An empirical analysis of user intention to use chatbots for airline tickets consultation

Abstract

Purpose

This study aims to empirically analyze the factors influencing users’ intention to use chatbots for airline ticket consultation. It seeks to introduce a comprehensive framework based on the technology acceptance model (TAM) that integrates key factors alongside traditional TAM constructs to understand what drives behavioral intention to use chatbots in the context of airline ticket consultation.

Design/methodology/approach

The study uses the partial least squares-structural equation modeling (PLS-SEM) approach to validate the proposed model empirically. Data were collected through a survey questionnaire distributed to potential users in Saudi Arabia, with 393 valid responses from a total of 409 received being included in the analysis.

Findings

The empirical analysis confirms the significance of perceived usefulness and user satisfaction as direct determinants of behavioral intention. Additionally, it reveals that factors such as perceived ubiquitous access, perceived completeness, perceived accuracy, perceived unbiased response and perceived convenience have both direct and indirect significant impacts on the behavioral intention to use chatbots for airline ticket consultation.

Originality/value

This research advances theoretical understanding and holds practical implications for designing and implementing effective chatbot services. By investigating the complex interplay of these factors, the study makes substantive contributions to both theoretical advancements and practical applications in the field, particularly in enhancing the user experience and acceptance of chatbots for airline ticket consultations.

Citation

Alotaibi, M. and Hidayat-ur-Rehman, I. (2025), "An empirical analysis of user intention to use chatbots for airline tickets consultation", Journal of Science and Technology Policy Management, Vol. 16 No. 1, pp. 204-228. https://doi.org/10.1108/JSTPM-03-2024-0087

1. Introduction

In the modern landscape characterized by technology-driven interactions, chatbots have emerged as a crucial tool across various industries, enabling efficient and effective customer engagement. Within this context, the aviation sector has notably integrated chatbots to enhance customer service experiences, particularly in the domain of airline ticket consultations. The incorporation of chatbots in this capacity presents unprecedented opportunities to streamline customer interactions, offers real-time assistance and optimizes the ticket booking process. However, the degree to which users are inclined to embrace these chatbot services depends on a complex interplay of psychological, functional and contextual factors. This research undertakes a comprehensive investigation aimed at empirically analyzing the factors that influence users’ intentions to use chatbots for airline ticket consultations.

Despite the promising potential of chatbots, the global chatbot market still faces challenges. For example, in 2022, only 38% of consumers worldwide reported a satisfactory chatbot experience, highlighting a gap in user expectations and actual service delivery (Ukpabi et al., 2019). In Saudi Arabia, the adoption rate of chatbots in the aviation sector remains relatively low compared to global standards, with only 25% of airlines having fully integrated chatbot services (Temsah et al., 2023). This underscores the need for a deeper understanding of the factors influencing user acceptance and satisfaction.

The existing research, including a systematic review by Wongyai et al. (2024), emphasizes the importance of chatbots in the airline industry, focusing on customer adoption, satisfaction and the technology acceptance model (TAM). Our study builds on this foundation to explore chatbots as emerging self-service technologies (SSTs) in new geographical areas. Garcia (2024) indicates that airline chatbots improve pre-flight satisfaction by offering personalized and credible information, but suggests further exploration into their emotional and decision-making capacities. Similarly, SP et al. (2024) apply social presence and flow theories to demonstrate how chatbot characteristics like usability, social presence and flow enhance e-satisfaction and patronage intentions among Indian online travel agency (OTA) users, noting the potential impact of anthropomorphism. Shawal et al. (2023) also use TAM to affirm that factors like chatbot usability, perceived playfulness and usefulness significantly boost user satisfaction in OTAs, while identifying research gaps in these factors within the OTA sector. Furthermore, Rady (2023) finds that EgyptAir chatbots enhance passenger experience by providing timely assistance, yet highlights the need for advancements in processing complex queries and emotions. Sidlauskiene et al. (2023) explore anthropomorphism theory, revealing that anthropomorphic chatbots improve perceived product personalization and increase the willingness to pay among lonely consumers, advocating for more research into the effects of anthropomorphism in various contexts.

The above mentioned research highlights the need to examine chatbot usability, social presence, emotional interaction and anthropomorphism in improving customer satisfaction in aviation and other sectors. Our study seeks to deepen understanding of what drives users to use chatbots for airline ticket consultations, thereby enriching discussions on technology adoption and customer engagement in digital settings.

In the context of chatbots, the TAM provides insights into the factors influencing user intention to use chatbots for airline ticket consultation in Saudi Arabia. Specifically, perceived usefulness and perceived ease of use are likely crucial in determining user intention. TAM’s robustness lies in its ability to identify key determinants of technology acceptance, making it ideal for analyzing user intentions across diverse technological applications. It offers a solid theoretical foundation for understanding how perceived usefulness, and ease of use influence user behavior, which is critical for new technologies like chatbots in the travel and tourism industry. By leveraging TAM, this study assesses the factors driving user satisfaction and adoption of chatbots for airline ticket consultation, providing actionable insights to enhance user experience and facilitate broader acceptance of this technology in Saudi Arabia.

This study uses the TAM framework, widely recognized for analyzing technology acceptance, and introduces an enhanced model that incorporates additional determinants shaping user intentions toward chatbots. These include perceived ubiquitous access, perceived completeness, perceived accuracy, perceived unbiased response, perceived convenience and user satisfaction, augmenting the traditional TAM constructs of perceived usefulness and behavioral intention. Perceived ubiquitous access, which denotes the accessibility of chatbots across multiple platforms, critically affects adoption decisions (Saif et al., 2024). Trust in chatbots is linked to their perceived completeness and accuracy of the information they provide (Chen et al., 2023), while the neutrality of responses is key for trustworthiness (Hutto and Gilbert, 2014). Additionally, the convenience of chatbot usage significantly influences engagement intentions (Kim et al., 2024). User satisfaction, indicative of chatbot interaction quality, plays a vital role in ongoing technology use, as indicated by Hidayat-ur-Rehman et al. (2021). This study aims to deepen understanding of user satisfaction drivers and chatbot adoption for airline ticket consultations, thereby providing insights to enhance user experiences and encourage broader acceptance of chatbots in Saudi Arabia’s travel and tourism sector.

To validate the proposed framework, this research uses partial least squares-structural equation modeling (PLS-SEM). This robust statistical method uncovers the relationships between the factors and their impact on users’ intentions to use chatbots for airline ticket consultations. The survey method is the primary data collection technique, focusing on potential users in Saudi Arabia. This context-specific approach adds depth to the research findings and enhances the practical relevance of the study.

This research advances theoretical understanding and holds practical implications for designing and implementing effective chatbot services. By investigating the complex interplay of these factors, the study contributes to both theoretical advancements and practical applications, particularly in enhancing the user experience and acceptance of chatbots for airline ticket consultations. The study emphasizes two key objectives:

1. To develop a model that identifies the factors influencing consumers' intentions to use chatbots for airline ticket consultations.

