Online Hotel Booking Intention: Integrating fsQCA and PLS-SEM to Explain the Roles of Electronic Word of Mouth, Customer Experience, and Trust

Abstract

The continuous expansion of digital tourism has increased the need to understand tourists’ intentions to book hotels online in competitive digital marketplaces. Previous studies have generally relied on symmetric analytical approaches, which have limited the understanding of the complex causal mechanisms underlying online hotel booking intention, especially in emerging tourism markets such as Thailand. This study aimed to investigate the determinants of online hotel booking intention (OBI) among Thai tourists by integrating partial least squares structural equation modeling (PLS-SEM) and fuzzy-set Qualitative Comparative Analysis (fsQCA). The study collected data from 1884 Thai tourists who had prior experience with online hotel booking. Furthermore, PLS-SEM was used to examine the linear relationships among the study constructs, whereas fsQCA was employed to identify multiple causal configurations leading to OBI. The findings indicate that electronic word of mouth (E-WOM) was the strongest direct driver of OBI, while customer experience (CEX) significantly enhanced perceived value (PCV). Trust (TRU) emerged as a key relational mechanism that strengthened perceived usefulness (PU) and facilitated the development of E-WOM, while electronic promotion (EPM) indirectly influenced OBI. In addition, fsQCA identified multiple sufficient configurational pathways leading to high OBI, supporting the principle of equifinality. Collectively, these findings advance the theoretical understanding of OBI by demonstrating that the integration of symmetric and asymmetric analytical approaches provides complementary insights into tourists’ decision-making processes.

1. Introduction

Rapid technological advancement has fundamentally transformed the global tourism industry, reshaping how tourists search for information, evaluate accommodation alternatives, and make hotel booking decisions. The widespread adoption of the Internet, smartphones, and digital platforms has accelerated the transition from traditional offline reservation systems to online hotel booking, making online booking channels an integral part of contemporary tourism consumption (Amin et al., 2021; Salameh et al., 2022). At the same time, the rapid expansion of online travel agencies (OTAs), such as Booking.com, Agoda, and Expedia, has intensified competition in digital tourism by providing travelers with real-time access to price comparisons, customer reviews, and personalized recommendations. Consequently, online hotel booking has evolved into a more complex decision-making process in which travelers evaluate large volumes of digital information under conditions of uncertainty and perceived risk. Understanding the factors that shape online hotel booking intention has therefore become increasingly important for both tourism researchers and industry practitioners seeking to understand consumer behavior in highly competitive digital environments (Saha & Biswas, 2024). Accordingly, understanding online hotel booking intention requires consideration of multiple technological, experiential, and relational factors rather than relying on a single perspective. Because consumers cannot physically evaluate accommodation before purchase, their booking decisions are increasingly influenced by technological, experiential, and social factors embedded within digital platforms. In contemporary online tourism markets, tourists increasingly rely on electronic word of mouth (E-WOM). Online and other forms of consumer-generated content play an important role in shaping tourists’ perceptions of hotel brands and service quality (Jeong et al., 2022). Consequently, electronic word of mouth (E-WOM) has become an influential source of social information that affects tourists’ online hotel booking decisions (Saha & Biswas, 2024; Salameh et al., 2022).

Online hotel booking behavior has become increasingly complex and can no longer be adequately explained by a single factor or a single theoretical perspective. Rather, booking intention emerges from the interaction of technological, experiential, relational, and marketing-related dimensions operating simultaneously within digital tourism environments (Bilan et al., 2024). Consequently, three important gaps remain in the existing literature. Previous studies have primarily relied on symmetric analytical approaches, such as regression analysis and structural equation modeling (SEM), to examine the net effects of individual predictors on online hotel booking intention (Terrah et al., 2024). Although these approaches provide valuable insights into direct relationships, they generally assume linear and independent effects among predictors and therefore provide only a limited understanding of causal complexity and equifinality in online booking behavior. Furthermore, existing studies have typically examined technological, experiential, relational, and social influence factors in isolation rather than integrating them within a comprehensive theoretical framework (Amin et al., 2021; Salameh et al., 2022; Saha & Biswas, 2024). As a result, current evidence remains limited in explaining how technological, experiential, relational, and marketing-related factors interact to shape online hotel booking intention. In addition, empirical evidence from emerging tourism markets remains relatively scarce, particularly in Thailand, where rapid digital transformation and the increasing adoption of online booking platforms may influence consumer behavior differently from that observed in developed economies (Sharafuddin et al., 2024).

This study aims to explain online hotel booking intention among Thai tourists by integrating partial least squares structural equation modeling (PLS-SEM) and fuzzy-set Qualitative Comparative Analysis (fsQCA) within a unified theoretical framework combining the Technology Acceptance Model (TAM), the Stimulus–Organism–Response (S-O-R) framework, and Signaling Theory. Specifically, the study examines how technological, experiential, relational, and marketing-related factors jointly shape online hotel booking intention through both symmetric and configurational causal relationships.

This study makes three primary contributions to the literature. First, this study extends the literature on online hotel booking behavior by integrating TAM, the S-O-R framework, and Signaling Theory into a unified explanatory framework. Rather than treating technological, experiential, relational, and marketing-related factors independently, the proposed framework explains how these complementary perspectives jointly contribute to online hotel booking intention.

Second, this study combines PLS-SEM and fsQCA to capture both net effects and configurational causality. While PLS-SEM identifies the relative importance of individual predictors, fsQCA reveals multiple causal pathways leading to the same behavioral outcome, thereby providing a more comprehensive understanding of online hotel booking intention.

Third, the findings provide practical insights for hotels, online travel agencies (OTAs), and digital marketers by demonstrating how technological functionality, customer experience, trust, electronic word of mouth, and digital marketing activities should be considered and managed as complementary rather than independent drivers of online hotel booking intention.

2. Literature Review and Hypotheses Development

2.1. Online Hotel Booking Intention (OBI)

OBI reflects consumers’ willingness to reserve accommodation via digital platforms and is a key outcome in electronic commerce and digital tourism (Chan et al., 2025). Previous studies have demonstrated that online hotel booking intention is influence by a wide range of technological, promotional, and Experiential factors including website quality, online reviews, perceived trust, promotional offers, and website convenience (Saha & Biswas, 2024; Amin et al., 2021). However, previous studies have generally examined these drivers independently, limiting understanding of how they interact to shape online hotel booking intention (Amin et al., 2021; Salameh et al., 2022). Accordingly, this study conceptualizes OBI as the outcome of interacting technological, experiential, relational, and market-signaling mechanisms rather than isolated determinants. This perspective provides the conceptual foundation for integrating TAM, the S-O-R framework, and Signaling Theory within the proposed research framework.

