Switzerland? The Best Choice for Accommodation in Europe for Skiing in the 2023 Season

Abstract

The study investigates the connections between tourists‘ hotel preferences and distance from resort expressed in meters, distance from ski lift expressed in meters, booking score, number of reviews, room type, feature of free cancellation, price expressed in Euro, type (private host/hotel), destination (ski-to-door access/travel sustainable property) from 10 highly appreciated European ski locations offered in from February 2023 by Booking, using a sustainable, electronic instrument for collecting and analyzing a large amount of data, Octoparse 8 and a multi-attribute decision model. Previous studies concerning tourist preferences and online behavior used traditional methods, such as questionnaires and surveys, being limited to a certain number of questions and respondents; thus, this study covers a research gap in the literature with regard to the use of a large amount of data, electronic instruments and multi-attribute models to rank the hotel locations, which derives from the difficulty in obtaining the necessary data to carry out an in-depth analysis. The results show that when selecting a hotel location from an exclusive ski resort, the tourists are interested in the number of reviews, the price and the distances from the resorts to the ski slopes, while the booking score is less important. This is translated into practical implications for marketers and hotel managers, presented at the end of the paper.

1. Introduction

The ski industry and winter tourism are very important economic development factors in the European Alps. They are discussion topics for both researchers and practitioners as they face challenges, such as climate change and changes in holiday preferences for other types of tourism, including city breaks and also, after the pandemic period, the custom of online reservation as a good practice. When facing these challenges, the practitioners are more and more interested in developing marketing models to attract more clients and also researchers are more and more interested in determining models of customer behavior when choosing the destination online. Even more, when talking about highly appreciated European touristic locations for ski, the interest is even higher.

The present study was applied to 10 highly appreciated European tourist locations for ski from France, Austria, Italy and Switzerland, offered by Booking for February 2023. They were chosen because they are considered [1] an example of the world’s biggest ski destinations and have a large number of facilities, and are highly appreciated.

There are studies examining the attributes prioritized by tourists and the factors that influence customers’ attitudes and purchasing behavior on online purchasing when choosing a winter sports destination (Liu, Dong, 2013 [2]; Lu et al. 2018 [3]; Miragaia, Martins, 2015 [4]; Mlađenović, Jovanović, 2019 [5]) but they are limited to some key factors, such as information, price and convenience, and they are focusing only on few food types and specific groups of tourists or locations.

Still, in terms of macro-correlations, between tourists’ preferences and distance from resort in meters, distance from ski lift in meters, booking score, number of reviews, room type, feature of free cancellation, price in Euro, type (private host/hotel), destination (ski-to-door access/travel sustainable property), the research remains in its infancy.

Additionally, the studies do not provide a comprehensive understanding of correlations among various large populations of tourists and their preferences are based on traditional and inefficient instruments, such as surveys and questionnaires.

The main objective of this paper is to use an innovative electronic instrument—an application—Octoparse 8.5.5, to extract a large amount of data from Booking.com, and then, to rank the hotel locations according to the data collected, using a multi-attribute decision model.

Consequently, the present work covers an important gap in the literature with regard to the use of a large amount of data, electronic instruments and multi-attribute models to rank the hotel locations, which derives from the difficulty in obtaining the necessary data to carry out an in-depth analysis. To this end, initially, we analyzed 368 records, obtained by eliminating incomplete data and redundant locations, from 10 highly appreciated European ski destinations from France, Austria, Italy and Switzerland in order to explore tourists’ preferences.

In the beginning, the authors provide the conceptual background, consisting of relevant theories on framework and criteria characterizing the best hotel location offers. Next, the authors discuss the instruments of the study, namely the multi-attribute decision model and the two normalization matrix methods. Then, the authors provide the research design, including data collection and data processing. Furthermore, the authors provide the findings and discussions. Finally, the authors provide the conclusion, in which the theoretical and managerial contributions are presented, as well as the research limitations and future directions.

2. Theoretical Background

2.1. Framework

In the 19th century, all winter sports promoted ski resort development [1,6]. It is a typical industry characterized by seasonality [1,7], which suffered a substantial loss in 2020 because of COVID-19 [8]. Nevertheless, the European Alps and Dolomites winter ski resorts are known worldwide, not only for their great skiing, but for the quality of the whole experience for skiers. The mountains of France, Italy, Switzerland, Austria and Germany [9,10] may be used by all categories of skiers, not only by the experienced ones. Most resorts have easy and intermediate terrain, and many have dedicated slopes and lifts just for learners and beginners. Extraordinary seems an insufficient description of the incomparable views of snowcapped peaks. Each resort has its own character and style, and the lift systems make possible a single trip—often one single day between several mountains.

