ALL THAT GLITTERS IS NOT GOLD: BEYOND ONLINE HOTEL REVIEWS

INTRODUCTION

Online customer reviews in the hospitality industry are rapidly increasing in volume and importance and thus require the constant attention of researchers and hotel managers. In fact, research on online reviews is so extensive thatBore et al. (2017) noted that it revolves around eight different topics, of which review analysis is one. At the same time, hotel managers place so much value on online customer reviews that some evaluate comments and manage online reputation in real time (Gössling et al. 2018, 2019). This is because their engagement with online reviews improves customer ratings (Liu et al. 2015), customer satisfaction and repurchase intentions (Zhou et al. 2014;Ho 2017), service quality (Gössling et al. 2018), and financial performance (Xie et al. 2017). However, despite the undeniable benefits of proper online review management, customer reviews are abundant and growing exponentially (Alaei et al. 2019). Therefore, hotel managers are under pressure to think in a so-called fast mode (cf. Kahneman 2013), i.e., to process information automatically while reducing creative solution seeking. When hotel managers give high priority to online reviews and yet processed them fast, there is a risk that they will overlook important points, especially when textual comments in online reviews are thematically mismatched with other available sources of information.

For example, a thematic mismatch could exist within online reviews - between textual comments and rating categories. That would be if predefined rating categories were thematically narrow and related to e.g., staff, comfort, price, and location, while textual comments freely written by customers addressed broader spectrum of themes, including e.g., facilities and food. The presence of such a mismatch would suggest that the easy-to-process information, i.e., the hotel ratings that potential guests process first when reading online reviews (Sparks and Browning 2011), is not based on themes that are most important to guests according to their textual comments. This mismatch may also suggest that hotel managers' improvements in what customers say is most important according to their textual comments do not translate into improvements in reviews" numerical ratings.

There could also be a thematic mismatch between textual review comments and other sources which provide information on consumer experience such as scales used in surveys. This could be the case because the scales in surveys often measure experience along a wide range of themes such as e.g., functional, emotional, social, and other values, whereas textual comments may be narrower and limited only to values which guests expect from a hotel the most like e.g., functional value. This thematic mismatch would imply either an overrepresentation of certain themes in the consumer experience scales or their underrepresentation in the textual comments, or both, and in either case implies misguided decisions by hotel managers who think fast and rely only on one or the other.

1. LITERATURE REVIEW

To build the arguments, we review three relevant lines of research. First, we introduce a dual system theory. It explains an important assumption of this research, namely that hotel managers are exposed to circumstances that often force them to think fast. Fast thinking makes them susceptible to decision-making problems that arise when the concepts under study are thematically mismatched. Second, we identify concepts, i.e., consumer experience scales and marketing tools, which are most important in the hospitality industry to ensure relevance of the analysis and the findings. Finally, we review the results of previous research on thematic scope of customer online reviews to point to the research gap of the past research and the contribution of the present.

1.1. Dual system theory

To explain why it is important to uncover potential mismatches between the thematic scope of textual review comments on the one hand and review rating categories, consumer experience scales, and marketing tools on the other, we rely on dual system theory. According to this theory, there are two modes of thinking, the so-called two systems (Kahneman 2013). The first is automatic, effortless, and fast, while the second focuses on complex mental activities, reasoned decisions, hypothetical thinking, and unconventional problem solving and is therefore slow (Evans and Stanovich 2013).McCallum (2012) points out that managers are expected to always use slow thinking when making important decisions, but they do not. Due to the ongoing pressure and repetitive nature of their work, they start thinking in fast mode. However, when they think fast, even if they are experienced, competent and with good intentions, they can make wrong decisions (Kahneman, et al. 2011).

Dual system information processing has been neglected in decision making research in tourism (McCabe et al. 2016), although its understanding provides insights of immense importance to researchers and destination managers (Stylos 2022). Recent research has introduced dual system theory to explain tourist decision making (McCabe et al. 2016;Stylos 2022), while also providing evidence for understanding why it is reasonable to assume that hotel managers also often think fast and react automatically when dealing with online reviews. First, online customer reviews exist in great volumes and growing exponentially (Alaei et al. 2019). Although large hotel chains employ data scientists to analyse their user-generated content, this is beyond the capabilities of smaller properties (Sparks and Bradley 2017). Second, hotel managers' compensation is increasingly tied to online reviews and achieving a minimum rating on a review platform (Gössling et al. 2018). Such incentives guarantee that managers are dedicated to monitoring and managing online reviews but could lead them to process them routinely and become blind to a broader perspective. Third, as it has become industry standard to write an online response to an online review within one to three days, preferably including a reference to the improvement actions taken or planned (Sparks and Bradley 2017), managers evaluate improvement opportunities within the same short time frame.