2.2. Technology Acceptance Model (TAM): PEOU and PU

The TAM posits that perceived ease of use (PEOU) enhances PU, which in turn drives behavioral intention (Davis, 1989). In online booking, easy-to-use platforms improve task efficiency perceptions and increase adoption likelihood (Davis, 1989; Nuryyev et al., 2021). Trust can further enhance PU by increasing consumers’ confidence in the effectiveness and reliability of online booking platforms (Azhar et al., 2025).

Although TAM provides the technological foundation for explaining OBI, it does not explicitly explain the psychological and market-signaling mechanisms underlying online hotel booking decisions. Therefore, TAM is integrated with the S-O-R framework and Signaling Theory in the proposed theoretical framework.

H1. 

PEOU positively influences PU.

H2. 

TRU positively influences PU.

H9. 

PU positively influences OBI.

2.3. Customer Experience (CEX) and Perceived Value (PCV)

CEX encompasses usability, service quality, and affective responses throughout the booking journey. Positive customer experiences enhance PCV, reflecting consumers’ evaluation of the trade-off between benefits and costs, thereby increasing intention (Hoo et al., 2024). CEX may also directly influence OBI by improving satisfaction and engagement (Amer, 2021). From the S-O-R perspective, CEX and PCV represent experiential mechanisms that complement the technological perspective of TAM within the proposed integrated theoretical framework.

H3. 

CEX positively influences PCV.

H8. 

PCV positively influences OBI.

H11. 

CEX positively influences OBI.

2.4. Electronic Word of Mouth (E-WOM) and Trust (TRU)

E-WOM, including online reviews, ratings, and peer recommendations, reduces uncertainty and exerts strong social influence in high-involvement services such as hotel booking (Garg & Kumar, 2021). Positive E-WOM enhances TRU, which in turn facilitates information acceptance and decision-making (Onobrakpeya & Otutuadum, 2024). E-WOM also directly increases booking intention due to its credibility (Saha & Biswas, 2024; Salameh et al., 2022). Within the proposed integrated theoretical framework, E-WOM and TRU represent relational mechanisms that complement the technological and experiential perspectives in explaining OBI.

H4. 

E-WOM positively influences TRU.

H10. 

E-WOM positively influences OBI.

2.5. Electronic Promotion (EPM) and Content Marketing (CMK)

EPM includes online advertising, discounts, and campaigns that signal value and quality (Velázquez et al., 2023). Rather than directly influencing OBI, EPM primarily operates indirectly by strengthening TRU and stimulating positive E-WOM (Anastasiei et al., 2024). Content marketing (CMK) refers to informative and engaging content created by a firm, which can directly motivate OBI by increasing awareness and purchase intention (Camilleri & Kozak, 2022). Within the proposed integrated theoretical framework, EPM and CMK represent market-signaling mechanisms that complement the technological, experiential, and relational perspectives in explaining OBI.

H5. 

EPM has a positive influence on E-WOM.

H6. 

EPM has a positive influence on TRU.

H7. 

CMK has a positive influence on OBI.

2.6. Theoretical Integration: TAM, S-O-R, and Signaling Theory

This study integrates three complementary theoretical perspectives to explain online hotel booking intention.

(1)

The Technology Acceptance Model (TAM) explains how technological perceptions, particularly perceived ease of use (PEOU) and perceived usefulness (PU), influence technology acceptance and behavioral intention (Davis, 1989; Li et al., 2024).

(2)

The S-O-R framework explains that external stimuli, such as EPM, E-WOM, and CMK, influence internal psychological states, including trust (TRU), perceived customer value (PCV), and perceived usefulness (PU), that ultimately shape online hotel booking intention (OBI) (Yuen et al., 2022).

(3)

Signaling Theory explains how market signals, such as electronic promotion (EPM), content marketing (CMK), and electronic word of mouth (E-WOM), reduce information asymmetry and strengthen consumer confidence (Palamidovska-Sterjadovska et al., 2024; Yuen et al., 2022).

Together, these three theoretical perspectives provide a comprehensive explanation of online hotel booking intention by integrating technological, psychological, and market-signaling mechanisms within a unified conceptual framework. Rather than offering three parallel theoretical explanations, the integrated framework demonstrates how these complementary perspectives collectively explain different dimensions of tourists’ online hotel booking decisions.

The integrated relationships among these complementary theoretical perspectives are illustrated in Figure 1.

2.7. Methodological Integration: PLS-SEM and fsQCA

PLS-SEM explains the net effects of individual predictors on online hotel booking intention. However, consumer decision-making in digital tourism often results from multiple interacting conditions rather than independent linear effects (Chen & Ye, 2023). Accordingly, fsQCA complements PLS-SEM by identifying multiple causal configurations leading to the same behavioral outcome, thereby recognising causal complexity and equifinality (Kumar et al., 2022). Therefore, integrating PLS-SEM and fsQCA enables the simultaneous examination of both net effects and configurational causality, thereby providing complementary insights into online hotel booking intention that cannot be achieved using either analytical approach alone.

Conceptual Framework

Drawing upon the integrated theoretical framework presented in Figure 1, this study proposes a conceptual framework explaining how technological, psychological, and market-signaling mechanisms jointly shape online hotel booking intention. Consistent with TAM, perceived ease of use enhances perceived usefulness, which subsequently influences online hotel booking intention (Davis, 1989; Li et al., 2024). Within the S-O-R framework, external stimuli influence internal psychological states that ultimately shape behavioral intentions (Yuen et al., 2022), whereas Signaling Theory explains how digital signals reduce information asymmetry and strengthen consumer confidence (Palamidovska-Sterjadovska et al., 2024). Based on this integrated theoretical perspective, the proposed conceptual framework specifies the hypothesized relationships among the study constructs. Specifically, perceived ease of use is proposed to enhance perceived usefulness, while customer experience strengthens perceived customer value and directly influences online hotel booking intention (Amoako et al., 2021). Electronic word of mouth contributes to trust and directly influences online hotel booking intention (Saha & Biswas, 2024; Salameh et al., 2022), whereas electronic promotion indirectly influences booking intention through trust and electronic word of mouth (Anastasiei et al., 2024). In addition, content marketing is proposed to directly enhance online hotel booking intention by increasing consumer awareness and engagement (Camilleri & Kozak, 2022).

