Hotel Guest Satisfaction: A Predictive and Discriminant Study Using TripAdvisor Ratings

Abstract

Understanding and promoting guest satisfaction is central to the economic sustainability of the hospitality industry. Satisfaction influences consumers’ booking intentions, hotel choice, loyalty, and the reputation and performance of accommodation establishments. Thus, accurate decision making by hotel managers relies on trustworthy and easily accessible information on the variables that affect guest satisfaction. Nowadays, this information is available through reviews and ratings provided by online platforms, such as TripAdvisor. Indeed, much research into guest satisfaction uses TripAdvisor reviews. However, this study aims to analyse guest satisfaction using only TripAdvisor ratings. These ratings can be more succinct and tractable indicators than reviews. A sample of 118 hotels in Cape Verde and the Azores, two archipelagos belonging to Macaronesia, and a descriptive, predictive, and discriminant methodology are employed for this purpose. Four main results are obtained. First, the rated items on TripAdvisor are consistent with the scientific literature on this topic. Second, TripAdvisor ratings are valid and reliable. Third, TripAdvisor ratings can predict guest satisfaction based on the perceived quality of hotel services. Fourth, there are significant differences in ratings depending on the tourism destination chosen. These results are of interest to researchers, tourists, as well as hotel, destination, and platform managers.

1. Introduction

The economic sustainability of tourism and the hospitality industry is at the forefront of tourism development due to the sector’s social and economic importance. This sector is also characterised by high competition, growing consumer demands, and recent crises and conflicts that have affected the industry and its sustainability (Alqaralleh et al., 2025; Laachach & Alhemimah, 2024; D. Zhang & Xie, 2023). This sustainability is even more significant in the case of island destinations, which have specific characteristics and suffer from tourism’s negative impacts more than other destinations (Guzmán-Pérez et al., 2023; Jacob et al., 2025; Ruggieri et al., 2022). In this context, understanding and promoting guest satisfaction is crucial to the economic sustainability of the hospitality industry (V. Sharma & Bhat, 2022). In fact, hotels are the reference frame for guests’ experience and hence where guest satisfaction is achieved (Josimović et al., 2025; Lim et al., 2024; Z. Zhao et al., 2024). In particular, research has found that hotel establishments are key for developing guests’ tourism experience and satisfying their needs, desires, and expectations (D. Zhang & Xie, 2023; Z. Zhao et al., 2024). Moreover, the hotel sector is particularly sensitive to any adverse effects on tourism, such as natural disasters (Song et al., 2022), terrorist attacks (Kubickova et al., 2019), political unrest (Ivanov & Stavrinoudis, 2018), and pandemics (Guo et al., 2021). All these events have a significant impact on guests (D. Zhang & Xie, 2023; Z. Zhao et al., 2024). Satisfaction is understood as the result of guests’ assessment of the difference between their experience at the hotel and their expectations (Kusumah, 2024; Oliver, 1980). If the experience exceeds expectations, there is confirmation, and if the opposite occurs, there is disconfirmation (Kolesárová et al., 2024; León et al., 2025; McMullen, 2024). The assessment of this difference is usually based on service (e.g., cleanliness, safety), attributes of service provision (e.g., speed, resolution), or elements related to the product (e.g., location) or price (Wu et al., 2024; Z. Zhao et al., 2024). The literature also highlights the importance of satisfaction due to its influence on guests’ hotel choices and loyalty, as well as on hotel performance and reputation (El-Adly et al., 2024; Jiang et al., 2024; Martín et al., 2020). Although there is a high degree of consensus on the consequences of satisfaction, the literature contains specific gaps and contradictions regarding the antecedents or drivers of guest satisfaction. For example, some authors suggest that perceived quality may be a precursor to guest satisfaction, while other researchers suggest the opposite (Nguyen & Nguyen, 2023; Verma et al., 2023). Furthermore, virtually no studies have analysed the combined influence of specific variables on guest satisfaction, including hotel attributes, room rates, perceived quality, and electronic word-of-mouth communication. These variables have generally been studied separately, not together.

The widespread use of e-WOM communication and digital tourism platforms, like TripAdvisor, has increased the focus on guest satisfaction (Feng et al., 2024; Solano-Sánchez et al., 2021). These platforms help guests plan their hotel stays and reduce uncertainty when making decisions about their stay (Islam, 2025; Zelenka et al., 2021; M. Zhang et al., 2025). On these platforms, consumers share their hotel experiences through reviews and ratings, reporting their satisfaction in a reliable, impartial, and credible manner (Arici et al., 2023; Gajić et al., 2024). It has been demonstrated in the literature that, albeit with certain contradictions, reviews and ratings relating to hotel service or attributes influence guest satisfaction (Huma et al., 2020; Pandey et al., 2023; N. Sharma & Arora, 2024; Jyoti et al., 2024). Consequently, e-WOM communication generates realistic expectations that are likely to be fulfilled, thus generating satisfaction (Mekoth et al., 2023; J. M. Kim & Han, 2023). In addition, platforms provide hotel managers with real-time information about guest satisfaction directly and at low cost, enabling them to make improvements and receive feedback (Deng et al., 2023; Feng et al., 2024; Yilmaz & Aytekin, 2018). Despite the advantages and benefits of e-WOM communication, this type of communication has not been without its criticisms and disadvantages. These have been related, among other things, to the accuracy of the information, the difficulty of case-by-case analysis, the time lag between data collection and the customer experience, and the failure to consider cultural differences (Hung et al., 2023; Kusawat & Teerakapibal, 2024).

