to Cognitive smart experience: 0.776 Hedonic smart experience points to Customization: 0.697 Localization points to Affinity: 0.362 Localization points to Cognitive smart experience: 0.392 Localization points to Customization: 0.49 Navigation affordance points to Active control: 0.539 Navigation affordance points to Argument quality: 0.676 Navigation affordance points to Connectivity: 0.53 Navigation affordance points to Hedonic smart experience: 0.673 Sense of safety points to Active control: 0.636 Sense of safety points to Argument quality: 0.758 Sense of safety points to Connectivity: 0.481 Sense of safety points to Hedonic smart experience: 0.813 Sense of safety points to Navigation affordance: 0.703 Source credibility points to Affinity: 0.393 Source credibility points to Cognitive smart experience: 0.533 Source credibility points to Customization: 0.621 Source credibility points to Localization: 0.478 Source credibility points to Sense of safety: 0.733 Stickiness points to Affinity: 0.761 Stickiness points to Cognitive smart experience: 0.74 Stickiness points to Customization: 0.615 Stickiness points to Localization: 0.377 Stickiness points to Sense of safety: 0.733 Synchronicity points to Active control: 0.639 Synchronicity points to Argument quality: 0.584 Synchronicity points to Connectivity: 0.52 Synchronicity points to Hedonic smart experience: 0.67 Synchronicity points to Navigation affordance: 0.67 Synchronicity points to Source credibility: 0.569 Two-way communication points to Active control: 0.56 Two-way communication points to Argument quality: 0.493 Two-way communication points to Connectivity: 0.387 Two-way communication points to Hedonic smart experience: 0.514 Two-way communication points to Navigation affordance: 0.414 Two-way communication points to Source credibility: 0.478 Two-way communication points to Synchronicity: 0.484 Note: All numerical data values are approximated.

HTMT ratio

Close modal
Table 4

Reliability and validity

FactorItemLoadingCronbach’s alphaCRAVE
Two-way communication (TWC)TWC10.8020.7170.7480.633
TWC20.743   
TWC30.837   
Active control (ACC)ACC10.8620.8420.8420.760
ACC20.872   
ACC30.881   
Synchronicity (SYN)SYN10.8300.8430.8530.761
SYN20.911   
SYN30.874   
Customization (CUS)CUS10.9100.8830.8830.810
CUS20.910   
CUS30.880   
Localization (LOC)LOC10.9110.9090.9100.846
LOC20.925   
LOC30.923   
Argument quality (ARQ)ARQ10.8410.8260.8280.742
ARQ20.855   
ARQ30.887   
Source credibility (SOC)SOC10.9080.8580.8610.781
SOC20.930   
SOC30.809   
Connectivity (CON)CON10.8930.9010.9030.835
CON20.909   
CON30.938   
Sense of safety (SOS)SOS10.8980.8880.8910.817
SOS20.927   
SOS30.886   
Navigation affordance (NVA)NVA10.8400.8480.8630.766
NVA20.885   
NVA30.900   
Hedonic smart experience (HSE)HSE10.9210.9250.9260.869
HSE20.923   
HSE30.953   
Cognitive smart experience (CoSE)CoSE10.9260.9140.9140.853
CoSE20.927   
CoSE30.918   
Affinity (AFF)AFF10.9070.8530.8870.773
AFF20.929   
AFF30.796   
Stickiness (STI)STI10.8020.9050.9160.779
STI20.899   
STI30.908   
STI40.917   

Source(s): Authors’ own work

The SM was used to verify hypotheses. Regarding outcomes, results indicated that interactivity (H1: β = 0.307***; t = 8.741), selectivity (H2: β = 0.177***; t = 5.665), information (H3: β = 0.098*; t = 2.173), association (H4: β = 0.274***; t = 6.871) and navigation affordance (H5: β = 0.081*; t = 2.303) impacts CSE, supporting H1H5. Further, the results showed that CSE positively influences affinity (β = 0.566***; t = 24.16), which in turn positively impacts stickiness (β = 0.690***; t = 37.82). Given these results, H6 and H7 are also supported. Finally, the analysis showed that control variables have insignificant impacts (Table 5). We further assessed our model under the coefficient of determinants/explanatory power (R2), predictive relevance (Q2) and effect size (f2). Our model explained 62.4% of the variance in CSE, 32% in affinity and 47.6% in stickiness, signifying a satisfactory outcome (Hair et al., 2019). In addition, as can be seen in Table 5, our model had good predictive relevance, leading to acceptable outcomes (Hair et al., 2019).