Overall, the proposed conceptual framework suggests that online hotel booking intention is jointly influenced by technological acceptance, psychological evaluations, and market-signaling mechanisms. These hypothesized relationships are subsequently examined using PLS-SEM to estimate net effects and fsQCA to identify multiple configurational pathways leading to online hotel booking intention.

The proposed conceptual framework derived from the integrated theoretical framework is presented in Figure 2. The model illustrates the hypothesized relationships (H1–H11) among the study constructs that are subsequently examined using PLS-SEM and fsQCA.

3. Materials and Methods

3.1. Research Design

This study employed a quantitative research design to examine the determinants of OBI among Thai tourists. A cross-sectional survey approach was adopted to collect empirical data from respondents at a single point in time. A quantitative approach was appropriate because the study sought to test the hypothesized relationships among multiple latent constructs and evaluate the proposed theoretical framework empirically. Furthermore, PLS-SEM and fsQCA were integrated to capture both net effects and configurational causality, providing complementary insights into the determinants of OBI (Rasoolimanesh et al., 2021).

3.2. Population and Sample

The target population comprised Thai tourists with prior experience using online hotel booking platforms. Eligible participants were required to have previously searched for, compared, or booked hotel accommodations through digital platforms, such as online travel agencies (OTAs) or hotel booking applications, within the previous 12 months. Purposive sampling was initially applied to ensure that respondents met these eligibility criteria. Following participant identification, convenience sampling was used to recruit eligible respondents for the online survey.

A total of 2030 questionnaires were returned. After applying the eligibility criteria and screening procedures, 146 questionnaires were excluded, resulting in 1884 valid responses for the final analyses. The final sample size was considered adequate for estimating the proposed PLS-SEM model and conducting robust configurational analyses using fsQCA (Sabol et al., 2023; Knight et al., 2021; Nguyen et al., 2023).

3.3. Measurement Development

The questionnaire was developed based on the proposed conceptual framework, relevant literature, and findings from a Fuzzy Delphi process involving 21 experts on online hotel booking intention among Thai tourists. Measurement items were adapted from previously validated scales reported in the tourism, hospitality, and information systems literature and were subsequently refined to align with the study objectives and conceptual framework. The draft questionnaire was reviewed by the research supervisors to improve its accuracy, clarity, and consistency with the study objectives. Prior to the main survey, the questionnaire was pilot-tested with 30 respondents who met the study eligibility criteria. Based on their feedback, minor wording revisions were made to improve clarity and comprehensibility without changing the underlying meaning of the measurement items. Internal consistency reliability was assessed using Cronbach’s alpha, yielding an overall coefficient of 0.95, indicating excellent reliability. The final questionnaire measured nine constructs: perceived ease of use (PEOU), perceived usefulness (PU), customer experience (CEX), perceived customer value (PCV), trust (TRU), electronic word of mouth (E-WOM), electronic promotion (EPM), content marketing (CMK), and online hotel booking intention (OBI). All constructs were measured using a seven-point Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree).

3.4. Data Collection

Data were collected between November 2025 and January 2026 using an online questionnaire administered through Google Forms (Google LLC, Mountain View, CA, USA). The survey link was distributed through Facebook (Meta Platforms, Inc., Menlo Park, CA, USA) travel communities, LINE groups (LY Corporation, Tokyo, Japan), and online travel agency (OTA)-related communities. Screening questions were included at the beginning of the questionnaire to verify participant eligibility before survey completion.

A total of 2030 questionnaires were returned. After screening the returned questionnaires, 146 questionnaires were excluded because of incomplete responses, failure to meet the eligibility criteria, or duplicate submissions, resulting in 1884 valid questionnaires for the final analyses.

To improve data quality, all completed questionnaires were screened for duplicate submissions and inconsistent response patterns before analysis. Ethical approval was obtained from the Ethics Committee of Rangsit University and received approval from the Institutional Review Board (IRB) on 7 July 2025 (Protocol No. RSUERB2025156). All participants provided informed consent prior to participation. Participation was voluntary, and all responses were collected anonymously and treated confidentially. The complete questionnaire used for data collection is provided in Appendix A.

3.5. Assessment of Common Method Bias

Because this study employed a cross-sectional survey using single-source self-reported data collected from a single source, the potential influence of common method bias (CMB) was assessed. Following the recommendations of Podsakoff et al. (2003), both procedural and statistical remedies were implemented. Procedurally, respondents were assured of anonymity and confidentiality, and measurement items were adapted from previously validated scales. Statistically, common method bias was assessed using Harman’s single-factor test and indicator variance inflation factor (VIF) values. The results of these assessments are presented in Section 4.

Descriptive statistical analyses and respondent profiling were performed using IBM SPSS Statistics version 21 (IBM Corp., Armonk, NY, USA). Measurement model assessment and structural model analysis were subsequently conducted using SmartPLS version 4.1.1.6 (SmartPLS GmbH, Oststeinbek, Germany), while configurational analysis was performed using fsQCA software version 4.1 (University of California, Irvine, CA, USA).

3.6. PLS-SEM Analysis

PLS-SEM analysis was employed to examine the hypothesized relationships among the study constructs. The analysis was conducted using SmartPLS version 4.1.1.6 (SmartPLS GmbH, Oststeinbek, Germany). PLS-SEM was considered appropriate because the study involved multiple latent constructs, predictive relationships, and a complex research framework integrating technological, experiential, relational, and marketing factors (Sabol et al., 2023).

The analysis was conducted in two stages. First, the measurement model was evaluated to assess reliability and validity using indicator loadings, Cronbach’s alpha, composite reliability (CR), average variance extracted (AVE), and discriminant validity (Hair et al., 2024; Henseler et al., 2016). Second, the structural model was assessed by examining path coefficients, coefficient of determination (R²), model fit indices, and the significance of hypothesized relationships. This approach enabled the study to evaluate both the measurement quality and the predictive capability of the proposed research model.