Compared to reviews, ratings involve simplified, reliable, and valuable information to guests, which reduces their search time and effort (Burkov & Gorgadze, 2023; Ntemi & Kotropoulos, 2022). For researchers, they are also simple, clear, and straightforward metrics that are very useful either in quantitative studies or in combination with reviews in mixed studies (Deng et al., 2023; Islam, 2025). Nonetheless, researchers have used ratings far less than reviews, particularly in predictive and discriminant studies on guest satisfaction (Manolitzas et al., 2022; Feng et al., 2024). In addition, the literature highlights the need to further study the asymmetry associated with reviews, that is, the existence of differences in reviews based on criteria such as the hotel guest’s origin or the establishment type (Davras & Caber, 2019; Galati & Galati, 2019). Finally, some authors also suggest the need to confirm the content validity and reliability of data from digital platforms to conduct rigorous scientific studies and make informed decisions about guest satisfaction (McMullen, 2024).

In light of the above, the overall objective of this study is to examine hotel guest satisfaction through ratings published on TripAdvisor (https://www.tripadvisor.com (accessed on 9 January 2025)) for hotels in the outermost region of Macaronesia, specifically the archipelagos of Cape Verde and the Azores. Both archipelagos have similar natural and tourism characteristics that allow for joint study, but they also differ in some respects. Cape Verde, an independent country, is characterised by its dry and warm climate, beaches, and desert landscapes. However, the Azores, an autonomous region of Portugal, have a cooler and rainy climate, and a landscape of mountains, lagoons, and volcanic activity.

Taking the above into account, this study raises the following research questions (RQ):

RQ1: Are TripAdvisor’s variables and metrics scientific, and are data from the ratings valid and reliable?

RQ2: What influence do hotel attributes, room rates, perceived quality, and e-WOM communication have on guest satisfaction?

RQ3: Are there differences between Cape Verde and the Azores regarding the variables in the study?

A literature review and an integrated quantitative methodology are used in this study. Firstly, a review of the literature on guest satisfaction and TripAdvisor information will provide insights into whether TripAdvisor’s system of variables and metrics aligns with scientific studies on this topic. Secondly, a descriptive analysis is undertaken to check the validity and reliability of the data provided by the digital platform. Thirdly, regression analysis will determine the predictive potential of TripAdvisor ratings concerning guest satisfaction. Finally, a discriminant analysis tests whether there are asymmetries and differences in the ratings according to the archipelago considered.

This study fills three gaps in the literature. Although the literature corroborates that guest satisfaction expressed on digital platforms is influenced by some variables, such as perceived quality, the findings reveal certain contradictions. The present study contributes by clarifying these contradictions. Furthermore, predictive studies of hotel guest satisfaction based solely on TripAdvisor ratings are virtually non-existent. This study only uses numerical ratings. Finally, some authors (e.g., Yang et al., 2023; Islam, 2025) have emphasised the importance of conducting further research on TripAdvisor metrics to corroborate their theoretical validity and reliability, thereby establishing a more robust scientific foundation. This study undertakes such an analysis.

This research is also valuable to five relevant stakeholders. First, it helps consumers understand the variables determining their hotel satisfaction, thus enabling better decision making. Second, it is helpful to hotel managers, who require predictive studies of guest satisfaction formation to optimise it and provide feedback. Third, the study can guide public tourism managers in formulating hospitality promotion policies and other policy measures. Fourth, it is of interest to researchers who need valid and reliable data and predictive causal models of guest satisfaction to perform their analyses. Finally, the results from this study can assist TripAdvisor managers, as they can be used to implement improvements to the platform based on scientific knowledge.