1.2. Relevant consumer experience scales and marketing tools

To find the answer to the research questions, we needed to identify relevant consumer experience scales and marketing tools. We found thatOh and Kim (2017) argue that customer satisfaction, perceived service quality (PSQ) and customer perceived value (CPV) are key marketing research concepts in business and hospitality research over the last three decades and represent critical indicators of overall customer experience. Due to the nature of the required analysis, we found multidimensional CPV and PSQ concepts suitable for the study.

CPV is a trade-off between total perceived benefits and sacrifices (Ledden et al. 2011;Zeithaml 1988).Sheth et al. (1991) defined it as a concept with five dimensions. These dimensions were later refined and extended to include additional dimensions, often subdivided into give (what a customer invests to receive a service) and get (what a customer receives from a service) dimensions (Zeithaml 1988;Ledden et al. 2011).Franzen and Bouwman (2001) extended the systematisation ofSheth et al. (1991) with that ofLai (1995) and proposed perhaps the most comprehensive model of get dimensions, which include: functional value (perceived utility based on functional and physical characteristics of the product), social value (perceived utility of interactions with social groups related to product consumption), emotional value (perceived utility from the product eliciting specific feelings or affective states), hedonistic value (perceived utility of sensory enjoyment, pleasure, or comfort), aesthetic value (perceived utility from perceiving the product as beautiful and stylish), epistemic value (perceived utility of aroused curiosity and discovered novelty), conditional value (perceived utility due to a specific occasion, such as weddings or anniversaries), and holistic value (perceived utility of the harmony of all previous values). Widely studied give dimensions are money (perceived monetary sacrifices) and time (perceived non-monetary efforts) (Zeithaml 1988;Ledden et al. 2011).

PSQ is a concept related to quality. AsZeithaml (1988) stated, the essence of quality is the superiority or excellence of something, while perceived quality is a higher abstraction and the consumer's judgement of a product or service, which may differ from objective (actual) quality. Numerous studies identified the PSQ dimensions (Tribe and Snaith 1998;Ceylan and Ozcelik 2016;Babić-Hodović et al. 2019), but the most famous are the five SERVQUAL scale dimensions (Parasuraman et al. (1988, 1991), which are still used extensively in hospitality research (e.g.,Bakirtzoglou et al. 2018;Lestari and Saputra 2018;Malik et al. 2020, Sangpikul 2021). They comprise tangibles (physical evidence, including equipment to ensure customer comfort), reliability (delivery of promised service and its correctness), responsiveness (dedication to customers, reaction to customer needs, willingness to help and provide prompt service), assurance (professionalism, having everything under control), and empathy (heartfelt behaviour toward customers, knowledge of customer needs, care, and individualized attention) (Parasuraman et al. 1991).

On the other hand, when it comes to marketing tools that marketing managers are taught to use, none is more common in marketing textbooks than the marketing mix (7Ps) (e.g.,Kotler et al. 2013, Iacobucci 2017;Marshall and Johnston 2019) and hence we find it appropriate to use in this study. 7Ps is an integrated marketing programme (Kotler et al. 2013) designed to deliver value to the customer. It consists of seven elements (Loo and Leung 2018;Ozretić Došen 2010). The first four originally proposed were product (the object of exchange comprising tangible and intangible attributes), price (the money customers pay for acquiring a product), place (the activity that enables delivery of the product to the customer at the appropriate time and place), and promotion (the activity aimed at informing, reminding, and encouraging customers to use the product) (Grbac 2012). The additional three later added elements of 7Ps are: people (knowledge and skills of the staff), physical environment (space and its characteristics experienced through various senses), and processes (background activities of service delivery and value co-creation) (Ozretić Došen 2010).

1.3 Previous research on the thematic scope of online customer reviews

Previous research on the content of online customer reviews has extensively applied data and text mining techniques to define themes, i.e., hotel features that are of most interest to consumers. Rooms, staff, location, breakfast, service, and cleanliness (Li et al. 2015); decoration and design, food, staff, and location (Calheiros et al. 2017); and staff, location, service, and cleanliness (Wang et al. 2020) were found to dominate consumer discourse.Zhou et al. (2014) not only found the most common features, but also divided them into satisfiers (e.g., public facilities), dissatisfiers (e.g., room size, cleanliness, and noise level), bidirectional forces (e.g., amenities in the room/bathroom, quality of food, and friendliness of staff) and neutrals (e.g., Wi-Fi services, entertainment facilities).