(1)

Measurement Model Assessment

The measurement model assessment was conducted to evaluate the reliability and validity of the latent variables included in the proposed framework. Indicator reliability was first examined using factor loadings, where values above the recommended criterion indicated satisfactory item reliability (Hair et al., 2024). Internal consistency reliability was assessed using Cronbach’s alpha and CR, with values exceeding the recommended criterion of 0.70 considered acceptable.

Convergent validity was evaluated using the AVE (Hair et al., 2024). AVE values greater than 0.50 indicated that the variables explained more than half of the variance of their respective indicators. Discriminant validity was assessed using the Fornell–Larcker criterion by comparing the square root of the AVE for each construct with its correlations with other variables (Fornell & Larcker, 1981). Discriminant validity was considered satisfactory when the square root of the AVE exceeded the inter-construct correlations.

(2)

Structural Model Assessment

The structural model assessment was performed to examine the hypothesized relationships among the variables within the proposed framework. Path coefficients were analyzed to evaluate the strength and direction of the relationships between variables (Hair et al., 2024). The significance of the structural relationships was assessed using bootstrapping procedures. The predictive capability of the model was evaluated using the coefficient of determination (R²), which indicates the proportion of variance explained by the predictor variables. Higher R² values indicate stronger explanatory power of the model (Hair et al., 2024).

In addition, model fit was assessed using standardized root mean square residual (SRMR) and normed fit index (NFI). An SRMR value below the recommended criterion indicated acceptable model fit, while NFI values were used to evaluate the overall fit of the proposed structural model (Hair et al., 2024).

3.7. fsQCA Methodology

To complement the symmetric analysis provided by PLS-SEM, this study employed fsQCA to examine asymmetric and configurational relationships associated with OBI. fsQCA is particularly suitable for investigating complex causal relationships because consumer decision-making in digital tourism environments often results from multiple interacting conditions rather than single independent effects (Fang et al., 2024). Unlike conventional symmetric approaches, fsQCA recognizes causal complexity, equifinality, and asymmetric relationships, allowing different combinations of conditions to produce the same outcome. The fsQCA analysis was conducted using fsQCA software version 4.1 (University of California, Irvine, CA, USA) and followed three major stages (Manosuthi et al., 2022; Slager et al., 2023): calibration, truth table construction, and sufficiency analysis.

(1) Calibration: all variables were calibrated into fuzzy-set membership scores ranging from 0 to 1 using the direct calibration method proposed by Charles Ragin (Ragin, 2008; Rihoux & Ragin, 2009). In the direct calibration procedure, the values of 0.95, 0.50, and 0.05 were used as the three qualitative anchors representing full membership, crossover point, and full non-membership, respectively. Following prior fsQCA studies, the calibration thresholds were determined using percentile values derived from the dataset. Specifically, the 95th percentile represented full membership, the 50th percentile represented the crossover point, and the 5th percentile represented full non-membership. This percentile-based calibration approach reflects the empirical distribution of the data and provides context-sensitive membership classification for large-sample survey research.

The percentile values were calculated using IBM SPSS Statistics version 21 (IBM Corp., Armonk, NY, USA) prior to calibration in fsQCA software version 4.1. This procedure enabled the transformation of latent construct scores into fuzzy-set membership values suitable for configurational analysis (Hossain et al., 2022).

(2) Truth table construction: a truth table was constructed to identify combinations of causal conditions associated with high OBI. The truth table analysis employed a frequency threshold of 5 and a consistency threshold of 0.88. The frequency threshold was used to eliminate configurations with insufficient empirical representation, while the consistency threshold ensured that only highly consistent causal configurations were retained for further analysis (Liozu et al., 2021; Villani et al., 2023).

(3) Sufficiency analysis: the sufficiency analysis was conducted to identify multiple causal configurations leading to high OBI. Unlike symmetric approaches that examine the independent net effects of variables, fsQCA focuses on the combined influence of interacting conditions. The intermediate solution was selected because it provides theoretically meaningful yet parsimonious causal interpretations while incorporating directional expectations derived from the conceptual framework (Islam et al., 2024).

The analysis evaluated raw coverage, unique coverage, and solution consistency for each causal configuration. Raw coverage indicates the proportion of the outcome explained by a configuration, whereas unique coverage reflects the distinct explanatory contribution of a specific pathway (Ragin, 2008; Rihoux & Ragin, 2009). Solution consistency assesses the degree to which the identified configurations consistently lead to high OBI (Terrah et al., 2024).

4. Results

4.1. Respondent Profile

Table 1. Demographic Profile of Respondents.

Demographic Characteristics	Category	Frequency	Percentage (%)
Gender	Male	751	39.90
	Female	1131	60.00
	Other/Prefer not to say	2	0.10
Age	18–29	1283	68.10
	30–41	485	25.70
	42–53	110	5.80
	54–65	6	0.30

presents the demographic characteristics of the respondents included in the study. A total of 1884 valid responses from Thai tourists with prior online hotel booking experience were analyzed. The respondent profile summarizes the gender and age distributions of the respondents.

Table 1 summarizes the demographic characteristics of the respondents. The sample consisted predominantly of female participants and young adults aged 18–29 years. These characteristics are broadly consistent with the demographic profile of users of online hotel booking platforms in Thailand.

To assess the potential influence of common method bias, Harman’s single-factor test and indicator variance inflation factor (VIF) values were examined. Harman’s single-factor test indicated that the first unrotated factor explained 35.395% of the total variance, which is below the recommended threshold of 50%. Furthermore, all indicator VIF values ranged from 1.326 to 2.834, remaining below the recommended threshold. These findings indicate that common method bias is unlikely to threaten the validity of the measurement model or the subsequent structural analyses.

4.2. Measurement Model Results

The measurement model was assessed to evaluate the reliability and convergent validity of the constructs included in the proposed framework. Reliability was examined using factor loadings, Cronbach’s alpha, and composite reliability (CR), while convergent validity was assessed using the average variance extracted (AVE). As presented in

Table 2. Reliability and Convergent Validity Assessment.