More than one third of all ski resorts are located in the Alps, and 80% of major resorts (over 1 million skier visits per winter season) worldwide are located there. Most of the industry is concentrated around the resorts that generate more than 100.000 skier visits per year and even if they only account for 20% of the resorts, they account for 80% of all skier visits [1].

On the other hand, internet has become an important distribution channel in the hotel industry [11,12]. In this context, online hotel reservation offers benefits to consumers, such as accessing photos, videos, a full description of the hotel property and location, better pricing and no additional booking fees [13,14]. Considering the convenience and cost/time saving, there is an increasing preference to use the internet to search for information regarding brand, price, service and to book hotels [15]. Thus, many hotels have noted this online trend and have provided access to secure online reservation systems. So, the use of the internet has not only changed the methods of how tourism products and services are distributed [16,17], but it has also changed tourists’ online behaviors [18].

As a consequence, all ski resorts hotels need to focus on providing online and also offline positive experiences, considering service quality, time, transportation, local development, natural resources, climate, economy and subjective experience to be memorable [19]. Once they are memorable, with no exception, they become commodities through the experience with the overall product or service being purchased [20,21,22], even more relevant when one considers the services (e.g., rooms). Thus, hotels need to stage experiences deliberately, and online booking systems and direct booking system hotels, are no exception.

The direct hotel online booking system has grown lately, and consequently, researchers [23] have examined the online hotel-booking behaviors considering channel characteristics [24], online information characteristics—content and types of information [25,26,27] consumer characteristics—demographics, branding and loyalty [28,29]. Researchers have extensively investigated also the customer engagement behavior [30], determinants of advanced booking [31], choice attributes [32] and the decision-making process [33].

Accordingly, the interest in the context of hotel choice has remained steadfast for its apparent importance to business success. Moreover, most studies focused mainly on buyers’ behavioral intentions, not on hotel owners’ holistic viewpoint meant to improve the online experience, hotel choice and increased intention to buy.

Ski destination choice criteria is a topic of research, and there are also, to some extent, some results showing that there are different segments of tourists [34], who are compared in order to ascertain possible differences in personal (gender and age) and situation-specific (type of visitor and traveling companion) characteristics between customer segments. There are studies using data collection from visitors of different ski resorts in Lapland, Finland during 2006 and 2007 identifying six different customer segments using the factor-cluster method: passive tourists, cross country skiers, want-it-all, all-but-downhill skiing, sports seekers and relaxation seekers.

Additionally, whilst there are studies focusing on the correlation between hotel reviews and pricing [35], there is scant attention paid to the impact of pricing on hotel review ratings, and there were research studies conducted based on a novel dataset of more than 44,000 guest review ratings linked to the prices paid for the rooms of a European hotel group [36]. Through applying a panel regression analysis, it was revealed that expectancy–disconfirmation is generally stronger than the placebo effect, and higher prices have a negative effect on review ratings, not only when assessing the perceived value for money but also when evaluating perceived quality.

Moreover, few studies investigate the effect of satisfaction on spending and even fewer relate this to visitor expectations, such as [37], which examines the case of the Falun, Sweden World Ski Championships 2015. A particular focus of the study is if and how visitor satisfaction influences visitors’ expenditures, and it was hypothesized and argued that spending depends on the satisfaction related to prior expectations.

The study is an early attempt to examine, using a large amount of data and a multi-attribute method, tourists’ hotel preferences from exclusive ski resorts from Europe. The study uses 10 highly appreciated European ski destinations from France, Austria, Italy and Switzerland in order to explore tourists’ preferences, and covers a gap in knowledge since previous studies used questionnaires and surveys to explore the hotel choices of tourists considering price, convenience, gender, age and type of visitors (Liu, Dong, 2013 [2]; Lu et al. 2018 [3]; Miragaia, Martins, 2015 [4]; Mlađenović, Jovanović, 2019 [5]; [30,31,32,33,34,35,36,37]) but only briefly explored the correlation between preferences, scores and prices from online booking (Zhao et al., 2015 [38]; Yang, K. 2005 [39]; Wu, Wang, 2005 [40]; Emam et al., 2021 [41]). Additionally, the study covers a knowledge gap and a relatively unexplored research area; it is an effort to expand the tourism literature.

In this paper, the researchers studied tourists’ food preferences for hotels from exclusive ski destinations from four European Alps countries, using a large amount of data gathered with an electronic application and a multi-attribute decision model. The findings showed that tourists are interested in the number of reviews, the price and the distances from the resorts to the ski slopes, while the booking score is less important.

The proposed method will help researchers and practitioners have insights of tourists’ hotel preferences by the use of booking data as an alternative information source to traditional surveys and questionnaires used to find out the tourists’ preferences.