More recently, researchers have also been interested in categorising online review themes according to the marketing concepts commonly studied. For example,Sangpikul (2021) categorised the comments according to SERVQUAL scale (PSQ) to find that tangibles and assurance were the most frequently mentioned. On the other hand,Loo and Leung (2018) andKwok et al. (2020) classified the review comments according to the 7Ps. According to the former, half of the issues relate to product, followed by processes and people, while according to the latter, product, physical environment, place, and people are most frequent.

At this point, it is also important to point out that in contrast to the presented inductive content-analysis research of online consumer reviews in which only some dimensions of the PSQ and 7Ps were found to dominate consumer discourse, deductive survey-based research shows that all or most the dimensions of these concepts are equally important to customers. For example,Lestari and Saputra (2018) andMalik et al. (2020) examined PSQ’s role in achieving satisfaction in the hospitality industry. The first study showed that hotel guests rated all dimensions as very important for overall satisfaction, while the second found that three of the five dimensions influence guest satisfaction. Similarly,Rasidah et al. (2017) showed that all the five studied CPV dimensions belong to a higher order concept that influences satisfaction, whileEl-Adly (2019) found that five of the seven studied CPV dimensions influence satisfaction. Although the 7Ps is a marketing tool rather than a measurement scale, past research also used it to measure consumer experience in the hospitality industry. AlAl Muala and Al Qurneh (2012) found that most of the 7Ps dimensions influence tourist satisfaction. Similarly,Harrington et al. (2017) found that the concept consisting of place and process best predicted purchase behaviour across the brands studied, while price and the concept consisting of product, physical environment, and people, also predicted purchase behaviour for some studied brands.

Research has so far not applied a comprehensive approach to understand the thematic scope of textual comments in online customer reviews in relation to a) review rating categories, b) consumer experience scales, and c) marketing tools to discover potential thematic mismatches between them. Discovery of these mismatches and discussion of their implications for hotel managers, researchers, and online platforms presents a contribution of this study to the existing body of knowledge.

2. METHODS

2.1. Data gathering

Since a thorough, manually conducted analysis of online reviews was necessary to achieve the purpose of our study, we decided to focus on one online review platform. The platform had to be globally relevant. According to the report byWohl (2017), most global online reviews (39%) come from Booking.com. Furthermore,Mellinas et al. (2016) describe Booking.com as a useful source of information for consumers, hoteliers, and researchers. Also, Booking.com’s review form requires rating of seven thematically specific categories and provides space for thematically unrestricted comments on positive and negative aspects of the overall stay. For these reasons, it was considered appropriate for the research.

Since the analysis required a contextual understanding of the meanings in the comments, we needed to be familiar with the destination and the hotels under review. We, thus, focused on destination Opatija. It is a year-round business and leisure destination on the Mediterranean, whose tourism dates back to 1844 (Opatija Riviera, n.d.a), among the destinations with the most tourists in Croatia MINT 2018), and with which both researchers are well familiar. Then, we searched for hotels that are:

1) four-star hotels, as they best represent Opatija (Opatija Riviera, n.d.b),

2) large enough to receive a considerable number of reviews, and

3) rated on average between 8 and 9 , to get positive and negative reviews.

FUS NOTA

Using these criteria, we selected two four-star hotels, both of which had an average rating of 8.5 and 100+ rooms. We chose two hotels to ensure that the themes in the comments are not hotel specific. All reviews for the two hotels from February 2015 to February 2017 with comments in Croatian or English were captured.

2.2. Data analysis

There were 287 reviews for the two hotels. Hotel A gave 45 reviews in Croatian and 90 in English and Hotel B 52 reviews in Croatian and 100 in English. Following procedures in similar data mining analyses (Berezina et al. 2016;Li et al. 2015), we split the reviews into single topic comments (hereafter: comments) to analyse them separately (Table 1). Accordingly,

Splitting reviews into comments revealed that most reviews (46%) comprised one to three comments, followed by four to six comments (40%). The rest were longer, with one consisting of 17 comments. In summary, 287 reviews produced 1,175 comments, of which 754 were positive and 421 were negative. Eleven comments addressed matters outside the hotels" control (e.g., "beautiful pathway along the sea") and were not further analysed.