Table 5

Structural model test results

Hypothesis and relationshipBeta (β)t-statisticsp-valuesSignificant?
Interactivity affordance → CSE0.307***8.7410.000Yes
Selectivity affordance → CSE0.177***5.6650.000Yes
Information affordance → CSE0.098*2.1730.040Yes
Association affordance → CSE0.274***6.8710.000Yes
Navigation affordance → CSE0.081*2.3030.031Yes
CSE → affinity0.566***24.160.000Yes
Affinity → stickiness0.690***37.820.000Yes
Control variables
Frequency → stickiness0.0170.6630.508No
Gender → stickiness0.0190.3660.715No
Age → stickiness−0.0140.5180.605No
Type → stickiness0.0281.1130.141No
Education → stickiness0.0361.3890.166No
Income → stickiness−0.0521.8040.072No
R2Q2
CSE0.6220.617
Affinity0.3200.227
Stickiness0.4760.325

Note(s): Customers’ smart experience (CSE), ***p < 0.001; **p < 0.01; *p < 0.05

Source(s): Authors’ own work

In this research, we set out to investigate the antecedents and consequences of CSEs in the chatbot context. To address our RQs, we developed a research model and conducted a quantitative study using a cross-sectional approach. Regarding the study’s outcomes, we found that chatbot affordances, including interactivity, selectivity, information, association and navigation significantly influence CSE. These findings indicate that when a chatbot facilitates two-way communication, offers customized information, delivers accurate/trustworthy information, allows customers to connect anytime/anywhere and provides concise and clear links, customers perceive they have positive experiences. The findings are consistent with extant technology studies stating that technology affordances, such as virtual reality, branded apps or chatbot affordances, can develop a strong relationship with customers and create unique customer experiences (Fang, 2019; Leung et al., 2022; Li et al., 2023).

Furthermore, our study highlighted the significant effect of CSEs on chatbot affinity. This outcome suggests that when customers perceive a positive experience with chatbots, they develop a strong attachment to and association with them, thereby enhancing chatbot affinity. This finding is supported by Xu and Du’s (2018) study, which observed that a system providing users with accurate and comprehensive information along with personalized and on-time services boosts users’ affinity with the system. Finally, we verified that affinity is positively associated with customer stickiness with chatbots; thus, affinity is a key determinant of stickiness that reflects customers’ perceptions of being committed to continued use of chatbots (Li et al., 2024; Lin, 2007). This finding implies that increased perceptions of the importance of chatbots (chatbot affinity) contribute to greater stickiness to chatbots, enriching our understanding of affinity in facilitating customer chatbot stickiness.

Our study provides evidence that supports and extends affordance and affinity theories by highlighting the roles of specific chatbot affordance factors that contribute to positive-smart experiences and subsequently enhance customer chatbot stickiness through affinity. First, while CSEs were studied in diverse domains, including mobile apps (Japutra et al., 2021), AR/VR services (Fan et al., 2020) and AI technology (Gao et al., 2022), limited studies examined the notion of CSEs in chatbot settings (Fan et al., 2022, 2023). To further explore this notion and subsequently respond to Fan et al.’s (2023) call to study smart experiences in the chatbot domain, we advance current research on chatbots by specifically studying the antecedents and consequences of CSEs.

Second, as an intelligence agent rather than merely a software, the chatbot not only provides the basic functions related to searching and retrieving information but it can significantly enhance customers’ experiences by engaging with them through personalized interactions and providing anytime/anyplace connectivity (Lee and Li, 2023; Shahzad et al., 2024b; Zhu et al., 2022). Likewise, we advance extant literature by showing that enhanced levels of chatbot affordances improve CSEs. While several studies investigated the impact of technology affordances on diverse factors, including customer inspiration (Zhou et al., 2023), brand competence/warmth (Lee and Li, 2023) and satisfaction (Shao et al., 2020), these studies have predominantly overlooked the impact of affordances on CSEs in technology research in general, and chatbot literature in particular. As such, the present study expands our understanding by studying the role of affordances (interactivity, selectivity, information, association, navigation) as antecedents of CSEs in the under-researched chatbot context.