2. Hotel Guest Satisfaction in the Context of Digital Platforms

Guest satisfaction is a crucial variable in the intangible context typical of services such as tourism and hospitality (Floričić & Jurica, 2023; Perić et al., 2023). Oliver’s (1980) proposal that guest satisfaction results from an evaluation of the difference between expectations and experience performance is generally accepted in the literature (El-Adly et al., 2024; Y. J. Kim & Kim, 2022). As highlighted in the study’s introduction, the difference between expectations and the outcome of a guest’s experience is associated with service elements, attributes of service delivery, or elements related to the product and price. This evaluation is included explicitly or implicitly in textual reviews and implicitly in ratings, as ratings summarise satisfaction (J. M. Kim & Han, 2023; Padma & Ahn, 2020). In contrast to the partial or multifactorial approach to satisfaction, the global approach proposed by Oliver (1980) predominates in the literature, according to which guest satisfaction refers to a guest’s enjoyment of the entire hotel experience (Ban et al., 2019; Moreno-Perdigón et al., 2021). The presence of the global approach in TripAdvisor reviews is evident, as guests refer to their satisfaction with the experience as a whole (Carvalho et al., 2024; C. Li & Agyeiwaah, 2023; Preziosi et al., 2022) as well as the partial approach since they relate their satisfaction or dissatisfaction to service or to specific attributes of the hotel or stay (Song et al., 2022; Williady et al., 2022; Wu et al., 2024). Global and multifactorial approaches to satisfaction are also present when guests evaluate the hotel as a whole or as the different attributes offered by the platform (e.g., sleep quality, cleanliness) (Bi et al., 2020; Moro et al., 2020). Similarly, there is a high consensus among authors that satisfaction allows gradations, which can be observed qualitatively or quantitatively in many TripAdvisor reviews (Ahmad & Sharma, 2023; El-Adly et al., 2024). The level of satisfaction is also easily observed in numerical ratings, as guests assign values between 1 and 5 (El-Adly, 2019; Wu et al., 2024). Finally, it is now accepted that satisfaction has behavioural, cognitive, and affective contents (Mehraliyev et al., 2020; Ye et al., 2022). Behaviour, emotion, and judgement are present explicitly or implicitly in the textual reviews on TripAdvisor, as well as synthetically and implicitly in the ratings (Perdomo-Verdecia et al., 2024; Wu et al., 2024).

The precursors and successors of hotel guest satisfaction have been the subject of numerous studies in the literature (X. Zhang et al., 2023; Z. Zhao et al., 2024). In contrast to the traditional questionnaire approach, information on guest satisfaction can now be consulted through reviews and ratings provided by online platforms (Mehraliyev et al., 2020; Perdomo-Verdecia et al., 2024). Recently, reviews have been widely used, primarily for sentiment analysis. Reviews entail textual information on guest satisfaction that is open, original, timely, impartial, and unbiased (Vo et al., 2021; Yang, 2022). However, studies on hotel guest satisfaction using ratings are very scarce, despite their usefulness for consumers, hotel managers, and researchers (Perdomo-Verdecia et al., 2024; Valenzuela-Ortiz et al., 2023).

Regarding the successors of satisfaction, the literature highlights its influence on consumers’ choices and purchase behaviour, their participation in e-WOM communication, and their loyalty (Deng et al., 2023; McMullen, 2024). Also noteworthy is the influence of guest satisfaction on hotel image, reputation, and profitability (El-Adly et al., 2024; X. Zhang et al., 2023). Regarding the precursors of guest satisfaction, the authors note the existence of a large number of personal and contextual variables, among which perceived quality, room rates, and hotel attributes stand out (X. Zhang et al., 2023; Z. Zhao et al., 2024). Despite the large number of studies, some gaps and inconsistencies suggest the need for further research. For example, for some authors, perceived quality is a precursor of satisfaction, while for others, the opposite is true (Guizzardi et al., 2020; Vives & Ostrovskaya, 2024). Additionally, it would be helpful to identify the variables used on digital platforms subject to ratings that provide information on guest satisfaction and to conduct further studies on satisfaction through ratings (Padma & Ahn, 2020; Z. Zhao et al., 2024).

3. Hypothesis Formulation

Regarding the need for more comprehensive studies on guest satisfaction through e-WOM communication, some authors suggest analysing the scientific basis of the theoretical and conceptual framework used by TripAdvisor, as well as the validity and reliability of its data (Shu et al., 2023; Zhou et al., 2024). Determining the scientific nature of the theoretical–conceptual framework used by TripAdvisor means determining whether the name, content, and metrics of the variables used by the platform are aligned with the literature in this field. Additionally, confirming the validity and reliability of the data provided by TripAdvisor involves verifying that they measure what they intend to measure, that they form a stable dimension, and that they are reproducible. It should be noted that, although the information provided by TripAdvisor is generally credible to consumers and used by hotel managers, some data published on the platform have been considered unhelpful, irrelevant, biassed, or asymmetrical (Choi & Leon, 2020; Konecnik & Petek, 2022). It should be noted that, although few published studies on tourism use TripAdvisor ratings, these studies are scientifically sound and have been published in prestigious journals. Nevertheless, the scientific nature of the quantitative information provided by TripAdvisor remains to be determined in these studies (e.g., Butler & Read, 2021; Manolitzas et al., 2022; Orea-Giner et al., 2022). Analysing these aspects would strengthen the descriptive, predictive, and discriminative studies by researchers and enable managers to make more rigorous and scientifically based decisions (Islam, 2025; Shu et al., 2025). Therefore, the first and second hypotheses state the following:

Hypothesis 1 (H1): 

TripAdvisor’s conceptual and metric framework is consistent with the literature in this field.

Hypothesis 2 (H2): 

Data related to TripAdvisor ratings are valid and reliable.

Regarding the background of hotel guest satisfaction, various studies have shown that both offline and online guest satisfaction depend, directly or indirectly, on tangible and intangible hotel attributes (Jang et al., 2018; Mamengko, 2022; Martín et al., 2020; D. Kim & Perdue, 2013). Among these attributes, the literature mentions the hotel category, the size of the establishment, and the number of services offered to guests (Srivastava & Kumar, 2021; Y. Zhao et al., 2019). Rhee and Yang’s (2015) review identified six hotel attributes: location, cleanliness, rooms, sleep quality, service, and value. This result coincides with the attributes that TripAdvisor presents to guests for evaluation through ratings on the platform. For their part, Jang et al. (2018) identified more than 30 hotel attributes that influenced satisfaction through text mining and TripAdvisor reviews, highlighting changes in the importance of hotel attributes over time.