2. To empirically analyze and validate the proposed model through comprehensive research methods.

2. Literature review

2.1 Chatbots

Chatbots are computer programs that can simulate human conversation and provide automated responses to user input. They have become increasingly popular in various industries, including the airline industry, where they can assist customers with booking flights, checking flight status and answering frequently asked questions. Several studies have examined the effectiveness of chatbots in various contexts. For instance, Zhang et al. (2019) investigated the use of chatbots in providing customer service in the airline industry and found them to be effective in responding to customer inquiries and assisting with tasks. Similarly, Wang et al. (2018) examined the use of chatbots for booking airline tickets and found them to be convenient and efficient. Shawal et al. (2023) investigated the factors influencing user acceptance, experience and satisfaction with chatbots in OTAs using the TAM. Numerous other studies have highlighted the necessity for additional research on chatbots (Auer et al., 2024; Sindhu and Bharti, 2023).

2.2 User intention to use chatbots

User intention to use chatbots is an important factor in determining their effectiveness. Several studies have examined the factors that influence user intention to use chatbots, such as perceived usefulness, ease of use and trust in the technology. In the airline industry, user intention to use chatbots for various tasks has also been examined. For instance, Chen et al. (2020) investigated the factors that influence user intention to use chatbots for flight booking and found that perceived usefulness and ease of use were the most important factors. Kuo et al. (2019) examined the factors that influence user intention to use chatbots for airline information inquiries and found that perceived usefulness, ease of use and trust in the technology were important factors. A study by SP et al. (2024) examined the effects of chatbot usability cues on e-satisfaction and patronage intention among users of Indian OTAs. They confirmed the indirect effects of chatbot use on purchase intention. A research by Wongyai et al. (2024) systematically reviewed literature on self-service technology (SST) in aviation, and highlighted research themes: customer adoption intention, satisfaction and experience.

2.3 Chatbots in the Saudi context

Chatbots are still a relatively new technology in the Saudi context, and there is limited research on their effectiveness in the country. However, some studies suggest that chatbots can be effective in providing customer service in the Saudi Arabian airline industry.

Alharbi and Aljumah (2019) investigated the use of chatbots in the Saudi Arabian airline industry and found that they can be effective in assisting with various tasks. Their study showed that chatbots can provide customers with quick and accurate responses to their inquiries, which can improve customer satisfaction. The study also found that chatbots can be helpful in reducing the workload of customer service representatives by handling routine inquiries and tasks.

In another study, Alshehri and Alqahtani (2020) examined the potential use of chatbots in the Saudi Arabian health-care system. They found that chatbots can be useful in providing basic health information and reducing the workload of health-care providers. The study also suggested that chatbots can be effective in improving patient engagement and satisfaction.

Recent research underscores the significant role of chatbots in enhancing customer service within the airline industry in Saudi Arabia. Chatbots provide instant, 24 / 7 support, streamline booking processes and offer real-time flight updates, greatly improving the customer experience (Wongyai et al., 2024). This technological shift not only enhances operational efficiency but also meets the expectations of tech-savvy travelers (Wongyai et al., 2024). However, researchers highlight the need for further investigation into the integration of advanced artificial intelligence (AI) features and personalization in chatbots to fully leverage their potential and ensure a seamless user experience (Wongyai et al., 2024).

However, there are some challenges to implementing chatbots in the Saudi context. One challenge is the language barrier, as the official language in Saudi Arabia is Arabic. Chatbots need to be programmed to understand and respond to Arabic language inputs to be effective in the Saudi context. Additionally, there may be cultural differences that need to be considered in the development of chatbots in Saudi Arabia.

2.4 Theoretical framework

This study uses TAM as the foundational theory to explore user acceptance and intention to use chatbots for airline ticket consultation in Saudi Arabia. TAM has been widely applied in the Saudi context to investigate user acceptance and intention to use various technologies. For instance, Al-Gahtani (2016) applied TAM to investigate the factors influencing user intention to use e-government services in Saudi Arabia. The study found that perceived usefulness, ease of use and trust in the technology were important factors influencing user intention. Al-Abdullatif (2023) conducted a study that explored students’ perceptions of chatbots in learning by integrating the TAM with the value-based adoption framework. The research revealed that perceived usefulness and perceived ease of use are key factors influencing students’ acceptance of chatbots. TAM has also been applied in the health-care industry in Saudi Arabia. For instance, a study by Alhur (2023) applied TAM to investigate nurses’ perceptions regarding the effectiveness and advantages of using medical records in Saudi Arabia. The study found that perceived usefulness and perceived ease of use were important factors influencing the intention to use technology. Huang and Chueh (2020) state that TAM assumes external variables like perceived usefulness and ease of use mediate and influence the willingness to use a system. Similar arguments have been given by Mokhtar et al. (2018) also in the context of learning management systems. According to Shawal et al. (2023), examining certain external variables alongside the mediators of the TAM framework is crucial for understanding the adoption of chatbots in online travel services.

In the context of chatbots, TAM can provide insights into the factors that influence user intention to use chatbots for airline ticket consultation in the Saudi context. Specifically, perceived usefulness and perceived ease of use are likely to be important factors in determining user intention to use chatbots for airline ticket consultation. Additionally, trust in the technology may also be an important factor, especially given that chatbots are a relatively new technology in the Saudi context. TAM’s robustness lies in its ability to consistently identify key determinants of technology acceptance, making it ideal for analyzing user intentions across diverse technological applications. It provides a solid theoretical foundation for understanding how perceived usefulness, ease of use and trust influence user behavior, which is critical for new and emerging technologies like chatbots in the travel and tourism industry. By leveraging TAM, this study can comprehensively assess the factors driving user satisfaction and adoption of chatbots for airline ticket consultation, thereby providing actionable insights to enhance user experience and facilitate broader acceptance of this technology in Saudi Arabia.

2.5 Incorporation of additional constructs to technology acceptance model framework

This study incorporates additional constructs like perceived ubiquitous access, perceived completeness, perceived accuracy, perceived unbiased response and perceived convenience to TAM framework. A detailed justification for including these variables and the development of the hypotheses is provided in the following section.

2.6 Hypotheses development and conceptual framework

2.6.1 Perceived usefulness (PU), user’s satisfaction (USAT) and behavioral intention to use chatbots (BI).

PU refers to the user’s perception of how useful the chatbot is in helping them perform various tasks or activities. Studies have consistently found that PU is a significant predictor of user intention to use chatbots, including in the airline industry (Chen et al., 2020; Kuo et al., 2019). The hypothesis H1a proposes that PU significantly influences BI, suggesting that users who perceive the chatbot as useful in helping them perform various tasks are more likely to intend to use it in the future.

USAT refers to the user’s satisfaction with the chatbot’s performance and service quality. Studies have found that satisfaction is a significant predictor of BI (Li et al., 2019). The hypothesis H2 proposes that satisfaction significantly influences BI, suggesting that users who are satisfied with the chatbot’s performance and service quality are more likely to intend to use it in the future.

In addition, the hypothesis H1b proposes that PU significantly influences USAT, suggesting that users who perceive the chatbot as useful are more likely to be satisfied with its performance and service quality.