Third, in contrast to earlier work focusing on examining the influence of CSEs on word-of-mouth (Gao et al., 2022; Roy et al., 2019) and customer patronage intention (Fan et al., 2022, 2023), our research takes this a step further by exploring the impact of CSEs on chatbot affinity, thereby proposing a novel relationship. According to Wang (2025), an article contributes to extant literature when it presents new information or expands current knowledge. As such, this study’s findings contribute to existing literature by showing how CSE enhances chatbot affinity. Lastly, this study proposes another new association that expands our understanding of the imperative role of affinity that drives customer chatbot stickiness. Several technology studies examined the factors that drive customer stickiness (e.g. Shao et al., 2020); however, no study investigated how affinity drives customer stickiness. We theorize affinity in the chatbot context and confirm that it is an important antecedent for enhancing customer chatbot stickiness.

Our study offers practical insights into the types of affordances that could increase CSEs toward chatbots, which would be beneficial for chatbot designers and developers. In particular, a chatbot’s ability to facilitate two-way communication, allowing users to connect anytime/anywhere connectivity, and providing customized and engaging interactions should be prioritized as core features. Interactivity, association and selectivity affordances are often considered important factors in technology settings generally, and chatbots particularly, to improve customer service and experience (Fang, 2019; Lee and Li, 2023; Li et al., 2023). Accordingly, managers should design chatbots by focusing on their affordance features, particularly emphasizing interactivity, selectivity and association of chatbot affordances to enhance CSEs and customer stickiness to chatbots. Accurate/reliable responses and ease of navigation provide feelings of comfort, and these are other core chatbot features that should be emphasized. Such chatbot affordance factors enhance CSEs, leading to customer chatbot stickiness through affinity, which would be useful insights for chatbot designers and developers.

Next, our study found that CSEs facilitate chatbot affinity, which in turn improves customer stickiness to chatbots. Managers should note that enhancing affordance features of chatbots not only boost CSEs but also increase customer stickiness to chatbots through affinity. Customers often become attached to chatbots when their services provide experiential value (Li et al., 2023). Besides, we also offer some industry-specific design recommendations. For instance, retail chatbots can focus on personalized recommendations with clear opt-in mechanisms to boost customer experience without overstepping privacy boundaries, while healthcare chatbots should prioritize strict privacy controls to align with sensitive patient information needs. Future research could explore how variables like social influence, information sensitivity and platform identification further shape the balance between privacy concerns and perceived benefits in chatbot adoption (Chen et al., 2024). By offering industry-specific recommendations, this study reinforces its contributions to understanding the complexities of AI-driven customer interactions while providing actionable insights for chatbot developers.

This study acknowledges certain limitations. First, the study does not categorize chatbot usage based on purposes such as transactional, informational or interactional. Future research could add contextual relevance to our findings by exploring the difference between transactional chatbots and informational chatbots and their impact on smart experiences and usage intentions. Second, future research may explore chatbot services in other industries, such as healthcare or finance, where user needs and privacy concerns vary, which could yield valuable insights. Third, this study did not include any potential moderating variable, thus we suggest considering potential moderating variables such as tech-savviness or the specific purpose of chatbot interactions (e.g. transactional versus informational) in future studies. Finally, the present research was primarily based on the affordance lens to explain customer chatbot stickiness, where we only included enabler factors in our model while ignoring the role of inhibitors such as privacy risks and technology anxiety that may discourage customers from sticking with chatbots. Future research can explore the balance between personalization and privacy concerns—a matter of growing importance in AI and digital interactions—using the privacy calculus model or BRT to contribute to chatbot adoption literature. For example, the Privacy Calculus Model emphasizes trade-offs users face between perceived benefits (e.g. personalized experiences) and risks (e.g. privacy concerns) in their interactions with chatbots (Chen et al., 2024). While chatbots provide several benefits (e.g. personalized interactions), these benefits often coexist with significant privacy concerns, such as data misuse or inadequate transparency in data handling (Wang, 2024). Consequently, scholars should prioritize developing a model to investigate trade-offs between perceived benefits and risks within a single framework in the chatbot setting.