Since our goal was not only to identify the most common themes, but to match each comment with the rating categories and context-rich dimensions of CPV, PSQ, and 7Ps, it was necessary to understand the inherent meaning of the comments. Thus, instead of using text mining software to identify keywords, as most previous research has done (e.g.,Li et al. 2015;Wang et al. 2020;Calheiros et al. 2017), coding was manual, as inLoo and Leung (2018) andSangpikul (2021). First, Booking.com's rating categories, the commonly used dimensions of PSQ and CPV, and the 7Ps elements were described in a codebook consisting of four coding sets.

The first coding set (Table 2) defined seven rating categories taken from the Booking.com review form: staff, facilities, cleanliness, comfort, value for money, location, and free Wi-Fi. As, in relation to RQ1, we aimed to understand whether textual comments cover broader thematic scope than rating categories, we searched for themes that appear in textual comments but not in the existing rating categories. We identified view and food and beverage as such themes and added them as two additional categories to the coding set. This is explained in more detail in Results.

1 added during the analysis

The other three coding sets were not expanded because the goal regarding the RQ2 was to determine if the textual comments were thematically narrower than the consumer experience scales (PSQ and CPV) and the marketing tools (7Ps). The coding scheme for CPV followed the systematisation and conceptualisation ofFranzen and Bouwman (2001) for the get andLedden et al. (2011) for the give dimensions (Table 3).

For PSQ, we relied on the conceptualisation and operationalisation of the often-cited SERVQUAL model ofParasuraman et al. (1988, 1991) to produce the coding scheme presented in Table 4.

Finally, for the operationalisation of the 7Ps (Table 5), we followed the definitions ofOzretić Došen (2010) andGrbac (2012).

To code the comments, we first recognised twenty hotel features (e.g., Room, Restaurant, Staff), then assigned each comment to a feature, and then coded feature by feature. To determine the codes, we first defined the general rule for each hotel feature (the primary codes). For example, the primary codes for the Restaurant were facilities within the rating categories, functional and hedonistic values and so on. All comments assigned to a hotel feature were coded with the associated primary codes. Then, within each hotel feature we distinguished three types of comments. The first type were short and simple comments (e.g., "no parking," "small room") for which coding ended once the primary codes were assigned. The second type were more detailed comments that converged on one or two themes within each hotel feature (e.g., parking-related comments were often about cost, room-related comments were about cleanliness). We then defined the rules (the secondary codes) for these themes and applied them consistently to all comments of this type. The third type of comments were complex and unique. Unlike the previous two types of comments, this type was the least frequent but required the most effort to resolve. Identifying relevant codes for each complex comment required ongoing discussions among authors, consultations with the field experts, and the application of a four-step procedure:

3. RESULTS

Within the rating categories (Figure 1), most comments related to comfort, emphasising issues such as "room too small" and "carpet old and dirty." More specifically, 325 comments (28%) in 176 reviews (61%) related to comfort. The second most frequently mentioned category was facilities (e.g., "the gym is poor" and "the outdoor pool is very nice") in 151 reviews (53%) and 219 comments (19%). Four hotel features did not fit into the existing rating categories. The first two, Hotel overall (e.g., "a fabulous hotel") and Overall experience (e.g., "stay was pleasant overall"), we thought reflect in all the rating categories, so we consider them covered by the existing rating categories. For the two remaining features, View and Food and beverage, after careful consideration of their relationship to the existing rating categories, we have concluded that they do not fit within any. Therefore, we added them as two new rating categories in the coding scheme (Table 2) and highlighted them in Figure 1. Food and beverage, the first added category, was mentioned third most often and occurred in 176 comments (15%) and 152 reviews (53%). View, the second added category, was mentioned in 7% of comments and 28% of reviews - more frequently than several existing rating categories, particularly free Wi-Fi, which was mentioned in only 2% of comments and 8% of reviews.

When comments are evaluated against CPV dimensions (Figure 2), most relate to functional value (e.g., "no parking" and "the location is fantastic"). Specifically, 277 reviews (97%) had 1,066 comments (91%) related to functional value. Hedonistic value with comments like "the breakfast room has a nice terrace with a sea view" and "the food was very varied and tasty" followed with 246 reviews (86%) and 744 comments (63%). All other values were far less represented. Emotional, social, and conditional values were found in less than 10% of the reviews and less than 3% of the comments.