While H1a, H1b and H2 propose important relationships between PU, USAT and BI, there are some limitations to these constructs. For instance, PU and BI may not fully capture the complexity of user behavior toward chatbots, as users may not always use chatbots for the intended purpose or may use them for different purposes than originally intended. Prior research has established significant impacts of PU on behavioral intention to use/net benefits/actual use (Hidayat Ur Rehman et al., 2023; Ibrahim and Hidayat-ur-Rehman, 2021). SP et al. (2024) have confirmed significant impacts of e-satisfaction on patronage intention while investigating chatbots users’ satisfaction in the context of OTAs. Additionally, while satisfaction is an important construct, it may not always lead to actual usage of chatbots, as users may be satisfied with the chatbot’s performance but may still prefer other modes of communication:

H1a.

PU significantly influences BI.

H1b.

PU significantly influences USAT.

H2.

Satisfaction significantly influences BI.

2.6.2 Perceived ubiquitous access (PUA), behavioral intention to use chatbots (BI) and perceived usefulness (PU).

Perceived ubiquitous access (PUA), behavioral intention to use chatbots (BI) and perceived usefulness (PU) are important constructs in understanding user acceptance and adoption of chatbots. The hypotheses H3a and H3b propose that PUA significantly influences BI and PU, respectively.

PUA refers to the user’s perception of how easily and conveniently they can access the chatbot from different locations and devices. In the context of chatbots for airline ticket consultation, PUA may be important for users who need to access the chatbot from different locations or devices, such as during travel or when on the go. Studies have found that PUA is a significant predictor of user intention to use chatbots, including in the airline industry (Gao et al., 2019). Saif et al. (2024) confirmed that students’ positive attitudes drive their engagement with Chat-GPT through ubiquitous learning, leading to increased actual usage. The ubiquitous access to this innovation plays an important role in influencing users’ intentions to use it.

BI refers to the user’s intention to use the chatbot in the future. Research has consistently shown that perceived usefulness, perceived ease of use and satisfaction are key predictors of BI, including in the airline industry (Chen et al., 2020; Li et al., 2019). Hypothesis H3a proposes that perceived ubiquitous access (PUA) significantly influences BI, suggesting that users who perceive the chatbot as easily accessible from different locations and devices are more likely to intend to use it in the future.

PU refers to the user’s perception of how helpful the chatbot is in assisting them with various tasks or activities. Studies have demonstrated that PU is a significant predictor of user intention to use chatbots, including in the airline industry (Chen et al., 2020; Kuo et al., 2019). Hypothesis H3b proposes that PUA significantly influences PU, indicating that users who perceive the chatbot as easily accessible from different locations and devices are more likely to find it useful for performing various tasks.

While H3a and H3b propose important relationships between PUA, BI and PU, there are some limitations to these constructs. For instance, PUA may not fully capture the complexity of user behavior toward chatbots, as users may prioritize other factors such as security or privacy over access. Additionally, while PU is an important construct, it may not always lead to actual usage of chatbots, as users may perceive the chatbot as useful but may prefer other modes of communication:

H3a.

PUA significantly influences BI.

H3b.

PUA significantly influences PU.

H3c.

PUA significantly influences USAT.

2.6.3 Perceived completeness (PCOM), perceived usefulness (PU) and user’s satisfaction (USAT).

Perceived completeness (PCOM), perceived usefulness (PU) and user’s satisfaction (USAT) are important constructs in understanding user acceptance and adoption of chatbots. The hypotheses H4a and H4b propose that PCOM significantly influences PU and USAT, respectively.

PCOM refers to the user’s perception of the completeness and comprehensiveness of the information provided by the chatbot. In the context of airline ticket consultation, PCOM may be important for users who need accurate and complete information about their flight details, such as departure time, gate number and baggage allowance. The hypothesis H4a proposes that PCOM significantly influences PU, suggesting that users who perceive the information provided by the chatbot as complete and comprehensive are more likely to perceive it as useful in helping them perform various tasks.

The completeness of a chatbot’s response significantly enhances perceived usefulness and user satisfaction by ensuring accurate information retrieval, fostering natural conversation flow and reducing user frustration (Rese and Tränkner, 2024). The completeness of a chatbot’s response enhances perceived usefulness and user satisfaction by ensuring accurate information retrieval and reducing user frustration. Research highlights that a complete and well-structured response positively impacts user satisfaction and continued use intention. Users perceive chatbots as more useful and are more satisfied when they receive comprehensive and accurate answers to their queries (Yu et al., 2024).

USAT refers to the user’s satisfaction with the chatbot’s performance and service quality. The hypothesis H4b proposes that PCOM significantly influences USAT, suggesting that users who perceive the information provided by the chatbot as complete and comprehensive are more likely to be satisfied with its performance and service quality.

While H4a and H4b propose important relationships between PCOM, PU and USAT, there are some limitations to these constructs. For instance, PCOM may not fully capture the complexity of user behavior toward chatbots, as users may prioritize other factors such as ease of use or trust in the technology over the completeness of the information provided. Additionally, while satisfaction is an important construct, it may not always lead to actual usage of chatbots, as users may be satisfied with the chatbot’s performance but may still prefer other modes of communication:

H4a.

PCOM significantly influences BI.

H4b.

PCOM significantly influences PU.

H4c.

PCOM significantly influences USAT.

2.6.4 Perceived accuracy (PA), perceived usefulness (PU) and user’s satisfaction (USAT).

Perceived accuracy (PA), perceived usefulness (PU) and user’s satisfaction (USAT) are important constructs in understanding user acceptance and adoption of chatbots. The hypotheses H5a and H5b propose that PA significantly influences PU and USAT, respectively.

PA refers to the user’s perception of the accuracy and correctness of the information provided by the chatbot. In the context of airline ticket consultation, PA may be important for users who need accurate and reliable information about their flight details, such as flight status, gate changes or delays. The hypothesis H5a proposes that PA significantly influences PU, suggesting that users who perceive the information provided by the chatbot as accurate and correct are more likely to perceive it as useful in helping them perform various tasks.

Rese and Tränkner (2024) have posited that higher correct response rates improve task completion likelihood and boost perceived conversational ability of text-based chatbots. Higher accuracy in chatbot responses leads to enhanced perceived usefulness, greater user satisfaction and increased intention to use. Accurate responses ensure users get relevant information, making the chatbot more reliable and effective. This leads to higher satisfaction and a greater likelihood of continued use (Yu et al., 2024). Several studies have investigated chatbot response accuracy. Verma et al. (2020) evaluated chatbots’ factual response accuracy and completeness. Kuligowska (2015) assessed correct responses using five basic knowledge questions, rated from 1 (very poor) to 5 (very good), for Polish commercial chatbots. Hung et al. (2009) measured task completion success using the Kappa coefficient k, developed by Walker et al. (1997), indicating the percentage of correct chatbot answers. According to Rese and Tränkner (2024), a conversation can be considered successful if the user concludes it upon receiving a satisfactory response.