We acknowledge the support of the College of Business and Law Melbourne and Vietnam Collaborative Project Support Scheme in funding this research.

Conflict of interest: All the co-authors have agreed to the inclusion of their names. All the authors declare that they have no conflict of interest.

Aldás‐Manzano
,
J.
,
Ruiz‐Mafé
,
C.
and
Sanz‐Blas
,
S.
(
2009
), “
Exploring individual personality factors as drivers of M‐shopping acceptance
”,
Industrial Management and Data Systems
, Vol. 
109
No. 
6
, pp. 
739
-
757
, doi: .
Ameen
,
N.
,
Tarhini
,
A.
,
Reppel
,
A.
and
Anand
,
A.
(
2021
), “
Customer experiences in the age of artificial intelligence
”,
Computers in Human Behavior
, Vol. 
114
, 106548, doi: .
Ashfaq
,
M.
,
Yun
,
J.
,
Yu
,
S.
and
Loureiro
,
S.M.C.
(
2020
), “
I, Chatbot: modeling the determinants of users' satisfaction and continuance intention of AI-powered service agents
”,
Telematics and Informatics
, Vol. 
54
, 101473, doi: .
Ashfaq
,
M.
,
Zhang
,
Q.
,
Ali
,
F.
,
Waheed
,
A.
and
Nawaz
,
S.
(
2021
), “
You plant a virtual tree, we’ll plant a real tree: understanding users' adoption of the Ant Forest mobile gaming application from a behavioral reasoning theory perspective
”,
Journal of Cleaner Production
, Vol. 
310
, 127394, doi: .
Ashfaq
,
M.
,
Yu
,
S.
,
Liu
,
Y.
and
Luo
,
K.
(
2023
), “
Usage intensity of mobile instant messengers for work and employees' anxiety
”,
International Journal of Mobile Communications
, Vol. 
22
No. 
3
, pp. 
278
-
295
, doi: .
Aslam
,
F.
(
2023
), “
The impact of artificial intelligence on chatbot technology: a study on the current advancements and leading innovations
”,
European Journal of Technology
, Vol. 
7
No. 
3
, pp. 
62
-
72
, 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
,
X.
,
Wu
,
M.
,
Cheng
,
C.
and
Mou
,
J.
(
2024
), “
Weighing user's privacy calculus on personal information disclosure: the moderating effect of social media identification
”,
Online Information Review
,
Vol. ahead-of-print No. ahead-of-print
. doi: .
Chung
,
M.
,
Ko
,
E.
,
Joung
,
H.
and
Kim
,
S.J.
(
2020
), “
Chatbot e-service and customer satisfaction regarding luxury brands
”,
Journal of Business Research
, Vol. 
117
, pp. 
587
-
595
, doi: .
Demers
,
E.
and
Lev
,
B.
(
2001
), “
A rude awakening: Internet shakeout in 2000
”,
Review of Accounting Studies
, Vol. 
6
, pp. 
2
-
3
, doi: .
Dong
,
X.
and
Wang
,
T.
(
2018
), “
Social tie formation in Chinese online social commerce: the role of IT affordances
”,
International Journal of Information Management
, Vol. 
42
, pp. 
49
-
64
, doi: .
Fan
,
X.
,
Chai
,
Z.
,
Deng
,
N.
and
Dong
,
X.
(
2020
), “
Adoption of augmented reality in online retailing and consumers' product attitude: a cognitive perspective
”,
Journal of Retailing and Consumer Services
, Vol. 
53
, 101986, doi: .
Fan
,
H.
,
Han
,
B.
,
Gao
,
W.
and
Li
,
W.
(
2022
), “
How AI chatbots have reshaped the frontline interface in China: examining the role of sales–service ambidexterity and the personalization–privacy paradox
”,
International Journal of Emerging Markets
, Vol. 
17
No. 
4
, pp. 
967
-
986
, doi: .
Fan
,
H.
,
Gao
,
W.
and
Han
,
B.