Categorizing comments into PSQ dimensions (Figure 3) shows that tangibles are by far the most commented on, with comments such as "the shower is very small" and "the gym is poor." Accordingly, 245 reviews (85%) contained 752 comments (64%) on tangibles, followed by assurance (e.g., "extremely kind staff" and "poorly trained reception staff") with 132 comments (11%) in 126 reviews (44%). The other three categories were mentioned in less than 10% of the comments and less than 20% of the reviews.

Finally, categorization of comments according to the 7Ps (Figure 4) shows that most reviews focused on product and physical environment, with most comments falling under both categories simultaneously (e.g., "comfortable bed" and "cold water in swimming pool"). Specifically, 263 reviews (92%) with 804 comments (68%) were related to product, while 233 reviews (81%) with 577 comments (49%) to physical environment. The other 7Ps elements were far less frequent, promotion almost absent.

4. DISCUSSION

4.1. The textual comments are thematically broader than the rating categories

We found many comments related to food and beverage and view that did not refer to the existing Booking.com rating categories, and we conclude that at least these categories are missing from the rating categories. At the same time, few comments referred to free Wi-Fi, an existing rating category. We compared our results with previous research and found that food and beverage is one of the most commented categories in similar qualitative studies (Sparks and Browning 2010;Li et al. 2015;Calheiros et al. 2017). Moreover, food and beverage contributes strongly to the overall travel experience (Beltrán et al. 2016) and is simultaneously satisfier and dissatisfier, unlike most hotel characteristics (Zhou et al. 2014). When it comes to view,Li et al. (2015) considered it an important emerging category, even though not yet among the top five. Along these lines,Fleischer (2012) found that rooms in the Mediterranean (the region of our study) are 10% more expensive if they offer a sea view. Finally, free Wi-Fi is not popular across comparative studies (Sparks and Browning 2010;Li et al. 2015;Calheiros et al. 2017). While Mellinas and Nicolau (2020) consider it as a Herzberg's hygiene factor that would not be mentioned if it is at the expected level or above,Zhou et al. (2014) found that free Wi-Fi is not even a hygiene factor, but rather irrelevant to satisfaction and dissatisfaction. The fact that Booking.com omits food and beverage and view while including free Wi-Fi is an attempt to provide consistent rating categories across regions and hotel types. Food and beverage is not offered at all listed hotels, and view is not relevant to all hotels, but, as our results show, neither is free Wi-Fi. Achieving uniformity and maintaining universal relevance in online review forms is challenging. However, omitting relevant categories has important implications, as we elaborate in Conclusion.

4.2 The textual comments are thematically narrower than the consumer experience scales and marketing tools

5. CONCLUSION

5.1. Theoretical implications

Although 7Ps, and particularly PSQ and CPV have been clearly operationalised previously (e.g., Sheth et al 1991; Zeithaml 1988;Kotler et al 2013), existing instruments (e.g., Parasuraman et al 1988, 1991;Ledden et al 2011) focused on the quantitative scales. Our research provides instruments (coding schemes) for qualitative research specific to the hospitality industry. Similarly, we propose a coding scheme for classifying comments into review rating categories. These instruments will allow researchers to replicate and further extend our research to better understand what hotel guests care most about and how it relates to their satisfaction and hotel ratings.

Further, we found a thematic mismatch between widely used consumer experience scales and online comments. Although previous deduction-based research (e.g., Lestari and Saputra 2018;Rasidah et al. 2017) finds that most or all dimensions of existing CPV and PSQ scales relate to consumer satisfaction, the deductive approach lacks information about which dimensions consumers pay most attention to when thinking about the overall hotel experience. Analysis of online comments is better suited for this purpose. The thematic mismatch we discovered suggests that existing scales could benefit from being refined in the dimensions that attract the most consumer attention and reduced in other dimensions. That is, a PSQ scale for the hospitality industry might have better construct validity if it included two or more tangibles-related dimensions and possibly excluded empathy. A HOLSAT scale (Tribe and Snaith 1998;Ceylan and Ozcelik 2016) might therefore be a suitable alternative to the still commonly used SERVQUAL. Similarly, in the hospitality industry, the CPV scale could better capture customers" experiences and provide more accurate insights if it focuses on and expands functional and hedonistic values and possibly reduces reference to social and conditional values.