USAT refers to the user’s satisfaction with the chatbot’s performance and service quality. The hypothesis H5b proposes that PA significantly influences USAT, suggesting that users who perceive the information provided by the chatbot as accurate and correct are more likely to be satisfied with its performance and service quality.

While H5a and H5b propose important relationships between PA, PU and USAT, there are some limitations to these constructs. For instance, PA may not fully capture the complexity of user behavior toward chatbots, as users may prioritize other factors such as ease of use or trust in the technology over the accuracy of the information provided. Additionally, while satisfaction is an important construct, it may not always lead to actual usage of chatbots, as users may be satisfied with the chatbot’s performance but may still prefer other modes of communication:

H5a.

PA significantly influences BI.

H5b.

PA significantly influences PU.

H5c.

PA significantly influences USAT.

2.6.5 Perceived unbiased response (PUR), PU, USAT and BI.

Perceived unbiased response (PUR), perceived usefulness (PU) and user’s satisfaction (USAT) are important constructs in understanding user acceptance and adoption of chatbots. The hypotheses H6a and H6b propose that PUR significantly influences PU and USAT, respectively.

PUR refers to the user’s perception of the chatbot’s response being unbiased and impartial. In the context of airline ticket consultation, PUR may be important for users who need reliable and impartial information about their flight details. The hypothesis H6a proposes that PUR significantly influences PU, suggesting that users who perceive the chatbot’s response as unbiased and impartial are more likely to perceive it as useful in helping them perform various tasks.

USAT refers to the user’s satisfaction with the chatbot’s performance and service quality. The hypothesis H6b proposes that PUR significantly influences USAT, suggesting that users who perceive the chatbot’s response as unbiased and impartial are more likely to be satisfied with its performance and service quality.

Advancements in AI have led to the widespread adoption of chatbots in various industries, including airline customer service. Perceived unbiased response (PUR) by chatbots has been identified as a crucial factor that can influence users’ perceptions and behaviors in the context of airline ticket consultations. Recent studies have argued that suggested that unbiased response generated by chatbots can positively impact user satisfaction and behavioral intention. Moreover, when users perceive chatbots as providing unbiased and impartial responses during airline ticket consultations, they are more likely to find the chatbots useful, express higher levels of satisfaction and demonstrate a greater intention to use the chatbot services in the future (Hidayat-ur-Rehman and Ibrahim, 2023; Hidayat-ur-rehman, 2024; Xue et al., 2024). This is particularly important in the airline industry, where customers often seek objective and reliable information to make informed decisions about their travel plans. Thus, this study believes that the perceived unbiased nature of chatbots can instill a sense of trust in users, leading to enhanced perceptions of usefulness and satisfaction, and ultimately, a stronger intention to engage with the chatbot for future.

While H6a, H6b and H6c propose important relationships between PUR, BI, PU and USAT, there are some limitations to these constructs. For instance, PUR may not fully capture the complexity of user behavior toward chatbots, as users may prioritize other factors such as ease of use or trust in the technology over the unbiased nature of the response. Additionally, while satisfaction is an important construct, it may not always lead to actual usage of chatbots, as users may be satisfied with the chatbot’s performance but may still prefer other modes of communication:

H6a.

PUR significantly influences BI.

H6b.

PUR significantly influences PU.

H6c.

PUR significantly influences USAT.

2.6.6 Perceived convenience (PCON), PU, USAT and BI.

Perceived convenience (PCON), perceived usefulness (PU), satisfaction (USAT) and behavioral intention to use chatbots (BI) are important constructs in understanding user acceptance and adoption of chatbots. The hypotheses H7a, H7b and H7c propose important relationships between PCON, PU, USAT and BI.

PCON refers to the user’s perception of how convenient and easy it is to use the chatbot. In the context of airline ticket consultation, PCON may be important for users who need quick and efficient access to their flight details. The hypothesis H7a proposes that PCON significantly influences PU, suggesting that users who perceive the chatbot as convenient and easy to use are more likely to perceive it as useful in helping them perform various tasks.

USAT refers to the user’s satisfaction with the chatbot’s performance and service quality. The hypothesis H7b proposes that PCON significantly influences USAT, suggesting that users who perceive the chatbot as convenient and easy to use are more likely to be satisfied with its performance and service quality.

BI refers to the user’s intention to use the chatbot in the future. The hypothesis H7c proposes that PCON significantly influences BI, suggesting that users who perceive the chatbot as convenient and easy to use are more likely to intend to use it in the future.

In other context, research has confirmed significant impacts of convenience both PU and BI (Hidayat-ur-Rehman et al., 2022a, 2022b). Prior research suggests that perceived convenience (PCON) significantly impacts perceived usefulness (PU), user satisfaction (SAT) and behavioral intention (BI) in airline ticket consultation. High PCON leads to easier access and faster transactions, enhancing PU. This convenience boosts SAT as users appreciate the efficiency, ultimately strengthening BI to use the service repeatedly (Amaro and Duarte, 2015; Kim et al., 2024).

While H7a, H7b and H7c propose important relationships between PCON, PU, USAT and BI, there are some limitations to these constructs. For instance, PCON may not fully capture the complexity of user behavior toward chatbots, as users may prioritize other factors such as accuracy or reliability of the information provided over convenience. Additionally, while satisfaction and intention to use are important constructs, they may not always lead to actual usage of chatbots, as users may have other preferences for communication channels:

H7a.

PCON significantly influences BI.

H7b.

PCON significantly influences PU.

H7c.

PCON significantly influences USAT.

Proposed model of the study is depicted in Figure 1 below.

3. Research methodology

3.1 Instrument development

The constructs within the proposed model were evaluated by using items primarily derived from existing literature, which were then modified to suit the specific context of this study. Fresh measurement items were formulated to gauge the variable “Perceived unbiased response”. We conducted a pretest to refine the questionnaire. After the pretest feedback, we revised the questionnaire to enhance clarity and improve the validity of the measures. Moreover, these newly developed items underwent evaluation by expert researchers, leading to improvements. An exploratory factor analysis (EFA) yielded favorable outcomes, as indicated by a Kaiser–Meyer–Olkin (KMO) value of 0.929, communalities surpassing 0.5 and cumulative variance reaching 70.034%, all of which affirmed the suitability of the measurement scale.

To evaluate “Perceived ubiquitous access,” four items were adopted from the research conducted by Hidayat-Ur-Rehman et al. (2021). Similarly, three items each were borrowed from the work of Huang and Chueh (2021) to assess “Perceived completeness,” “Perceived accuracy,” “Perceived convenience,” “User satisfaction” and “Behavioral intention to use chatbots”. Additionally, five items were adapted from the study performed by Kasilingam (2020) to measure “Perceived usefulness”. The questionnaire encompassed sections related to demographics and opinions, using Likert scales (ranging from 1 to 5) for participant responses.

3.2 Data collection and sample

The primary objective of this research is to explore the various factors that influence users’ intentions to use chatbots for the purpose of seeking advice on airline ticket matters. The data for this investigation was obtained through the distribution of a survey questionnaire among potential users located in Saudi Arabia. The data collection process involved a combination of both in-person and digital approaches. In the in-person data collection method, researchers personally approached respondents and requested their voluntary participation after providing a detailed explanation of the study’s purpose. Each participant provided informed consent, which facilitated open communication for addressing any queries they might have had.