(
2023
), “
Are AI chatbots a cure-all? The relative effectiveness of chatbot ambidexterity in crafting hedonic and cognitive smart experiences
”,
Journal of Business Research
, Vol. 
156
, 113526, doi: .
Fang
,
Y.-H.
(
2019
), “
An app a day keeps a customer connected: explicating loyalty to brands and branded applications through the lens of affordance and service-dominant logic
”,
Information and Management
, Vol. 
56
No. 
3
, pp. 
377
-
391
, doi: .
Ferreira
,
M.
,
Loureiro
,
S.M.C.
,
Ashfaq
,
M.
and
Pereira
,
H.
(
2023
), “
Tourist-virtual assistant communication: the power of attachment and authenticity
”,
Anatolia
, Vol. 
34
No. 
4
, pp. 
566
-
578
, doi: .
Gao
,
J.
,
Ren
,
L.
,
Yang
,
Y.
,
Zhang
,
D.
and
Li
,
L.
(
2022
), “
The impact of artificial intelligence technology stimuli on smart customer experience and the moderating effect of technology readiness
”,
International Journal of Emerging Markets
, Vol. 
17
No. 
4
, pp. 
1123
-
1142
, doi: .
Gibson
,
J.J.
(
1979
),
The Ecological Approach to Visual Perception
,
Houghton Mifflin., Psychology Press
,
Boston
.
Gnewuch
,
U.
,
Morana
,
S.
,
Hinz
,
O.
,
Kellner
,
R.
and
Maedche
,
A.
(
2023
), “
More than a bot? The impact of disclosing human involvement on customer interactions with hybrid service agents
”,
Information Systems Research
, Vol. 
35
No. 
3
, pp. 
936
-
955
, doi: .
Guha
,
A.
,
Bressgott
,
T.
,
Grewal
,
D.
,
Mahr
,
D.
,
Wetzels
,
M.
and
Schweiger
,
E.
(
2023
), “
How artificiality and intelligence affect voice assistant evaluations
”,
Journal of the Academy of Marketing Science
, Vol. 
51
No. 
4
, pp. 
843
-
866
, doi: .
Hair
,
J.F.
,
Risher
,
J.J.
,
Sarstedt
,
M.
and
Ringle
,
C.M.
(
2019
), “
When to use and how to report the results of PLS-SEM
”,
European Business Review
, Vol. 
31
No. 
1
, pp. 
2
-
24
, doi: .
Henseler
,
J.
,
Ringle
,
C.M.
and
Sarstedt
,
M.
(
2015
), “
A new criterion for assessing discriminant validity in variance-based structural equation modeling
”,
Journal of the Academy of Marketing Science
, Vol. 
43
No. 
1
, pp. 
115
-
135
, doi: .
Jan
,
I.U.
,
Ji
,
S.
and
Kim
,
C.
(
2023
), “
What (de) motivates customers to use AI-powered conversational agents for shopping? The extended behavioral reasoning perspective
”,
Journal of Retailing and Consumer Services
, Vol. 
75
, 103440, doi: .
Japutra
,
A.
,
Utami
,
A.F.
,
Molinillo
,
S.
and
Ekaputra
,
I.A.
(
2021
), “
Influence of customer application experience and value in use on loyalty toward retailers
”,
Journal of Retailing and Consumer Services
, Vol. 
59
, 102390, doi: .
Kushwaha
,
A.K.
,
Kumar
,
P.
and
Kar
,
A.K.
(
2021
), “
What impacts customer experience for B2B enterprises on using AI-enabled chatbots? Insights from Big data analytics
”,
Industrial Marketing Management
, Vol. 
98
, pp. 
207
-
221
, doi: .
Lee
,
K.-W.
and
Li
,
C.-Y.
(
2023
), “
It is not merely a chat: transforming chatbot affordances into dual identification and loyalty
”,
Journal of Retailing and Consumer Services
, Vol. 
74
, 103447, doi: .
Leung
,
W.K.S.
,
Cheung
,
M.L.
,
Chang
,
M.K.
,
Shi
,
S.
,
Tse
,
S.Y.
and
Yusrini
,
L.
(
2022
), “