Variables	Item	Factor Loading	Cronbach’s Alpha	Composite Reliability (CR)	Average Variance Extracted (AVE)
Content Marketing (CMK)	CMK1	0.655	0.804	0.863	0.559
	CMK2	0.799
	CMK3	0.711
	CMK4	0.747
	CMK5	0.816
Perceived Ease of Use (PEOU)	PEOU1	0.832	0.875	0.909	0.666
	PEOU2	0.820
	PEOU3	0.830
	PEOU4	0.782
	PEOU5	0.816
Customer Experience (CEX)	CEX1	0.816	0.876	0.909	0.668
	CEX2	0.819
	CEX3	0.830
	CEX4	0.818
	CEX5	0.802
Electronic Promotion (EPM)	EPM1	0.779	0.859	0.898	0.639
	EPM2	0.790
	EPM3	0.823
	EPM4	0.832
	EPM5	0.772
Perceived Value (PCV)	PCV1	0.839	0.878	0.911	0.673
	PCV2	0.837
	PCV3	0.844
	PCV4	0.775
	PCV5	0.804
Perceived Usefulness (PU)	PU1	0.777	0.865	0.903	0.650
	PU2	0.815
	PU3	0.846
	PU4	0.787
	PU5	0.805
Trust (TRU)	TRU1	0.829	0.883	0.915	0.682
	TRU2	0.841
	TRU3	0.845
	TRU4	0.794
	TRU5	0.819
Electronic Word of Mouth (E-WOM)	E-WOM1	0.767	0.871	0.906	0.659
	E-WOM2	0.810
	E-WOM3	0.842
	E-WOM4	0.821
	E-WOM5	0.818
Online Booking Intention (OBI)	OBI1	0.801	0.887	0.917	0.689
	OBI2	0.816
	OBI3	0.851
	OBI4	0.862
	OBI5	0.819

, all factor loadings exceeded the recommended threshold, indicating satisfactory indicator reliability. In addition, the Cronbach’s alpha and CR values for all constructs were above the recommended threshold of 0.70, confirming satisfactory internal consistency reliability. Convergent validity was also supported, as all AVE values exceeded the recommended threshold of 0.50, indicating that the constructs explained a substantial proportion of variance in their indicators. Overall, the results demonstrate that the measurement model possesses satisfactory reliability and convergent validity.

Table 2 presents the measurement model assessment results, including factor loadings, Cronbach’s alpha, CR, and AVE. The results indicate that all factor loadings exceeded the recommended threshold of 0.60, ranging from 0.655 to 0.862, confirming satisfactory indicator reliability. Although the loading of CMK1 (0.655) was slightly lower than the remaining indicators, it remained within the acceptable range for exploratory and consumer behavior research. Furthermore, all Cronbach’s alpha values ranged from 0.804 to 0.887, while CR values ranged from 0.863 to 0.917, exceeding the recommended threshold of 0.70. These findings confirm satisfactory internal consistency reliability across all constructs. In addition, the AVE values ranged from 0.559 to 0.689, all above the recommended threshold of 0.50, indicating adequate convergent validity. Collectively, these findings support the adequacy of the measurement model for subsequent structural model analysis.

4.3. Discriminant Validity Assessment

Discriminant validity was assessed using both the Fornell–Larcker criteria and the heterotrait–monotrait ratio (HTMT). According to Fornell and Larcker (1981), discriminant validity is established when the square root of the average variance extracted (AVE) for each construct exceeds its correlations with other constructs. HTMT was further employed to provide a more rigorous assessment of discriminant validity, with values below the recommended threshold indicating that the constructs are empirically distinct.

Table 3. Discriminant Validity Assessment Using the Fornell–Larcker Criterion.

Variables	CMK	PEOU	CEX	EPM	PCV	PU	TRU	E-WOM	OBI
CMK	0.748
PEOU	0.020	0.816
CEX	0.032	0.706	0.817
EPM	0.034	0.628	0.732	0.799
PCV	0.013	0.478	0.514	0.511	0.820
PU	0.015	0.509	0.484	0.553	0.520	0.806
TRU	0.049	0.529	0.644	0.620	0.495	0.588	0.826
E-WOM	0.000	0.467	0.532	0.549	0.456	0.508	0.681	0.812
OBI	0.064	0.443	0.490	0.460	0.452	0.456	0.526	0.604	0.830

presents the results of the discriminant validity assessment using the Fornell–Larcker criterion. The diagonal values represent the square root of the AVE for each construct, while the off-diagonal values represent the correlations among variables. The results indicate that all square root AVE values were greater than the corresponding inter-construct correlations, satisfying the recommended Fornell–Larcker criterion for discriminant validity. Discriminant validity was further assessed using the heterotrait–monotrait ratio (HTMT). HTMT values ranged from 0.022 to 0.841, all below the recommended threshold of 0.90, providing additional evidence that the constructs are empirically distinct and that the measurement model demonstrates satisfactory discriminant validity. Collectively, these findings confirm that each construct has a unique conceptual domain and is empirically distinguishable from the other constructs, supporting the suitability of the measurement model for subsequent structural model analysis.

4.4. Structural Model Results

4.4.1. Estimated PLS-SEM Structural Model

The structural model was estimated using partial least squares structural equation modeling (PLS-SEM) in SmartPLS to examine the hypothesized relationships among the proposed constructs. Figure 3 presents the estimated PLS-SEM structural model with standardized path coefficients, factor loadings, and coefficients of determination (R²) for the endogenous variables. The estimated model demonstrates satisfactory explanatory capability for OBI among Thai tourists. The results indicate that technological, experiential, relational, and marketing factors encourage OBI within the digital tourism context.

4.4.2. Hypothesis Testing

Table 4. Results of Hypothesis Testing.

Hypothesis	Relationship	Path Coefficient (β)	t-Value	p-Value	Result
H1	PEOU → PU	0.275	12.632	0.000	Supported
H2	TRU → PU	0.443	19.538	0.000	Supported
H3	CEX → PCV	0.514	25.179	0.000	Supported
H4	E-WOM → TRU	0.488	22.645	0.000	Supported
H5	EPM → E-WOM	0.549	30.008	0.000	Supported
H6	EPM → TRU	0.352	16.740	0.000	Supported
H7	CMK → OBI	0.055	3.407	0.001	Supported
H8	PCV → OBI	0.132	5.400	0.000	Supported
H9	PU → OBI	0.105	4.349	0.000	Supported
H10	E-WOM → OBI	0.410	17.302	0.000	Supported
H11	CEX → OBI	0.151	6.301	0.000	Supported

shows the structural model results, including the standardized path coefficients (β), t-values, and p-values for the hypothesized relationships. The results indicate that all hypothesized relationships were statistically significant, providing support for all proposed hypotheses.