For the digital data collection process, platforms such as Facebook Messenger, WhatsApp, LinkedIn and email were used. Participants received a survey link along with an attached letter and consent form. These documents clarified the voluntary nature of participation and outlined the measures in place to ensure privacy. Only individuals who agreed and provided consent took part in the study.

The study targeted all users of chatbots for airline ticket consultations. Roscoe (1975) posited that effective sample sizes range from 30 to 500, cautioning against the accuracy of very large samples. Similarly, Stevens (2002) suggested a rule of 15 cases per predictor for multiple regression analysis, which aligns with SEM approaches. Given these precedents, and considering SEM’s similarities to multiple regression, 15 cases per construct was deemed suitable. Thompson (2012) advised a minimum of 384 participants for studies in similar contexts to ensure generalizability. Aligning with these guidelines and the specifics of PLS-SEM, this study aimed for at least 384 responses but successfully gathered data from 409 participants. Following a thorough screening, 16 responses were discarded due to missing information, ensuring the robustness and reliability of the analysis.

By adopting this appropriate sample size, the research assures robust representation and meaningful analysis despite the varying population of chatbot users in Saudi Arabia. Convenience sampling was selected for its practicality and efficiency in accessing chatbot users in Saudi Arabia for timely data collection. The distribution of questionnaires took place through diverse channels, involving students, acquaintances, employees from various organizations and individuals from different walks of life. In the demographics section of the questionnaire, a screening question was incorporated to ask participants if they had ever interacted with chatbots. Throughout the process, the study maintained transparency, minimized biases and ensured participant anonymity.

In terms of response rates, the in-person survey achieved a 47.3% rate, while the online survey yielded a 4.9% response rate. The survey was conducted from May 2023 to July 2023. This meticulous methodological approach underscores the reliability of the data collected, facilitating a comprehensive and rigorous examination of the subject matter. Details about demographics of the respondents are listed below in Table 1.

3.3 Data analysis and results

At the beginning of our model evaluation, we carried out a one-sample Kolmogorov–Smirnov test to assess the distribution of the data, which revealed a nonnormal distribution. Hew et al. (2017) have indicated that PLS-SEM is effective for analyzing data that does not follow a normal distribution. Consequently, this research applied the PLS-SEM approach to evaluate the suggested model. Urbach and Ahlemann (2010) have argued that PLS-SEM is superior to covariance-based SEM for managing intricate models with numerous variables. This technique, widely recognized for theory testing, measures the psychometric properties of the metrics and verifies the existence of hypothesized relationships (Ringle et al., 2015). We used SmartPLS 4 to conduct our data analysis including measurement model structural model analysis.

3.3.1 Measurement model analysis.

The study utilized the PLS algorithm, using its standard settings, to assess the credibility and robustness of the examined constructs. The outcomes, outlined for both “reliability” and “convergent validity,” are comprehensively detailed in Table 2. These findings undeniably indicate that the values for “Cronbach’s alpha,” “composite reliability” and “indicator reliability” all exceed the 0.7 threshold, firmly establishing a high level of reliability within the measurement framework. Additionally, the “average variance extracted (AVE)” values, showcased in the fifth column of Table 2, surpass 0.5, further confirming the convergent validity demonstrated by the measurement indicators.The assessment of discriminant validity was conducted meticulously, using both the Fornell−Larcker criterion and the heterotrait-monotrait ratio (HTMT) criterion. The outcomes derived from the application of the Fornell−Larcker criterion have been thoroughly presented in Table 3. Within this table, the diagonal elements take a prominent role by prominently displaying the square root values of AVE across various constructs. It is crucial to highlight the consistent pattern wherein these values consistently outshine the corresponding correlations observed with other variables. This observation serves as compelling evidence indicating the successful establishment of discriminant validity. Similarly, Table 4 elucidates the HTMT ratios, which consistently remain below the predefined threshold of 0.9. This adherence to the standard set by Henseler et al. (2015) underscores the unmistakable presence of discriminant validity within the framework.

3.3.2 Common method bias.

Within the context of potential common method bias (CMB), a phenomenon arising when data is gathered from a solitary source for both independent and dependent variables − often encountered in survey research − the manifestation of CMB becomes conspicuous when a solitary factor exerts dominance over the variability among these variables (Philip et al., 2003). In the context of our study, the scrutiny of CMB encompassed the utilization of Harman’s single-factor examination, which laid bare that a lone factor elucidated 38.9% of the comprehensive variance. This percentage is notably situated beneath the critical 50% threshold, thereby underscoring the nonexistence of CMB within our milieu. Intensifying the level of scrutiny, a comprehensive evaluation of collinearity unfurled, divulging that the entirety of the measurements for the variance inflation factor (VIF) rested comfortably beneath the threshold of 3.3. This outcome reverberates with the perspective presented by Kock’s (2015), asserting that VIF values lower than 3.3 distinctly indicate the dearth of CMB.

3.3.3 R² (coefficient of determination).

The function of the R² value extends as a precise yardstick, illuminating the degree to which the underlying latent variables (constructs) contained within the framework unravel the intricate variability entrenched within their corresponding observed counterparts (indicators). In the case of the dependent variable SP, the R² value commands attention at an impressive 0.707, casting a luminous beam upon the fact that BI adeptly demystifies a substantial 70.7% of the intricate fluctuations woven into the exogenous constructs. This deduction translates into the compelling inference that the model, with prowess, encompasses a commendable grasp over a striking 70.7% of the intricate tapestry of variations that inherently grace the realm of the users’ behavioral intentions to use chatbots for airlines tickets consultation.

3.3.4 Model-fit indices.

To gauge the full extent to which the measurement model adeptly envelops and encapsulates the intricate reservoir of amassed data, we harnessed well-established benchmarks, encompassing a spectrum of metrics ranging from the standard root mean square residual (SRMR) to the normed fit index (NFI). The ascertained SRMR value, impressively poised at 0.043, elegantly nestles itself beneath the pivotal demarcation points of 0.10 or 0.12, a keen discernment that masterfully conveys the model’s impeccable alignment and concordance with the intricate nuances encapsulated within the expansive data set (Hair et al., 2017). As we widen the aperture of examination, the NFI value, measuring a resolute 0.868, gracefully mirrors its proximity to the cherished threshold of surpassing 0.9. In harmonious symphony, these indices, emblematic of the model’s tenacious fit, interweave into a resounding testament, a definitive attestation that the observed data finds profound resonance and substantively harmonizes with the intricately conceived model, thereby projecting an exceptionally robust and harmonious alignment.

3.3.5 Effect size (f²).

According to Hair et al. (2014), f² values 0.02, 0.15 and 0.35 correspond to small, medium and large effects, respectively, of the exogenous latent construct on the endogenous latent construct. The numbers in Table 5 reveal that the PUA has medium level effect on BI while all others have small effect on their corresponding endogenous latent constructs.