The role of virtual reality interactivity in building tourists' memorable experiences and post-adoption intentions in the COVID-19 era
”,
Journal of Hospitality and Tourism Technology
, Vol. 
13
No. 
3
, pp. 
481
-
499
, doi: .
Li
,
C.-Y.
and
Zhang
,
J.-T.
(
2023
), “
Chatbots or me? Consumers' switching between human agents and conversational agents
”,
Journal of Retailing and Consumer Services
, Vol. 
72
, 103264, doi: .
Li
,
C.Y.
,
Fang
,
Y.H.
and
Chiang
,
Y.H.
(
2023
), “
Can AI chatbots help retain customers? An integrative perspective using affordance theory and service-domain logic
”,
Technological Forecasting and Social Change
, Vol. 
197
, 122921, doi: .
Li
,
G.
,
Zhao
,
Z.
,
Li
,
L.
,
Li
,
Y.
,
Zhu
,
M.
and
Jiao
,
Y.
(
2024
), “
The relationship between AI stimuli and customer stickiness, and the roles of social presence and customer traits
”,
Journal of Research in Interactive Marketing
, Vol. 
18
No. 
1
, pp. 
38
-
53
, doi: .
Lin
,
J.C.C.
(
2007
), “
Online stickiness: its antecedents and effect on purchasing intention
”,
Behaviour and Information Technology
, Vol. 
26
No. 
6
, pp. 
507
-
516
, doi: .
Lindell
,
M.K.
and
Whitney
,
D.J.
(
2001
), “
Accounting for common method variance in cross-sectional research designs
”,
Journal of Applied Psychology
, Vol. 
86
No. 
1
, pp. 
114
-
121
, doi: .
Liu
,
Y.
(
2003
), “
Developing a scale to measure the interactivity of websites
”,
Journal of Advertising Research
, Vol. 
43
No. 
2
, pp. 
207
-
216
, doi: .
Loureiro
,
S.M.C.
,
Ashfaq
,
M.
and
Rodrigues
,
M.O.B.
(
2021
), “AI meaning and applications in the consumer sector of retailing, hospitality, and tourism”, in
Handbook of Research on Applied Data Science and Artificial Intelligence in Business and Industry
,
IGI Global
, pp. 
291
-
303
.
Molinillo
,
S.
,
Navarro-García
,
A.
,
Anaya-Sánchez
,
R.
and
Japutra
,
A.
(
2020
), “
The impact of affective and cognitive app experiences on loyalty towards retailers
”,
Journal of Retailing and Consumer Services
, Vol. 
54
, 101948, doi: .
Mostafa
,
R.B.
and
Temerak
,
M.S.
(
2024
), “
Does consumer empowerment enhance brand page stickiness? The role of brand page experience and brand love
”,
Journal of Research in Interactive Marketing
, Vol. 
18
No. 
6
, pp. 
1136
-
1154
, doi: .
Nadeem
,
W.
,
Alimamy
,
S.
and
Ashraf
,
A.R.
(
2024
), “
Unraveling the factors that influence connectedness and relationship performance with augmented reality apps
”,
Information Technology and People
,
Vol. ahead-of-print No. ahead-of-print
. doi: .
Niu
,
B.
and
Mvondo
,
G.F.N.
(
2024
), “
I am ChatGPT, the ultimate AI Chatbot! Investigating the determinants of users' loyalty and ethical usage concerns of ChatGPT
”,
Journal of Retailing and Consumer Services
, Vol. 
76
, 103562, doi: .
Ou
,
C.X.
,
Pavlou
,
P.A.
and
Davison
,
R.M.
(
2014
), “
Swift guanxi in online marketplaces: the role of computer-mediated communication technologies
”,
MIS Quarterly
, Vol. 
38
No. 
1
, pp. 
209
-
230
, doi: .
Perse
,
E.M.
(
1986
), “
Soap opera viewing patterns of college students and cultivation
”,
Journal of Broadcasting and Electronic Media
, Vol. 
30
No. 
2
, pp. 
175
-
193
, doi: .
Pillai
,