The importance and influence of hotel room rates are well established in the literature (e.g., Martin & Hall, 2020). In particular, the direct influence of hotel room rates on guest satisfaction has been demonstrated (e.g., Aryal et al., 2023; Demydyuk & Carlbäck, 2024; Wijaya & Purba, 2021). This is due to the expectations that price generates, as consumers use the room rate as an indicator in hotel selection decisions to predict their satisfaction with intangible experiences (Huang et al., 2018; Nicolau et al., 2020; Rady et al., 2023). Price has also been found to indirectly influence satisfaction through variables such as perceived service quality or perceived value, particularly in service companies (Konuk, 2019; Rady et al., 2023). A better understanding of how room rates relate to satisfaction allows hotel managers to obtain feedback and develop better pricing policies and strategies while better meeting customers’ needs, desires, and expectations (Demydyuk & Carlbäck, 2024; Prum et al., 2024). In light of the above, the third and fourth hypotheses propose the following:

Hypothesis 3 (H3): 

Hotel attributes positively affect guest satisfaction.

Hypothesis 4 (H4): 

Room rates positively affect guest satisfaction.

The literature also confirms that perceived quality and perceived service quality are the most important determinants of guest satisfaction (Gheibdoust et al., 2024; Ríos-Martín et al., 2020; Z. Zhao et al., 2024). According to Zeithaml (1988), the concept of perceived quality refers to the excellence of a good or service (Dias & Lavaredas, 2024; Gálvez-Ruiz et al., 2023; Ju et al., 2019), while service quality is a component of perceived quality and refers to the assessment of a specific aspect of service provision (Assaker et al., 2020; Šerić, 2018). Several works have identified location, sleep quality, room, service, and cleanliness as essential elements of perceived quality, which are precisely those used by TripAdvisor (Brochado & Brochado, 2019; Lee et al., 2020; Padma & Ahn, 2020). For example, in a study conducted in five-star hotels in Spain, using content generated by TripAdvisor users, several hotel quality attributes were identified that influenced guest satisfaction: services, rooms, and location (Ríos-Martín et al., 2020). Another study conducted in New York using TripAdvisor data identified the following quality attributes to influence satisfaction: sensory experience, sleep quality, location, room, service, and cleanliness (Lee et al., 2020). Similarly, numerous authors have studied the effects and impact of e-WOM communication on attitude and purchase intention (Premananto et al., 2023; Khan et al., 2024), consumer loyalty (Madi et al., 2024; Prasetio et al., 2024), and even the adoption of e-WOM communication by consumers themselves (Liang et al., 2021; Verma et al., 2023). Likewise, the influence of e-WOM communication on consumer satisfaction has been confirmed in a general context (Huma et al., 2020; Jyoti et al., 2024; Pandey et al., 2023; N. Sharma & Arora, 2024) and, specifically, in the hotel sector (R. Liu et al., 2022; S. Singh & Kathuria, 2019; X. Zhang et al., 2023). This is because reviews and ratings create expectations in consumers that are motivated by the credibility of this type of communication, particularly among young people, as well as by certain characteristics specific to this type of communication (e.g., volume, valence, semantics) (Berger et al., 2022; Kusawat & Teerakapibal, 2024; Yoon et al., 2019). In light of the above, the fifth and sixth hypotheses state the following:

Hypothesis 5 (H5): 

Perceived quality positively influences guest satisfaction.

Hypothesis 6 (H6): 

e-WOM communication positively influences guest satisfaction.

Some authors have demonstrated that asymmetry in the reviews and ratings of certain aspects or variables has distinct influences. Asymmetry means that the exact change in the positive or negative performance of certain variables (e.g., hotel attributes) leads to different changes in other variables (e.g., guest satisfaction) (H. Li et al., 2020; Y. Liu et al., 2017). It has been shown, for example, that reviews with positive asymmetry, i.e., those with more positive than negative aspects, increase consumers’ perceptions of the product or service before consumption, thereby having a very positive influence on their purchasing decision (Y. Zhao et al., 2019; Konecnik & Petek, 2022). Asymmetry may be determined by the segment to which the guest belongs, the type of hotel, and the type of traveller, among other factors (Davras & Caber, 2019; Galati & Galati, 2019; Jang et al., 2018; Nunkoo et al., 2020; Radojevic et al., 2018; Ying et al., 2020). Based on the above, the seventh hypothesis states the following:

Hypothesis 7 (H7): 

There are significant differences in the ratings of the variables included in this study depending on the archipelago (Cape Verde and the Azores).

4. Methodological Design

4.1. Research Context

This study uses only quantitative data on hotels in Cape Verde and the Azores published on the TripAdvisor platform and obtained in September 2024. Cape Verde and the Azores are two archipelagos that form part of Macaronesia, along with the Canary Islands and the island of Madeira. The archipelagos of the Azores and Cape Verde were chosen as there are concerns about the economic sustainability of the hospitality industry in these tourism destinations (Cadima Ribeiro et al., 2023; Fernandes Neves Barbosa et al., 2024). Both archipelagos share a mass sun-and-beach tourism model, limited accommodation capacity, similar source markets, and the same accommodation types (Mendes et al., 2024; Sarmento & Monteiro, 2023).