Among the structural relationships, E-WOM demonstrated the strongest direct effect on OBI (β = 0.410, t = 17.302, p < 0.001). CEX exerted the strongest effect on PCV (β = 0.514, t = 25.179, p < 0.001), while TRU significantly influenced PU (β = 0.443, t = 19.538, p < 0.001). In addition, EPM significantly enhanced both E-WOM (β = 0.549, t = 30.008, p < 0.001) and TRU (β = 0.352, t = 16.740, p < 0.001), suggesting that promotional activities indirectly strengthen OBI through relational and social influence mechanisms.

Furthermore, PCV (β = 0.132, t = 5.400, p < 0.001), PU (β = 0.105, t = 4.349, p < 0.001), CEX (β = 0.151, t = 6.301, p < 0.001), and CMK (β = 0.055, t = 3.407, p < 0.01) also positively influenced OBI. Overall, the findings support all proposed hypotheses and demonstrate that technological, experiential, relational, and marketing factors collectively explain OBI among Thai tourists.

4.5. Coefficient of Determination (R²)

The coefficient of determination (R²) was used to assess the explanatory power of the proposed structural model. As presented in

Table 5. R² Values of Endogenous Constructs.

Variables	R2
Perceived Value (PCV)	0.246
Perceived Usefulness (PU)	0.400
Trust (TRU)	0.550
Electronic Word of Mouth (E-WOM)	0.301
Online Booking Intention (OBI)	0.433

, the proposed framework demonstrated satisfactory explanatory capability across the endogenous constructs.

Among the endogenous constructs, TRU exhibited the highest explanatory power (R² = 0.550), followed by OBI (R² = 0.433), PU (R² = 0.400), E-WOM (R² = 0.301), and PCV (R² = 0.246). The model explained 43.3% of the variance in OBI, indicating that technological, experiential, relational, and marketing factors jointly contribute to tourists’ online booking intention. Overall, these findings demonstrate satisfactory explanatory capability for OBI among Thai tourists.

4.6. Model Fit and Predictive Assessment

4.6.1. Model Fit

Model fit was assessed using the standardized root mean square residual (SRMR) and the normed fit index (NFI). As presented in

Table 6. Model Fit Assessment.

Model Fit Index	Saturated Model	Estimated Model	Recommended Criterion	Assessment
SRMR	0.042	0.079	<0.08	Acceptable
NFI	0.877	0.870	>0.80	Acceptable

, the estimated model achieved an SRMR value of 0.079, which is below the recommended threshold of 0.08, indicating acceptable model fit. The model also achieved an NFI value of 0.870, indicating an acceptable level of overall model fit (Hair et al., 2024). Overall, these results support the adequacy of the proposed structural model.

4.6.2. Predictive Assessment (PLS Predict)

The predictive capability of the proposed model was further evaluated using the PLS predict procedure. All endogenous constructs produced positive Q²predict values, ranging from 0.261 to 0.383, indicating satisfactory out-of-sample predictive relevance. Among the endogenous constructs, TRU demonstrated the highest predictive capability (Q²predict = 0.383), followed by PU (Q²predict = 0.336) and E-WOM (Q²predict = 0.300). Overall, these findings indicate that the proposed model demonstrates satisfactory out-of-sample predictive performance.

4.7. Effect Size (f²)

Effect size (f²) was assessed to evaluate the relative contribution of each exogenous construct to the endogenous constructs. The largest effect sizes were observed for the relationships between EPM → E-WOM (f² = 0.431), E-WOM → TRU (f² = 0.370), and CEX → PCV (f² = 0.359), indicating large effects. Moderate effect sizes were identified for TRU → PU (f² = 0.235), EPM → TRU (f² = 0.192), and E-WOM → OBI (f² = 0.184). The remaining structural relationships demonstrated small or negligible effect sizes. Overall, these findings indicate that the proposed structural model includes relationships with varying levels of practical significance, ranging from negligible to large effects.

4.8. fsQCA Results

The fsQCA analysis was conducted to identify multiple configurational pathways leading to high OBI among Thai tourists. All study constructs were calibrated into fuzzy-set membership scores using the direct calibration method. The results identified multiple sufficient configurations associated with high OBI, supporting the principle of equifinality and indicating that tourists may achieve high booking intention through alternative combinations of technological, experiential, relational, and marketing conditions. The findings also demonstrate causal complexity, suggesting that the influence of an individual condition relies on its interaction with other causal conditions within specific configurations. In addition, E-WOM, TRU, CEX, and EPM usually appeared as core causal conditions, highlighting their central role in the digital tourism context. The overall solution coverage was 0.763, indicating that the identified configurations explained a substantial proportion of the cases associated with high OBI, while the overall solution consistency of 0.815 demonstrated acceptable configurational consistency.

4.8.1. Calibration and Necessity Analysis

Prior to the fsQCA analysis, all variables were calibrated into fuzzy-set membership scores ranging from 0 to 1 using the direct calibration method. Following recommended fsQCA procedures, three qualitative anchors were specified for each construct: full non-membership, the crossover point, and full membership.

Table 7. Calibration criterion for fsQCA analysis.

Variables	Variable Code	Full Non-Membership	Crossover Point	Full Membership
Content Marketing	CMK	4.2	5.8	6.8
Perceived Ease of Use	PEOU	4.0	5.6	6.8
Customer Experience	CEX	4.2	5.6	6.8
Electronic Promotion	EPM	4.2	5.6	7.0
Perceived Value	PCV	4.0	5.6	6.8
Perceived Usefulness	PU	4.0	5.6	6.8
Trust	TRU	4.0	5.6	6.8
Electronic Word of Mouth	EWOM	4.0	5.6	6.8
Online Booking Intention	OBI	3.8	5.6	6.8

presents the calibration threshold used in this study. Following calibration, necessity analysis was conducted to determine whether any single condition was necessary for high OBI. According to fsQCA guidelines, a condition is considered necessary when the consistency value exceeds 0.90. The results indicated that none of the individual conditions exceeded the recommended threshold, suggesting that no single factor was necessary to explain high OBI. These findings support the configurational logic of fsQCA and indicate that booking intention emerges from multiple combinations of technological, experiential, relational, and marketing conditions.

4.8.2. Sufficient Configurations

Table 8. Sufficient Configurations Leading to High Online Hotel Booking Intention.