3.3.6 Predictive relevance Q².

Q² (also known as Stone-Geisser’s Q²) quantifies the predictive relevance of the structural model. A model demonstrating predictive relevance can effectively forecast the data points for indicators of an endogenous construct within a reflective model. Q² value greater than 0 suggest that the model possesses predictive relevance for a specific endogenous construct (Götz et al., 2010). Conversely, values below 0 signify a lack of predictive relevance. The Q² scores obtained for all endogenous constructs USAT (0.297), PU (0.302) and BI (0.671) are above the threshold value zero that show the predictive relevance of our model.

3.3.7 Structural model analysis.

Our approach to validating our hypotheses involved using the bootstrapping technique within SmartPLS 4. We applied this method by using 5,000 subsamples along with the default settings. To determine the significance of relationships, we examined beta (β) values along with their corresponding p-values and t-values. For a more detailed breakdown of the bootstrapping results, please consult Table 6. These findings collectively provide strong support for all of our hypotheses, except for hypotheses H6c (PUR → USAT) and H7b (PCON → PU), which did not receive empirical support.

The empirical results confirm the substantial impact of perceived usefulness (PU) on behavioral intention and user satisfaction (USAT). This is evident from the notable values for β, p and t in H1a (β = 0.101, t = 2.683, p = 0.008) and H1b (β = 0.098, t = 1.857, p = 0.064). These outcomes validate H1a and H1b, suggesting that users’ perceptions of the usefulness of chatbots significantly enhance their intention to use and their satisfaction levels. Similarly, the significant and direct influence of USAT on BI, highlighted by the bootstrapping results (β: 0.117, p = 0.002, t = 3.189), provides support for hypothesis H2. These findings are consistent with prior research (Chen et al., 2020; Kuo et al., 2019; Li et al., 2019).

Hypothesis H3 proposed substantial impacts of PUA on BI, PU and USAT, all of which are confirmed by the empirical results: H3a (β = 0.295, t = 6.579, p = 0.000), H3b (β = 0.103, t = 1.975, p = 0.049) and H3c (β = 0.119, t = 2.465, p = 0.014). These outcomes align with previous research findings (Gao et al., 2019) and suggest that PUA significantly affects both the direct and indirect aspects of BI in using chatbots.

Moving on, hypothesis H4 suggested significant effects of perceived completeness (PC) on BI, PU and USAT. The results for H4a (β = 0.167, t = 3.653, p = 0.000), H4b (β = 0.267, t = 4.360, p = 0.000) and H4c (β = 0.235, t = 3.902, p = 0.000) support this hypothesis, indicating a positive and noteworthy relationship between general functionality and service performance. Hypothesis H5 aimed to establish significant direct and indirect effects of PC on BI. The empirical findings underscore the critical role of PC in influencing users’ intention to use chatbots for airline ticket consultations.

Furthermore, hypothesis H5 suggested significant effects of perceived accuracy (PA) on BI, PU and USAT, all of which were corroborated by the bootstrapping results: H5a (β = 0.119, t = 3.079, p = 0.002), H5b (β = 0.130, t = 2.330, p = 0.020) and H5c (β = 0.110, t = 2.012, p = 0.045). This supports the idea that PA plays a noticeable role in influencing users’ intentions.

Moreover, hypothesis H6 proposed significant effects of perceived unbiased response (PUR) on BI (H6a), PU (H6b) and USAT (H6c). The empirical results supported H6a (β = 0.235, t = 4.203, p = 0.000), and H6b (β = 0.191, t = 2.708, p = 0.007). However, hypothesis H6c did not receive empirical support (β = 0.035, t = 0.563, p = 0.574).

Continuing, hypothesis H7 investigated the significant impacts of perceived convenience (PCON) on BI (H7a), PU (H7b) and USAT (H7c). The empirical outcomes confirmed significant impacts of PCON on BI and USAT: H7a (β = 0.139, t = 3.688, p = 0.000), and H7c (β = 0.196, t = 3.771, p = 0.000), while the impact of PCON on USAT in H7b (β = 0.064, t = 1.302, p = 0.194) was deemed insignificant. Consequently, our propositions for H7a and H7c obtained empirical support, whereas H7b did not.

To summarize, the empirical analysis effectively reinforces the proposed framework, highlighting the crucial direct and indirect influences of the mentioned external constructs on determining the internal construct of BI in using chatbots for airline ticket consultations. Ultimately, the model, validated through this study, accounts for 70.7% of the observed variance in service performance within the banking sector (Figure 2).

4. Discussion

This research aimed to empirically analyze the determinants of users’ intention to use chatbots for airline ticket consultation. The study constructed a comprehensive conceptual framework based on existing literature and validated it using the PLS-SEM approach. The results provide valuable insights into the relationships between various factors influencing users’ intentions and behaviors regarding chatbots usage for airline ticket consultations. The empirical findings provide robust support for the majority of the hypotheses, which contributes significantly to understanding users’ perceptions and behaviors related to chatbots utilization.

The research findings confirm a strong positive relationship between perceived usefulness and behavioral intention (H1a). Users who perceive chatbots as useful are more inclined to intend to use them for airline ticket consultations. This aligns with prior studies that emphasize the role of perceived utility in driving users’ intentions to adopt technology (Hidayat Ur Rehman et al., 2023; Ibrahim and Hidayat-ur-Rehman, 2021; Kuo et al., 2019; Li et al., 2019). Additionally, the significant impact of PU on user satisfaction (USAT) reinforces the idea that when users find chatbots useful, they are likely to have higher satisfaction levels.

The empirical results validate the positive influence of user satisfaction on behavioral intention (H2). This suggests that users who are satisfied with chatbots interactions are more likely to continue using them for airline ticket consultations. This finding resonates with previous research by SP et al. (2024), further solidifying the notion that ensuring positive user experiences contributes to sustained intention to use.

The study’s findings support the significance of perceived ubiquitous access (PUA) on behavioral intention (H3a), perceived usefulness (H3b) and user satisfaction (H3c). This implies that users’ perceptions of easy and widespread access to chatbots positively affect their intentions, perception of usefulness and satisfaction levels. These results are consistent with prior studies that highlight the importance of accessibility and convenience in shaping users’ attitudes and intentions (Chen et al., 2020; Li et al., 2019; Kuo et al., 2019; Saif et al., 2024).

The research underscores the positive impact of perceived completeness on behavioral intention (H4a), perceived usefulness (H4b) and user satisfaction (H4c). These findings reinforce the conclusions of Rese and Tränkner (2024) and Yu et al. (2024). This suggests that users who perceive chatbots as offering comprehensive and effective functionality are more likely to intend to use them and be satisfied with their services. The relationship between PC and behavioral intention reinforces the notion that perceived service quality is a key driver of technology adoption (Venkatesh et al., 2003).

The empirical results affirm the significant effects of perceived accuracy on behavioral intention (H5a), perceived usefulness (H5b) and user satisfaction (H5c). This emphasizes the importance of users’ perceptions of chatbots’ accuracy in influencing their intentions and satisfaction levels. Accurate responses contribute to users’ confidence in the technology, ultimately shaping their behaviors and attitudes. The results correspond with the conclusions of Yu et al. (2024), Rese and Tränkner (2024) and Kuligowska (2015).