TripAdvisor (https://www.tripadvisor.com) is an American tourism platform and a pioneer in travel-related e-WOM (Sánchez-Vargas et al., 2022; Yang et al., 2023). TripAdvisor’s popularity has been growing among consumers, hotel managers, and researchers because the platform offers free, easy-to-use information that covers a wide range of aspects regarding the characteristics of tourism establishments (Abeysinghe & Bandara, 2022; Ertaş & Karakan, 2024; Mokgehle & Fitchett, 2024).

4.2. Data Collection Procedure

Data were obtained by directly consulting the information on the TripAdvisor platform. This data collection was carried out in August and September 2024.

4.3. Population and Sample

The hotel population in this study is composed of 201 hotels with three or more stars. The sample includes 58.71% of the population, that is, 118 hotels for which TripAdvisor provided information on all variables used in this study at the time of data collection. (

Table 1. Sample details.

Hotel Category	Archipelago		Total	%
	Azores	Cape Verde
3 stars	19	27	46	38.98%
>3 stars	41	31	72	61.02%
Total	60	58	118	100%
%	50.85%	49.15%	100%	---
Chi square < 0.05

). The sample size is adequate for a 95% confidence level and a 5% margin of error (Jhantasana, 2023; Zickar & Keith, 2023).

4.4. Variables

The observed variables that TripAdvisor provides quantitative data for are listed in

Table 2. Results of descriptive analysis.

Group	Label	Variables	Min	Max	∑	Mean	SD	Cramér p
HAT	1	Category	3	5	442	3.75	0.68	0.000
	2	Number of rooms	5	1151	13,534	114.69	170.90	0.000
	3	Number of room amenities	0	30	747	6.33	4.34	0.000
	4	Number of property amenities	6	67	3105	26.31	14.19	0.000
EWQ	1	Room tips	0	100	1828	15.49	23.15	0.000
	2	Number of reviews	7	11,516	91,165	772.58	1652.81	0.000
	3	Number of questions and answers	0	3983	19,336	163.86	517.70	0.000
RRA	1	Highest price	36	1272	16,876	143.02	129.02	0.000
	2	Lowest price	26	199	8003	67.82	29.32	0.000
	3	TripAdvisor price	34	672	12,450	105.51	72.56	0.000
PEQ	1	Cleanliness	3.5	5	507	4.30	0.37	0.000
	2	Service	3	5	491.50	4.17	0.44	0.000
	3	Hotel location	3	5	510.50	4.33	0.40	0.000
	4	Sleep quality	3	5	497.50	4.21	0.42	0.000
	5	Room	3	5	513.30	4.35	0.41	0.000
SAT	OCR	Overall customer rating	3	5	472.50	4.00	0.43	0.000
	VFM	Value for money	3	5	467.00	3.96	0.44	0.000

. It should be noted that the grouping of variables is for illustrative and ordering purposes only, since the analyses were carried out using individual variables, not constructs. These include four variables related to hotel attributes (HAT), three variables associated with the amount of e-WOM (EWQ), three variables related to room rates (RRA), five variables related to perceived quality (PEQ), and two variables related to guest satisfaction (SAT). It should be noted that two proxy dependent variables of satisfaction are used: “Overall customer rating” (OCR) and “Value for money” (VFM). Both variables are present on the TripAdvisor platform and fulfil the key criteria that different authors propose for this variable category. First, according to the value–perception gap theory, the two variables involve an assessment of expectations (Moreno-Perdigón et al., 2021). Second, both variables summarise satisfaction’s cognitive and affective characteristics (Valenzuela-Ortiz et al., 2023). Third, both variables capture the relevant causal effects (De Luna et al., 2017). Lastly, the two variables derived from the sample are pertinent to the research and are stable, easily measurable, and present a strong correlation (Nilsson et al., 2023). OCR is an all-inclusive indicator of satisfaction (Glaveli et al., 2022), and VFM reflects the overall guest experience and hence entails a suitable predictor of satisfaction (Albayrak & Caber, 2015; Chen & Chen, 2010; Zhuo & Wang, 2022).

4.5. Methods

Hypotheses 2 through 7 were tested using descriptive, predictive, and discriminant methods, using the software SPSS v29 (https://www.ibm.com).

First, data validity and reliability were checked through Bartlett’s test and Cronbach’s alpha indicator, respectively. Second, a descriptive data analysis was carried out, including variable significance. Third, two multiple linear regression analyses were performed, following the guidelines and suggestions of other authors (e.g., P. Singh et al., 2024). In line with the literature on customer satisfaction, the variable OCR was taken as a proxy for guest satisfaction in the former regression. At the same time, VFM was the proxy for the latter one. In the predictive study, a stepwise linear regression method was used, which involved an iterative step-by-step selection of independent variables. This method allows for the best model to be obtained based on the statistical significance of the variables and the model’s performance in terms of its fit to the data. Finally, a discriminant analysis was undertaken to test whether there were significant differences in the ratings according to the archipelago in which the hotels are located.