Finally, by relying on dual system theory (see Kahneman 2013) we explain why we believe the identified thematical mismatches may pose problems to hotel managers. Circumstances lead managers to closely monitor online reviews (cf. Alaei et al. 2019;Gössling et al. 2018;Sparks and Bradley 2017), but these same circumstances lead them to think fast, potentially blinding them to a broader perspective and different marketing tools available. Researchers should apply this knowledge when exploring the usefulness of online reviews as a source of ideas for improvement investments and when exploring how prioritising online reviews in hotel managers' compensation packages affects hotels' long-term benefits.

5.2. Managerial implications

Online reviews are a useful source of information that managers can use to improve the overall satisfaction of future guests (Ho 2017), future online ratings, and future online comments (Gössling et al. 2018). However, because hotel managers are often under pressure from a huge and increasing number of online reviews (Alaei et al. 2019), in addition to other responsibilities, they tend to process online reviews fast and automatically rely on them. This makes them vulnerable to falling into the traps described below.

First, when managers improve aspects that are mentioned in review comments but fall outside the existing rating categories on review forms (e.g., food and beverage), the impact of the improvements made may not be reflected as strongly in future ratings as improvements in aspects that fall into a rating category (e.g., comfort or staff). On the other hand, the former (e.g., food and beverage) might improve overall satisfaction more than the latter. Ideally, managers should improve both types of issues, but resources are scarce. Therefore, managers should be cautious, take a critical stance, and periodically carefully examine online reviews (i.e., think slowly) to understand which improvements will best translate into the desired benefits (better future overall guest satisfaction, higher future online ratings, and better future online comments). Managers should be aware that the issues that need the most attention or the most urgent action are not necessarily the ones that are most frequently mentioned in consumer review comments.

Second, most consumer comments relate only to some aspects of the consumer experience scales (i.e., functional and hedonistic values or tangibles). This means that to better understand consumers, managers should not rely only on review comments, but a combination of quantitative, survey-based data and qualitative, review-based data. At the same time, most consumer comments relate only to some marketing tools (i.e., product and physical environment). That means that fast-thinking hotel managers who focus their attention on online reviews may get constrained within the narrow customer perspective when considering hotel improvements. In fact,Sparks and Bradley (2017) found that when hotel managers explain an action undertaken related to a consumer complaint in online review responses,they mostly refer to refurbishments (product and physical environment), while very rarely mentioning staff training or process changes. It seems that narrow perspective of online comments prevents managers from considering all the marketing tools (7Ps) at their disposal to make improvements that increase long-term benefits for customers and the hotel. We therefore urge managers to think slowly and seek creative solutions outside customer perspective. For example, when confronted with comments about lack of parking, they should consider not only the feasibility of building a garage, but also solutions such as shuttle transfers to a remote parking lot, staff assistance with parking, promoting the hotel as a no parking property for guests arriving by charter buses, or other solutions that employ different elements of 7Ps to go beyond product and physical environment. A hotel culture that encourages a creative and thorough search for solutions (slow thinking) is also better prepared for unforeseen situations that require immediate responses.

Our findings may also be useful to Booking.com managers. That is, if guests find certain themes important to comment on in online reviews, we can assume that they will also find those themes important when looking for their next hotel. If rating categories do not cover important themes, the hotel search will be less effective for future guests, and they may turn away from the platform when new, potentially better-performing ones enter the market. Since it is challenging to choose rating categories that are relevant to all hotel types and all guests, it might make sense not to have universal rating categories. The platform could require the reviewers to evaluate more than the seven rating categories currently available and allow each review reader to decide which categories will be included to create a customised overall rating. Basic categories (e.g., comfort, facilities, and staff) could always be included in the overall rating, while others (e.g., view, free Wi-Fi) could remain discretionary.

5.3. Limitations and directions for future research

Since we studied leisure hotels in the Mediterranean region, the results are arguably determined by the region, especially regarding the importance of the view. Future studies may examine other regions and other types of hotels to discover other relevant rating categories. We also focus our analysis on a specific platform. Analysis of other platforms that offer differently structured review forms may help to understand how a review form influences comments, future rating improvements, and overall hotel performance. Even though the data was collected several years ago, the study's conclusions are still valid because the concepts studied have remained unchanged in the meantime. Also, our intent was not to provide an exhaustive list of changes needed in terms of rating categories or marketing scales, but to uncover the mismatches. Nevertheless, more recent data may provide further ideas for refining the scales and rating categories. Finally, interviews with hotel managers should provide insights into different approaches to reacting to online reviews.