The findings support the positive impacts of perceived unbiased response on behavioral intention (H6a) and perceived usefulness (H6b). This study’s outcomes resonate with the conclusions of Hidayat-ur-Rehman and Ibrahim (2023), Hidayat-ur-rehman (2024) and Xue et al. (2024). However, the lack of empirical support for its effect on user satisfaction (H6c) suggests that while users might value unbiased responses, this might not significantly impact their overall satisfaction. This could be due to other factors influencing satisfaction that were not captured in the model.

The research findings confirm the influence of perceived convenience on behavioral intention (H7a) and user satisfaction (H7c), highlighting that users who find chatbots interactions convenient are more likely to intend to use them and be satisfied with the service. However, the lack of significant impact on perceived usefulness (H7b) suggests that convenience might not directly translate into perceived utility, implying that ease of use might not always correlate with usefulness. The conclusions are in agreement with those of Kim et al. (2024), Hidayat-ur-Rehman et al. (2022a, 2022b) and Amaro and Duarte (2015).

In conclusion, this study empirically validates the proposed framework’s key relationships and sheds light on the determinants of users’ intention to use chatbots for airline ticket consultation. The findings underscore the significant roles of perceived usefulness, user satisfaction, accessibility, completeness, accuracy, unbiased response and convenience in shaping users’ intentions and perceptions. These insights provide airlines and service providers valuable guidance in designing and implementing chatbots systems that cater to user needs and preferences.

5. Implications of the study

5.1 Theoretical implications

The empirical analysis of the determinants of users’ intention to use chatbots for airline ticket consultation offers substantial theoretical insights into user behavior, perceptions and preferences regarding chatbots utilization in the airline industry. The findings of this study have important implications for academia, contributing to the understanding of technology adoption. This section discusses the theoretical implications of the study’s results.

This study integrates the TAM framework to explore factors influencing chatbot usage for airline ticket consultations in Saudi Arabia. By incorporating additional constructs − perceived ubiquitous access, completeness, accuracy, unbiased response, convenience and user satisfaction − the research provides a robust theoretical foundation for understanding and enhancing technology acceptance and user experience in the travel sector. The findings align with TAM principles, confirming the significance of perceived usefulness, user satisfaction and convenience in influencing users’ behavioral intentions. This validation strengthens the foundation of these theories and extends their applicability to chatbots for airline ticket consultation.

This study introduces new measurement items for “Perceived unbiased response,” enhancing theoretical understanding and providing a novel approach to assessing chatbot neutrality in influencing user trust and acceptance in the travel sector. This contribution broadens the TAM framework by incorporating the critical aspect of perceived neutrality.

The study introduces and validates constructs like perceived ubiquitous access, perceived completeness and perceived unbiased response in the context of chatbots usage. These new constructs expand the theoretical understanding of technology adoption and user perceptions by considering factors that are particularly relevant in the age of AI-driven services.

The study highlights the pivotal role of perceived accuracy in driving users’ intentions and perceptions. This emphasizes the significance of ensuring that chatbots provide accurate and reliable information to users, as inaccuracies can erode users’ trust and confidence in the technology.

The study demonstrates the intricate interplay of factors influencing users’ intentions and perceptions. For instance, the interaction between perceived ubiquitous access and perceived completeness underscores the importance of not only providing widespread access but also ensuring a comprehensive and effective service.

5.2 Practical implications

The empirical analysis of the determinants of users’ intention to use chatbots for airline ticket consultation offers substantial managerial insights into user behavior, perceptions and preferences regarding chatbots utilization in the airline industry. The findings of this study have important implications for industry professionals, guiding practical strategies for service improvement. This section discusses the managerial implications of the study’s results.

The study underscores the importance of perceived usefulness, user satisfaction and convenience in driving users’ intentions to use chatbots. Airlines and service providers can focus on optimizing these factors to enhance the overall user experience and encourage sustained usage. This could involve refining chatbots’ functionalities to ensure they effectively meet user needs and consistently deliver accurate information.

Given the significant impact of perceived accuracy and perceived unbiased response on behavioral intentions and perceived usefulness, organizations should prioritize training chatbots to provide accurate and unbiased information. This not only enhances users’ confidence in the technology but also fosters positive perceptions and intentions.

The findings emphasize the role of accessibility and convenience in shaping users’ attitudes. Service providers should ensure that chatbots are easily accessible across various platforms and devices while focusing on designing user-friendly interfaces that promote convenience and ease of use.

Perceived completeness emerges as a crucial factor in determining user intentions and satisfaction. Airlines should invest in developing chatbots that offer comprehensive functionalities, addressing a wide range of user needs related to airline ticket consultation. This could include features like itinerary planning, fare comparisons and real-time updates.

To manage users’ satisfaction and expectations, organizations should transparently communicate the capabilities and limitations of chatbots. This can help prevent users from forming unrealistic expectations and subsequently becoming dissatisfied with the service.

The study’s findings indicate that user perceptions and intentions are influenced by various factors. Therefore, organizations should adopt a continuous improvement approach, gathering user feedback, analyzing user interactions and iteratively refining the chatbots system to align with evolving user preferences.

Considering the positive impact of accurate and unbiased responses, organizations should invest in training chatbots using high-quality data and machine learning techniques. This ongoing training can lead to improved accuracy and a better understanding of user queries over time.

In conclusion, this study provides valuable insights into the determinants of users’ intention to use chatbots for airline ticket consultation. The theoretical implications extend existing technology adoption theories, while the managerial implications offer actionable strategies for airlines and service providers to enhance their chatbots services. By focusing on factors of the model, organizations can effectively design, implement and refine chatbots to meet user needs and preferences in the airline industry.

6. Limitations and future research avenues

While this research offers valuable insights into the determinants of users’ intention to use chatbots for airline ticket consultation, several limitations and future research directions should be noted. First, the study’s focus on airline ticket consultation limits the generalizability of its findings to other domains. Future research should explore the framework’s applicability in various industries and service contexts. Second, the quantitative approach using PLS-SEM is suitable for identifying relationships between variables but may not capture the depth of user experiences. Complementary qualitative studies could provide richer insights into user attitudes and behaviors. Third, the study considered individual factors in isolation, overlooking the complex interplay of individual, social and contextual factors that influence user intentions. Future research could incorporate social influences, such as social norms and peer recommendations, and contextual factors like technological infrastructure and cultural differences.

The study focused on determinants such as perceived usefulness, satisfaction, accessibility, completeness, accuracy, unbiased response and convenience. However, it did not include factors like privacy concerns, security perceptions and trust in chatbots, which could also impact user intentions. Exploring these factors could lead to a more comprehensive understanding of user decision-making. Additionally, the reliance on self-reported data may introduce common method bias and social desirability effects. Future research could adopt mixed-method approaches, combining self-report data with behavioral observations or physiological measures to enhance validity.