5. Results

5.1. Data Validity and Reliability

As shown in the literature review and the formulation of hypotheses, the variables used by the TripAdvisor platform are very similar to those considered in research studies on guest satisfaction (e.g., hotel attributes, room price, and perceived quality) (V. Sharma & Bhat, 2022). Similarly, the metrics used by TripAdvisor are based on a 5-point scale, similar to those used by many authors when using Likert-type scales (Ayyildiz et al., 2025).

Regarding construct validity, the Kaiser–Meyer–Olkin (KMO) and Bartlett’s tests were applied in a factorial analysis framework. The KMO test yielded a value of 0.88, while the significance value of the Bartlett test was less than 0.05 (p ≤ 0.05). These results confirm that the observed data form an invariant structure that can serve as a basis for interpreting the results (Ntalakos et al., 2025).

Regarding overall data reliability, Cronbach’s alpha indicator was calculated with a result of 74%, and the minimum acceptable value is 70% or even 60% in exploratory or preliminary studies (Koç et al., 2023). The alpha coefficient was calculated using only the independent variables. The small sample size, the number of variables, and the high diversity of aspects these variables address suggest acceptable reliability. Furthermore, no variables whose removal would increase the overall reliability of the data were observed.

5.2. Results of Descriptive Analysis

Table 2 presents the results of the descriptive analysis. The factor analysis performed effectively confirmed the existence of the factors that group the variables. It presents the minimum and maximum values reached by each variable, the total sum of values (Σ), the arithmetic mean, the standard deviation (SD), and Cramér’s p value of significance. Two characteristics of the data are worth noting: first, the wide range of values taken by the observed variables, and second, all observed variables related to perceived quality reached values equal to or higher than 3.

5.3. Results of Predictive Analysis of Overall Customer Rating

The “Overall customer rating” (OCR) was chosen as the dependent variable. The stepwise regression method provided four models, of which Model 4, with the highest coefficient of determination R², was selected. As shown in

Table 3. Results of regression analysis 1.

Model (R2)	Variables	Non-Standardised Coefficients		Standardised Coefficients	t	Sig. (p)
		B	Desv. Error	Beta
Model 4 (71%)	Service (PEQ2)	0.465	0.076	0.479	6.116	0.000
	Cleanliness (PEQ1)	0.323	0.085	0.281	3.788	0.000
	Lowest price (RRA2)	0.002	0.001	0.150	2.871	0.005
	Hotel location (PEQ3)	0.146	0.062	0.139	2.338	0.001
Dependent variable: Overall customer rating (OCR)

, this model explains 71% of the dependent variable, with the minimum recommended value being 50% and even 30% in social sciences. Furthermore, the Durbin–Watson coefficient is 1.92, demonstrating the model’s statistical validity. As can be seen in the table, “Service” is the variable with the highest predictive weight (β = 0.479) in the model, followed by “Cleanliness” (β = 0.281), “Lower price” (β = 0.150), and “Location” (β = 0.139). It should be noted that the observed variable “Service”, which guests rate on a scale of 1 to 5, represents a measure of overall service quality. In addition, the location rating does not refer to the archipelago, but rather to the hotel’s location within the destination. Guests rate the location section based on whether the hotel is centrally located. Furthermore, the weights of the four variables in the regression are significant, with p less than 0.05 (p < 0.05).

5.4. Results of the Predictive Analysis of the Value for Money

The “Value for money” (VFM) was taken as the dependent variable. The stepwise regression analysis yielded two models, of which Model 2, with the highest coefficient of determination, was selected. Model 2 predicts 55% of the dependent variable with only two variables, namely, “Service” and “Cleanliness”. The Durbin–Watson coefficient is 2.10, which shows the statistical validity of the chosen regression model. As can be seen in

Table 4. Results of regression analysis 2.

Model (R2)	Variables	Non-Standardised Coefficients		Standardised Coefficients	t	Sig. (p)
		B	Desv. Error	Beta
Model 2 (55%)	Service (PEQ2)	0.505	0.091	0.507	5.534	0.000
	Cleanliness (PEQ1)	0.337	0.108	0.287	3.130	0.000
Dependent variable: Value for money (VFM)

, the variable “Service” has a higher predictive weight (β = 0.507) than “Cleanliness” (β = 0.287).

The results in Table 3 and Table 4 show that, using either OCR or VFM as a proxy, “Service” and “Cleanliness” are the variables that most influence guest satisfaction in the context of TripAdvisor ratings. These two variables have the greatest weight in the regression analysis in Table 3 and are the only ones included in the best model in Table 4.

5.5. Results of Discriminant Analysis According to Archipelago

Discriminant analysis is a parametric technique widely used to determine the quantitative variables that best discriminate between two or more groups. The analysis produces a discriminant function that is a linear combination of the weights and scores of these variables. Like multiple linear regression, discriminant analysis can predict an outcome from a given set of predictors. The technique also allows the researcher to test whether cases are classified as expected theoretically. The attributes used to separate the groups must be discriminated between and known in advance. The discriminant analysis in this study followed the guidelines suggested by other authors (Park et al., 2022).