R.
,
Ghanghorkar
,
Y.
,
Sivathanu
,
B.
,
Algharabat
,
R.
and
Rana
,
N.P.
(
2023
), “
Adoption of artificial intelligence (AI) based employee experience (EEX) chatbots
”,
Information Technology and People
, Vol. 
37
No. 
1
, pp. 
449
-
478
, doi: .
Ringle
,
C.M.
,
Wende
,
S.
and
Becker
,
J.-M.
(
2022
),
SmartPLS 4
,
SmartPLS GmbH
,
Oststeinbek
.
Roy
,
S.K.
,
Singh
,
G.
,
Hope
,
M.
,
Nguyen
,
B.
and
Harrigan
,
P.
(
2019
), “
The rise of smart consumers: role of smart servicescape and smart consumer experience co-creation
”,
Journal of Marketing Management
, Vol. 
35
Nos
15-16
, pp. 
15
-
16
, doi: .
Rubin
,
A.M.
(
1981
), “
An examination of television viewing motivations
”,
Communication Research
, Vol. 
8
No. 
2
, pp. 
141
-
165
, doi: .
Sarstedt
,
M.
,
Ringle
,
C.M.
and
Hair
,
J.F.
(
2017
),
Partial Least Squares Structural Equation Modeling
,
Research and Markets
,
Dublin
.
Schuetzler
,
R.M.
,
Grimes
,
G.M.
,
Giboney
,
J.S.
and
Rosser
,
H.K.
(
2021
), “
Deciding whether and how to deploy chatbots
”,
MIS Quarterly Executive
, Vol. 
20
No. 
1
, pp. 
1
-
15
, doi: .
Shahzad
,
K.
,
Ashfaq
,
M.
,
Zafar
,
A.U.
and
Basahel
,
S.
(
2024a
), “
Is the future of the metaverse bleak or bright? Role of realism, facilitators, and inhibitors in metaverse adoption
”,
Technological Forecasting and Social Change
, Vol. 
209
, 123768, doi: .
Shahzad
,
M.F.
,
Xu
,
S.
,
An
,
X.
and
Javed
,
I.
(
2024b
), “
Assessing the impact of AI-chatbot service quality on user e-brand loyalty through chatbot user trust, experience and electronic word of mouth
”,
Journal of Retailing and Consumer Services
, Vol. 
79
, 103867, doi: .
Shao
,
Z.
,
Zhang
,
L.
,
Chen
,
K.
and
Zhang
,
C.
(
2020
), “
Examining user satisfaction and stickiness in social networking sites from a technology affordance lens: uncovering the moderating effect of user experience
”,
Industrial Management and Data Systems
, Vol. 
120
No. 
7
, pp. 
1331
-
1360
, doi: .
Shen
,
X.-L.
,
Sun
,
Y.
and
Wang
,
N.
(
2013
), “
Recommendations from friends anytime and anywhere: toward a model of contextual offer and consumption values
”,
Cyberpsychology, Behavior, and Social Networking
, Vol. 
16
No. 
5
, pp. 
349
-
356
, doi: .
Sussman
,
S.W.
and
Siegal
,
W.S.
(
2003
), “
Informational influence in organizations: an integrated approach to knowledge adoption
”,
Information Systems Research
, Vol. 
14
No. 
1
, pp. 
47
-
65
, doi: .
Swani
,
K.
(
2021
), “
To app or not to app: a business-to-business seller's decision
”,
Industrial Marketing Management
, Vol. 
93
, pp. 
389
-
400
, doi: .
Thakur
,
R.
and
AlSaleh
,
D.
(
2018
), “
A comparative study of corporate user-generated media behavior: cross-cultural B2B context
”,
Industrial Marketing Management
, Vol. 
73
, pp. 
125
-
136
, doi: .
Upadhyay
,
N.
and
Kamble
,
A.
(
2023
), “
Why can't we help but love mobile banking chatbots? Perspective of stimulus-organism-response
”,
Journal of Financial Services Marketing
, Vol. 
29
No. 
3
, pp. 
1
-
18
, doi: .
Wagner
,
D.
,
Vollmar
,
G.
and
Wagner
,
H.-T.
(
2014
), “
The impact of information technology on knowledge creation: an affordance approach to social media
”,