Configurations	E-WOM	TRU	PU	PCV	EPM	CEX	PEOU	CMK	Raw Coverage	Unique Coverage	Consistency
1		⊗	⊗	⊗	⊗	●	●		0.296	0.006	0.878
2	●	⊗	⊗	⊗	⊗			●	0.303	0.024	0.863
3	●	●	●	●	●	●			0.519	0.009	0.932
4	●	●	●	●		●	●		0.506	0.006	0.937
5	●	●	⊗	⊗	⊗	⊗	⊗		0.302	0.017	0.887
6	⊗	⊗	⊗	⊗		●	●	⊗	0.253	0.003	0.876
7	⊗	⊗	●	●	⊗	⊗	⊗		0.263	0.019	0.861
8	●	●	●	●	●	⊗	●		0.272	0.003	0.896
9	●	●	⊗	⊗	●	●	●		0.297	0.003	0.906
10	●	●	⊗	⊗	●	●		●	0.270	0.002	0.898
11		●	⊗	⊗	●	●	●	●	0.268	0.002	0.902
12	⊗		●	●	●	●	●	●	0.247	0.006	0.903
13	⊗	⊗	⊗	⊗	●	●	⊗	●	0.234	0.002	0.882
14	⊗	⊗	●	●	⊗	●	⊗	●	0.226	0.004	0.899
15	●	●	●	●		●		●	0.391	0.000	0.930
16	●	●	●	●	⊗		⊗	●	0.255	0.000	0.914

|Solution Coverage|0.763| Solution Consistency |0.815|. Note: ● indicates the presence of a causal condition, whereas ⊗ indicates its absence. Blank cells indicate “don’t care” (irrelevant) conditions. Raw coverage represents the proportion of cases explained by each configuration, whereas unique coverage indicates the exclusive explanatory contribution of each configuration. Solution coverage represents the overall proportion of cases explained by all identified configurations, while solution consistency reflects the overall consistency of the configurational solution. Abbreviations: EWOM = Electronic Word of Mouth; TRU = Trust; PU = Perceived Usefulness; PCV = Perceived Value; EPM = Electronic Promotion; CEX = Customer Experience; PEOU = Perceived Ease of Use; and CMK = Content Marketing.

presents the sufficient configurational pathways associated with high OBI. The results demonstrate that multiple combinations of causal conditions can jointly produce the same outcome, supporting the principle of equifinality in fsQCA analysis.

Configurations 3 and 4 exhibited the highest raw coverage values, indicating that these pathways explain the largest proportion of cases associated with high OBI. Both configurations consistently included E-WOM, TRU, PU, PCV, and CEX, suggesting that the simultaneous presence of relational, technological, and experiential factors plays a critical role in shaping tourists’ booking decisions. Across all identified configurations, E-WOM, TRU, PU, PCV, CEX, and EPM frequently appeared as important causal conditions, highlighting their central role in explaining high OBI. Overall, the solution coverage of 0.763 and solution consistency of 0.815 indicate satisfactory configurational performance, demonstrating that multiple alternative combinations of technological, experiential, relational, and marketing conditions can explain high OBI among Thai tourists.

5. Discussion

5.1. Discussion of PLS-SEM Structural Results

5.1.1. TAM Perspective

The findings support the Technology Acceptance Model (TAM) by demonstrating that perceived ease of use positively influences perceived usefulness, while perceived usefulness positively affects OBI (Li et al., 2024). These results indicate that tourists are more likely to adopt online hotel booking platforms when the systems are perceived as user-friendly and functionally beneficial. The significant relationship between perceived usefulness and booking intention further indicates that technological functionality remains an important determinant of online consumer behavior in digital tourism environments (Terrah et al., 2024). However, technological perceptions alone were insufficient to fully explain OBI. Although PU significantly influenced booking intention, its effect size was smaller than those of E-WOM and CEX (Baydeniz et al., 2024; Salameh et al., 2022). This finding suggests that online hotel booking behavior extends beyond technological evaluations and encompasses broader experiential and relational dimensions.

5.1.2. S-O-R Perspective

The findings support the Stimulus–Organism–Response (S-O-R) framework (Morrison et al., 2023). CEX, EPM, and E-WOM, as external stimuli, significantly influenced tourists’ internal states, such as TRU, PCV, and PU, which subsequently influenced OBI (Al-Saad et al., 2024). In particular, CEX exerted a strong positive effect on PCV, suggesting that favorable digital interactions enhance tourists’ perceptions of the overall value of online booking platforms (Bouchareb, 2025). Furthermore, TRU significantly strengthened PU, indicating that tourists are more likely to perceive booking platforms as useful when they trust the platform and its information. E-WOM also positively influenced TRU and OBI, highlighting the important role of social interaction and peer-generated information within digital tourism decision-making processes. Collectively, these findings indicate that OBI is shaped by experiential and relational mechanisms in addition to technological factors, supporting the central proposition of the S-O-R framework (González-Mohíno et al., 2024).

5.1.3. Signaling Theory Perspective

The findings are also consistent with Signaling Theory by demonstrating the significant effects of EPM and CMK on OBI (Sharma et al., 2023). EPM positively influenced both TRU and E-WOM, indicating that promotional activities function as important informational signals that reduce uncertainty and strengthen tourists’ confidence in online booking platforms (Garg & Kumar, 2021). In digital tourism environments characterized by information asymmetry, promotional communication serves as an important mechanism for enhancing platform credibility and encouraging consumer engagement. In addition, CMK positively influenced online booking intention, although with a relatively smaller effect size. This finding suggests that informative and persuasive digital content contributes to tourists’ booking decisions by improving information transparency and strengthening platform attractiveness. Collectively, these findings support the central proposition of Signaling Theory, demonstrating that digital marketing signals reduce information asymmetry and encourage OBI (Chandra & Mishra, 2025).