The discriminant analysis considers the archipelago to identify the groups, with groups 1 and 2 being hotels located in the Azores and Cape Verde, respectively. As can be seen in

Table 5. Basic indicators of discriminant analysis.

% of Variance	Sig. of M. Box	Eigenvalue	Canonical Correlation	Wilks Lambda	Sig. (p)	Centroids
						Azores	Cape Verde
100%	0.000	0.492	0.574	0.670	0.000	0.683	−0.707

, the resulting discriminant function explains 100% of the variance. Table 5 shows that the eigenvalues and the canonical correlation are far from zero, and that the Wilks Lambda indicator is far from one, with a significance level of less than 0.05 (p ≤ 0.05). Therefore, it can be said that there are some significant differences between the two archipelagos in terms of the variables observed, although these differences are not excessive.

The signs of the centroids tell us whether differences exist between the two archipelagos. As shown in Table 5, the Azores’ centroid has a positive sign, while Cape Verde’s centroid has a negative sign in the discriminant function. The standardised coefficients in

Table 6. Standardised coefficients of the discriminant analysis.

Group	Label	Variables	Standardised Coefficients
HAT	1	Category	0.371
	2	Number of rooms	−0.425
	3	Number of room amenities	0.144
	4	Number of property amenities	−0.056
EWQ	1	Room tips	0.473
	2	Number of reviews	1.013
	3	Number of questions and answers	0.076
RRA	1	Highest price	0.435
	2	Lowest price	−0.792
	3	TripAdvisor price	0.212
PEQ	1	Cleanliness	−0.104
	2	Service	0.692
	3	Hotel location	−0.124
	4	Sleep quality	0.208
	5	Room	0.114
SAT	OCR	Overall customer rating	−0.140
	VFM	Value for money	0.472

allow us to identify the differences between the two archipelagos, taking into account the signs of the centroids. Based on the results, and considering only standardised coefficients greater than 0.350 in absolute terms, the following can be stated:

(a)

Hotels located in the Azores score higher on the following variables:

HAT1: Category

EWQ1: Room tips

RRA1: Highest price

PEQ2: Service

SAT: Value for money (VFM)

(b)

Hotels located in the Cape Verde archipelago score higher on the following variables:

HAT2: Number of rooms

RRA2: Lowest price

6. Discussion

The results of this study confirm the first two hypotheses (H1 and H2) because the similarity of TripAdvisor variables and metrics to scientific proposals in the literature on this topic is verified and the validity and reliability of the TripAdvisor data are confirmed. This addresses the concerns and suggestions of different authors (Islam, 2025; Shu et al., 2025; Zhou et al., 2024). However, the third hypothesis (H3) is not confirmed because the regression analyses did not find that hotel attributes directly and significantly impact guest satisfaction, which contradicts the suggestions of some authors (e.g., Jang et al., 2018; Mamengko, 2022). There are three possible reasons for this result. First, previous studies may have used guest reviews rather than ratings. Second, regression analysis examines direct rather than indirect relationships. Therefore, the influence of hotel attributes on satisfaction may be indirect through other intermediate variables, such as price or perceived quality. Third, in a stepwise regression analysis, some variables are left aside. Thus, it could be possible that hotel attributes have a relative weight in a standard regression analysis with all variables, though a small one.

The results of this study confirm the fourth hypothesis (H4), as the first regression analysis shows that lower room prices have a significant influence on customers’ overall evaluation, as proposed by Aryal et al. (2023) and Demydyuk & Carlbäck (2024). However, this influence was not confirmed, at least not directly and significantly, concerning the second dependent variable, “Value for money”. As argued previously, this result could arise because stepwise regression analysis does not include all variables under study. Moreover, it could also be the case that the price effect on satisfaction is indirect through other intermediate variables.

The results of both regression analyses support the fifth hypothesis (H5), confirming that perceived quality is the variable that most influences satisfaction, whether understood as “Customer overall evaluation” or “Value for money”. This result confirms other authors’ view regarding the influence of quality on satisfaction (e.g., Gheibdoust et al., 2024; Z. Zhao et al., 2024), as well as the argument of some authors that quality precedes satisfaction (e.g., Barnes & Krallman, 2019; Gheibdoust et al., 2024).

However, the results do not confirm the sixth hypothesis (H6), as it was not possible to verify that the amount of e-WOM communication has a direct and significant impact on satisfaction, contradicting the findings of authors such as Berger et al. (2022) and Kusawat and Teerakapibal (2024). This may be because the types of variables used to measure e-WOM communication differ from those considered in other studies. For example, while other authors examined the amount of e-WOM communication using the number of times a guest rates an attribute as “Very good” or “Excellent”, this study used (the number of) “Room tips”, “Number of reviews”, and “Number of questions and answers”. In addition, as mentioned with other hypotheses, a stepwise regression analysis was performed instead of a standard regression with all variables, and only the direct influence on satisfaction was examined.