Journal of Enterprise Information Management
, Vol. 
27
No. 
1
, pp. 
31
-
44
, doi: .
Wang
,
C.L.
(
2022
), “
Editorial–The misassumptions about contributions
”,
Journal of Research in Interactive Marketing
, Vol. 
16
No. 
1
, pp. 
1
-
2
, doi: .
Wang
,
C.L.
(
2023
), “Interactive marketing is the new normal”, in
The Palgrave Handbook of Interactive Marketing
. doi: .
Wang
,
C.L.
(
2024
), “
Editorial – what is an interactive marketing perspective and what are emerging research areas?
”,
Journal of Research in Interactive Marketing
, Vol. 
18
No. 
2
, pp. 
161
-
165
, doi: .
Wang
,
C.L.
(
2025
), “
Demonstrating contributions through storytelling
”,
Journal of Research in Interactive Marketing
, Vol. 
19
No. 
1
, pp. 
1
-
4
, doi: .
Wei
,
Y.
,
Syahrivar
,
J.
and
Simay
,
A.E.
(
2024
), “
Unveiling the influence of anthropomorphic chatbots on consumer behavioral intentions: evidence from China and Indonesia
”,
Journal of Research in Interactive Marketing
, Vol. 
19
No. 
1
, pp. 
132
-
157
, doi: .
Westaby
,
J.D.
(
2005
), “
Behavioral reasoning theory: identifying new linkages underlying intentions and behavior
”,
Organizational Behavior and Human Decision Processes
, Vol. 
98
No. 
2
, pp. 
97
-
120
, doi: .
Wolf
,
M.
,
Ritz
,
W.
and
McQuitty
,
S.
(
2023
), “
The role of affinity for prosumption experiences and values
”,
Journal of Consumer Behaviour
, Vol. 
22
No. 
1
, pp. 
43
-
55
, doi: .
Xie
,
T.
,
Pentina
,
I.
and
Hancock
,
T.
(
2023
), “
Friend, mentor, lover: does chatbot engagement lead to psychological dependence?
”,
Journal of Service Management
, Vol. 
34
No. 
4
, pp. 
806
-
828
, doi: .
Xu
,
F.
and
Du
,
J.T.
(
2018
), “
Factors influencing users' satisfaction and loyalty to digital libraries in Chinese universities
”,
Computers in Human Behavior
, Vol. 
83
, pp. 
64
-
72
, doi: .
Zhang
,
P.
and
Von Dran
,
G.M.
(
2001
), “
User expectations and rankings of quality factors in different Web site domains
”,
International Journal of Electronic Commerce
, Vol. 
6
No. 
2
, doi: .
Zhou
,
F.
,
Zhang
,
N.
,
Wang
,
N.
and
Mou
,
J.
(
2023
), “
Design affordance in VR and customization intention: is customer inspiration a missing link?
”,
Technological Forecasting and Social Change
, Vol. 
192
, 122594, doi: .
Zhou
,
Y.
,
Zhang
,
Y.
,
Furuoka
,
F.
and
Kumar
,
S.
(
2024
), “
The antecedents and outcomes of electronic customer-to-customer interaction: a PLS-SEM and fsQCA approach
”,
Journal of Research in Interactive Marketing
, Vol. 
18
No. 
5
, pp. 
836
-
864
, doi: .
Zhu
,
H.
and
Zhang
,
Y.
(
2007
), “Navigability design and measurement”, in
Encyclopedia of Portal Technologies and Applications
,
IGI Global
, pp. 
642
-
646
.
Zhu
,
Y.
,
Janssen
,
M.
,
Wang
,
R.
and
Liu
,
Y.
(
2022
), “
It is me, chatbot: working to address the COVID-19 outbreak-related mental health issues in China. User experience, satisfaction, and influencing factors
”,
International Journal of Human-Computer Interaction
, Vol. 
38
No. 
12
, pp. 
1182
-
1194
, 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 http://creativecommons.org/licences/by/4.0/legalcode

or Create an Account

Close Modal
Close Modal