5.2. Discussion of fsQCA Findings

5.2.1. Technology-Driven Configurations

The findings support the Technology Acceptance Model (TAM) by demonstrating that PU and PEOU frequently appeared in sufficient configurations leading to high OBI (Davis, 1989). These findings suggest that tourists are more likely to develop high OBI when booking platforms are perceived as useful and easy to use. Configurations involving PU and PEOU indicate that technological functionality remains an important pathway to online booking intention within digital tourism environments (Mohamad et al., 2021; Terrah et al., 2024). However, the fsQCA results further indicate that technological perceptions alone were insufficient to explain high OBI. In several configurations, high booking intention emerged only when technological conditions interacted with experiential, relational, or marketing-related conditions. These findings highlight the configurational complexity of online hotel booking behavior, whereby technological conditions operate in combination with other causal conditions rather than as isolated determinants (Saha & Biswas, 2025; Salameh et al., 2022).

5.2.2. Experience-Centered Configurations

Multiple sufficient configurations highlighted the combined roles of CEX, PCV, and TRU, supporting the S-O-R framework (Gahler et al., 2022). CEX appeared frequently across the identified pathways, indicating that positive digital experiences represent a central mechanism influencing tourists’ booking decisions. In addition, PCV and TRU jointly contributed to high OBI in several configurations, emphasizing the importance of internal psychological states within online tourism environments (Tiamiyu et al., 2022). The findings further demonstrate that high OBI can emerge even when certain technological conditions are absent, provided that strong experiential and relational conditions are present (Baydeniz et al., 2024). These findings reinforce the importance of experiential and relational mechanisms in shaping tourists’ online booking decisions.

5.2.3. Social Influence and Signaling Configurations

E-WOM, EPM, and CMK frequently appeared across the sufficient configurations, highlighting the importance of social influence and signaling mechanisms in online hotel booking behavior (Chen & Ye, 2023). E-WOM consistently appeared in several high-coverage pathways, indicating that online reviews and user-generated content strongly shape tourists’ perceptions and booking decisions (Adekuajo et al., 2023). Furthermore, EPM and CMK functioned as important market signals that reduced uncertainty and strengthened tourists’ confidence in booking platforms (Warintarawej et al., 2024). The fsQCA findings also support the principle of equifinality, demonstrating that multiple alternative combinations of technological, experiential, relational, and marketing conditions can jointly produce high OBI (Terrah et al., 2024). These findings further highlight the causal complexity of online hotel booking behavior and complement the symmetric relationships identified through PLS-SEM analysis.

5.2.4. Comparison Between PLS-SEM and fsQCA Findings

The integration of PLS-SEM and fsQCA provides complementary insights into OBI by explaining both net effects and configurational causality. The PLS-SEM findings identified E-WOM as the strongest direct predictor of online booking intention, while fsQCA further confirmed that E-WOM consistently appeared across several sufficient configurations. This convergence suggests that E-WOM functions as a critical conversion trigger in digital tourism environments, supporting the important role of social influence and user-generated information in consumer decision-making (Aghazada, 2023; Tùng & My, 2023).

TRU and CEX also demonstrated complementary findings across both analytical approaches. PLS-SEM revealed that trust significantly enhanced PU, while CEX positively influenced PCV and OBI. The fsQCA results showed that TRU, CEX, and PCV usually appeared together in multiple configurations leading to high booking intention. These findings indicate that online booking behavior is shaped not only by isolated relationships but also by the interaction of experiential and relational conditions within broader causal structures (Amin et al., 2021; Emam & Abdelaal, 2021).

The comparison between SEM and fsQCA highlights the methodological value of adopting a hybrid analytical approach in online tourism research. While PLS-SEM explains the strength and significance of individual relationships, fsQCA captures the configurational complexity and equifinality underlying consumer behavior by identifying multiple pathways leading to high OBI. Together, these complementary approaches provide a more comprehensive understanding of digital consumer decision-making in online hotel booking platforms, thereby strengthening the methodological contributions of this study and providing richer theoretical insights into online hotel booking intention (Hair et al., 2024; Terrah et al., 2024).

6. Limitations and Future Research

Despite its contributions, this study has several limitations that should be acknowledged. First, the study employed a cross-sectional research design, which limits the ability to establish causal relationships among the investigated constructs. Future studies may adopt longitudinal or experimental designs to examine how tourists’ online hotel booking intention evolves over time.

Second, the data were collected exclusively from Thai tourists. Although this provides important empirical evidence within the Thai tourism context, the generalizability of the findings to other cultural and tourism settings may be limited. Future research should validate the proposed framework across different countries and tourism markets to examine its cross-cultural applicability.

Third, the study relied on self-reported questionnaire data. Although common method bias was assessed and found not to pose a significant concern, future studies may benefit from incorporating multiple data sources or behavioral data to further strengthen the robustness of the findings.

Finally, this study examined online hotel booking intention across online booking platforms without distinguishing among specific platforms or service providers. Future research could compare different online travel agencies (OTAs) and booking platforms to determine whether technological, experiential, relational, and marketing factors operate differently across digital tourism environments.

Overall, future studies are encouraged to extend the present framework by incorporating additional contextual factors, diverse research settings, and longitudinal evidence to further advance the understanding of online hotel booking behavior.

7. Conclusions

This study examined the determinants of online hotel booking intention (OBI) among Thai tourists by integrating PLS-SEM and fsQCA within a unified framework based on the Technology Acceptance Model (TAM), the Stimulus–Organism–Response (S-O-R) framework, and Signaling Theory. The findings demonstrate that technological, experiential, relational, and marketing factors jointly influence OBI within digital tourism environments. The PLS-SEM analysis identified E-WOM as the strongest direct predictor of OBI, while CEX significantly enhanced PCV, TRU strengthened PU, and EPM indirectly influenced OBI through TRU and E-WOM. Complementing these findings, the fsQCA analysis demonstrated that multiple configurational pathways can lead to high OBI, supporting the principles of equifinality and causal complexity. Rather than being driven by a single dominant factor, high OBI emerged through different combinations of technological, experiential, relational, and marketing conditions, with E-WOM, TRU, CEX, and EPM frequently appearing across the identified configurations. Theoretically, this study extends the literature by integrating TAM, the S-O-R framework, and Signaling Theory to explain online hotel booking behavior within a unified conceptual framework. Methodologically, the findings demonstrate the complementary value of combining PLS-SEM and fsQCA, providing insights into both net effects and configurational causality. Practically, the findings suggest that hotel businesses and online booking platforms should prioritize customer experience, trust-building strategies, E-WOM management, and digital promotional activities to strengthen tourists’ online booking intention. In addition, this study provides empirical evidence from the Thai tourism context, contributing to a broader understanding of online consumer behavior in emerging digital tourism markets.