Finally, the seventh hypothesis (H7) regarding the existence of a certain asymmetry in the values of the variables observed depending on the archipelago (Azores and Cape Verde), as proposed by some researchers (e.g., Nunkoo et al., 2020; Ying et al., 2020), is confirmed by discriminant analysis. The discriminant analysis results suggest that, in contrast to Cape Verde, hotels in the Azores are of a higher category (number of stars), offer a more highly rated service, and are more expensive, with a higher number of opinions and recommendations about the rooms. On the other hand, hotels in Cape Verde are larger, less costly, and more prone to mass tourism.

7. Conclusions and Implications

This study achieved its overall objective of deepening both the theoretical and practical knowledge of hotel guest satisfaction in the context of digital platforms and through the exclusive use of TripAdvisor ratings. The starting point was that hotel guest satisfaction is a critical variable to achieve economic sustainability in hotels, as it influences consumers’ participation in e-WOM communication, as well as their booking intentions and loyalty. From a business perspective, satisfaction also affects a hotel’s reputation, business, and financial outcomes. Furthermore, achieving economic sustainability in tourism is particularly important in the case of island destinations due to the limitations associated with this type of destination and the negative impact that tourism can have on islands.

To date, the analysis of guest satisfaction has relied more on textual reviews from digital tourism platforms than on ratings. Reviews hold significant interest and trust for guests, who consult them while organising their visit or compose them to share their experiences. Reviews are likewise valuable to hotel managers, as they use them to take measures to promote satisfaction. Despite the advantages of synthetic indicators, researchers have used online ratings significantly far less than reviews. TripAdvisor ratings summarise guests’ assessments of quality and satisfaction and are valid, reliable, readily available, and cost-free indicators. Ratings also have great practical potential for analysis, building predictive models, and gaining insight into satisfaction, quality, and other variables. It should be noted that the literature shows that there are many internal and external consumer variables that directly or indirectly influence guest satisfaction. These include hotel attributes, prices, and most importantly, perceived quality. The influence of these variables on guest satisfaction is not without asymmetries, mainly due to the hotel type and guests’ origins. Finally, digital platforms such as TripAdvisor represent a meeting point to enrich the guest experience, promote satisfaction, and develop the economic sustainability of the hotel industry. Digital tourism platforms encourage the co-creation of value and coopetition among different stakeholders.

7.1. Theoretical Implications

This work contributes to the knowledge on the economic sustainability of the hospitality industry on islands through guest satisfaction as reflected in e-WOM communication through digital tourism platforms. The study fills a theoretical deficit in the literature on satisfaction, as it is based solely on ratings published by TripAdvisor, rather than reviews. The proximity of the variables and metrics used by TripAdvisor to those proposed by authors provides a scientific basis for decision-making processes and research based on TripAdvisor data. In addition, the literature review on hotel guest satisfaction in this study revealed evidence that these theoretical approaches are present in guest reviews and ratings, providing realism and applicability. Likewise, the study’s results reinforce in some cases or refute others in the weight given to the satisfaction of certain relationships in the literature, allowing for better understanding and complementarity with other studies. This work also clarifies specific theoretical gaps and contradictions present in the literature. For example, it confirms that perceived quality is a precursor of satisfaction in digital platforms and ratings. Finally, another theoretical contribution of this study refers to the inclusion of two theoretically valid proxy indicators.

7.2. Practical Implications

This study is of practical interest to consumers, hotel managers, public authorities, and online platforms, as it shows that TripAdvisor ratings are valid and reliable metrics for research and decision making. The information provided here is helpful for consumers when making hotel choices, thereby increasing their satisfaction, trust, and participation in e-WOM communication. The work is also valuable for hotel managers because its findings allow them to improve guest satisfaction. In this respect, perceived quality turns out to be the crucial variable for achieving this goal. Moreover, the improvement of guest satisfaction increases consumer loyalty, which results in more favourable reviews and ratings. This process bolsters the hotel’s reputation and its financial advantages, rendering the company more economically viable. This research also holds practical significance for public managers of the tourism destination, as it offers helpful insights for promoting tourism’s economic sustainability. Indeed, more satisfied guests recommend and/or return to the hotel and, consequently, to the destination. Moreover, the results can be useful for policy design. Finally, the study is valuable to TripAdvisor managers, as it confirms that the platform’s model is rigorous and scientifically sound. This enhances the platform’s reputation, as its services can contribute to both guest satisfaction and the economic sustainability of hotels and destinations.

7.3. Limitations and Future Research Lines

This study has three main limitations. Firstly, only hotels in Cape Verde and the Azores are considered, which limits the possibility of making comparisons that would be possible with a larger and more diverse sample. Secondly, this research is based exclusively on TripAdvisor’s variables, so other variables considered by other authors were not included. This may have limited the richness and scope of the study. Finally, despite the advantages of ratings, their quantitative nature only provides a summary of the evaluative and emotional processes associated with satisfaction. These processes are better captured by reviews, which are more detailed.

These limitations open avenues for future research. In particular, it would be interesting to analyse satisfaction using both qualitative and quantitative approaches. The analysis could be performed considering other tourism destinations, hotel types and consumer segments. In addition, the impact of digitization on satisfaction is particularly relevant in the context of online platforms. Finally, exploring the consistency between the sentiments conveyed in reviews and ratings also emerges as a promising research line.