The study did not investigate potential moderating effects of individual differences, such as age, gender and technological expertise, on the proposed relationships. Investigating these moderating influences could provide a nuanced understanding of how determinants vary across different user groups. Finally, future research could explore long-term adoption and usage patterns of chatbots, and the integration of AI-driven personalization within chatbot interactions to enhance user experience.

7. Conclusion

This study provides a thorough empirical analysis of the factors influencing users’ intention to use chatbots for airline ticket consultations. By incorporating key constructs from the TAM alongside novel factors such as perceived ubiquitous access, completeness, accuracy, unbiased response and convenience, the research validates a robust model for understanding user behavior in this phenomenon. The findings indicate that perceived usefulness and user satisfaction are significant determinants of behavioral intention, underscoring the importance of these factors in driving chatbot acceptance for airline ticket consultations. Moreover, the study emphasizes that consumers’ intentions to use chatbots are significantly influenced by their perceptions of ubiquitous availability, completeness, accuracy and unbiased responses. These factors affect not only users’ behavioral intentions but also their perceptions of usefulness and satisfaction, showing that users value chatbots that provide accessible, accurate and unbiased information. However, the research also suggests that while convenience contributes to satisfaction, it is not necessarily a direct translation of perceived usefulness, highlighting the nuanced nature of user perceptions. The results contribute valuable theoretical insights by broadening the TAM framework with additional constructs, reinforcing the importance of perceived utility, satisfaction and accuracy in technology adoption. Practically, the findings offer actionable direction for airlines and service providers aiming to enhance chatbot services. By focusing on improving accessibility, accuracy and user experience, organizations can better meet user needs, promoting greater adoption and satisfaction.

In summary, this study highlights the complex interplay of factors driving user intentions and provides a solid foundation for further research on chatbot utilization in the travel industry. It also offers practical implications for improving chatbot systems, ensuring that they serve effectively to users’ evolving expectations and preferences.

Figures

Figure 1.

Proposed model of the study

Figure 2.

Bootstrapping results

Table 1.

Demographic information of the sample

Item	Characteristics	Percentage (%)
Gender	Male	57.0
	Female	43.0
Age-group	15–25 years	32.6
	26–35 years	27.2
	36–45 years	18.1
	Above 45 years	22.1
Education	High school	24.2
	Higher secondary school	29.5
	Bachelor degree	27.2
	Masters degree	14.0
	PhD	3.3
	Other	1.8
Employment	Full-time	29.3
	Part-time	9.4
	Self-employed	24.2
	Student	31.8
	Retired	5.3

Source: Authors’ calculations based on using R program

Table 2.

Reliability and convergent validity tests summary

Construct	Cronbach’s alpha > 0.7	Composite reliability > 0.7	Items	Indicators’ reliability>=0.7	AVE > 0.5
BI	0.913	0.914	BI1 BI2 BI3	0.922 0.914 0.934	0.853
PA	0.839	0.849	PA1 PA2 PA3	0.825 0.893 0.891	0.757
PC	0.881	0.881	PC1 PC2 PC3	0.882 0.911 0.903	0.808
PCON	0.891	0.892	PCON1 PCON2 PCON3	0.908 0.917 0.893	0.821
PU	0.909	0.911	PU1 PU2 PU3 PU4 PU5	0.850 0.871 0.892 0.789 0.879	0.735
PUA	0.887	0.893	PUA1 PUA2 PUA3 PUA4	0.879 0.808 0.887 0.882	0.747
PUR	0.888	0.893	PUR1 PUR2 PUR3 PUR4	0.882 0.807 0.889 0.880	0.748
USAT	0.857	0.864	USAT1 USAT2 USAT3	0.877 0.911 0.857	0.778

Source: Authors’ calculations based on using R program

Table 3.

Outcomes of Fornell−Larcker criterion

	BI	PA	PC	PCON	PU	PUA	PUR	USAT
BI	0.923
PA	0.546	0.870
PC	0.625	0.465	0.899
PCON	0.542	0.420	0.395	0.906
PU	0.540	0.397	0.495	0.342	0.857
PUA	0.640	0.366	0.398	0.339	0.366	0.865
PUR	0.659	0.410	0.527	0.434	0.460	0.454	0.865
USAT	0.543	0.398	0.478	0.424	0.385	0.371	0.388	0.882

Source: Authors’ calculations based on using R program

Table 4.

Heterotrait-monotrait ratio (HTMT)

	BI	PA	PC	PCON	PU	PUA	PUR	USAT
BI
PA	0.622
PC	0.696	0.539
PCON	0.600	0.485	0.445
PU	0.587	0.448	0.551	0.378
PUA	0.709	0.416	0.448	0.380	0.401
PUR	0.729	0.474	0.595	0.484	0.509	0.511
USAT	0.611	0.468	0.547	0.483	0.433	0.424	0.440

Source: Authors’ calculations based on using R program

Table 5.

Effect size (f²)

	BI	PA	PC	PCON	PU	PUA	PUR	USAT
BI
PA	0.032				0.017			0.016
PC	0.053				0.067			0.054
PCON	0.045				0.004			0.046
PU	0.023							0.012
PUA	0.211				0.012
PUR	0.111				0.033			0.004
USAT	0.031

Source: Authors’ calculations based on using R program

Table 6.

Bootstrapping test results

Hyp.#.	Path.	Beta (β)	Confidence interval 95% bias corrected
			Lower limit	Upper limit	T-values	P-values	Remarks
H1a	PU → BI	0.101	0.031	0.176	2.683	0.008	***
H1b	PU → USAT	0.098	0.000	0.206	1.857	0.064	*
H2	USAT → BI	0.117	0.046	0.187	3.189	0.002	***
H3a	PUA → BI	0.295	0.211	0.392	6.579	0.000	***
H3b	PUA → PU	0.103	0.007	0.198	1.975	0.049	**
H3c	PUA → USAT	0.119	0.023	0.213	2.465	0.014	**
H4a	PC → BI	0.167	0.075	0.258	3.653	0.000	***
H4b	PC → PU	0.267	0.120	0.368	4.360	0.000	***
H4c	PC → USAT	0.235	0.113	0.354	3.902	0.000	***
H5a	PA → BI	0.119	0.043	0.195	3.079	0.002	***
H5b	PA → PU	0.130	0.019	0.235	2.330	0.020	**
H5c	PA → USAT	0.110	−0.003	0.211	2.012	0.045	**
H6a	PUR → BI	0.235	0.127	0.352	4.203	0.000	***
H6b	PUR → PU	0.191	0.059	0.324	2.708	0.007	***
H6c	PUR → USAT	0.035	−0.080	0.166	0.563	0.574	N.S.
H7a	PCON → BI	0.139	0.062	0.212	3.688	0.000	***
H7b	PCON → PU	0.064	−0.025	0.151	1.302	0.194	N.S.
H7c	PCON → USAT	0.196	0.087	0.299	3.771	0.000	***

Notes:

*p < 0.1;

**p < 0.05;

***p < 0.01; N.S. = not significant

Source: Authors’ calculations based on using R program