2.2. Hypothesis

Rather than analyzing consumers’ perceptions, intentions and satisfaction, it is essential to understand the factors influencing the tourists’ online choices by analyzing the importance of the distance from the resorts to the ski slope, the booking score, the reviews and the price and the correlations between them, when decisions are made by the tourists.

To address this gap, this study investigates how these factors influence the tourists’ decision, using a multi-attribute decision model and two normalization methods. The instruments enable marketers to understand consumer choice behavior based on the factors of decision processes, as they can evaluate the preferences and alternative sets. This technique provides estimates of the importance and influence of different variable choices across different hotel types (e.g., hotel location or privately managed host unit).

2.2.1. Hotel Location Choices

In reference [14], an experimental design to investigate four independent variables was conducted: the review target, overall valence of a set of reviews, review framing of and whether a consumer’s generated numerical rating is provided together with the written text. Consumers seem to be more influenced by early negative information, especially when the overall set of reviews is negative. However, positively framed information together with numerical rating details increases both booking intentions and consumer trust.

Additionally, the impact of the content analysis on the star rating given to hotels was examined for all the hotels in Istanbul that have received a maximum of 25 reviews on the TripAdvisor website, with a total of 12,000 comments assessed, the results showing that the hotel location, the access to transport facilities, the food and beverage concept, the quality of staff/service and the cleanliness of the facilities all affected the star ratings given to hotels [14].

Moreover, searching for information for hotel booking has become a dispensable step in travelers’ decision-making process [42,43,44], as consumers tend to rely and influence each other on information about hotel products and services provided by fellow customers [45]. This indicates the power and persuasiveness of online product reviews [43,46] and has shown that consumers tend to rely more on peer reviews than on information provided by business entities because peer customers are more independent and trustworthy [47]. Furthermore, consumers are believed to have no vested patterns when posting a review online, and there is no structured pattern for them to post their experiences on the Web [48].

There are studies exploring the influence of different factors on the online decision of the tourists, such as age, income, education level, cultural background, positive/negative information, e-WOM, prices or scores ([14,43,46], Zhao et al., 2015 [38]; Yang, K., 2005 [39]; Wu, Wang, 2005 [40]; Emam et al., 2021 [41]) have been shown to influence consumer behavior, but still there is not a general opinion among researchers that they are the only factors used to choose an online hotel. Considering the previous findings, we considered the following hypothesis.

H1. 

Tourists of exclusive ski resorts do not always choose hotels by online scores and prices.

2.2.2. Online Scores and Reviews

According to the data available, 81% of travelers frequently or always read reviews before booking a hotel [49], 52% would never book a hotel that had zero reviews [49] and 55% read several pages worth of reviews to obtain a better sense of public opinion [49]. Additionally, 66% of travelers plan to spend more time reading reviews about destinations [50], 65% are inspired by online searches (Google) and 72% frequently or always read reviews before deciding where to stay or what to do [50]. When deciding between two similar properties, 79% of travelers are more likely to reserve a room at a higher rated hotel [49]. Moreover, travelers will value guest ratings over a hotel’s brand 72% of the time [51], over 40% of travelers will leave a review if they have a positive experience at a hotel [52]. Additionally, 48% of travelers will leave a review after a bad hotel experience [52], travelers read an average of nine reviews before deciding to book a hotel [49] and 91% of travelers want property owners to respond to negative reviews [53].

The research shows that online recommendation systems offered by online retailers are more influential than the recommendation from consumers [54] but the results are moderated when considering the influence on buying decisions regarding hotel stays [55,56,57]. Reviews also act as anchors of consumer experience and encourage subsequent reviews, as shown in [58]. Moreover, the results of some studies [59] revealed that positive feedback reviews do not affect booking intention, while negative feedback reviews have a strong impact.

There are studies exploring the type of behavior factors for the online decision of the tourists, such as expectations and scores (Bandara, Silva, 2016 [60]; Jeon, Jeong, 2016 [61]; Cho, Sagynov, 2015 [62]), but among researchers there is not a general opinion that the price is the only factor influencing decisions about a hotel or a privately managed host, but also scores are a subject of discussion.

Considering the above, we formulated a second hypothesis.

H2. 

Tourists pay attention to the score when choosing a hotel or a privately managed host unit.

3. Method

In order to decide upon the best hotel location offers for ski in February 2023, all touristic offers from Booking.com in 10 highly appreciated European locations were analyzed [10].

As some of these touristic resorts are very close to each other—a maximum of a few kilometers distance—for a certain resort there were recommended closest touristic resort hotel locations. In this situation doubled data of hotel locations were eliminated, maintaining the closest locations to ski slope hotel location.

3.1. Data Collection

The 10 ski resorts were chosen using an online ranking list [10]. There are many online rankings of ski resorts, but the authors’ choice was made based on the popularity of the website mentioned. Even so, most of these rankings are determined by users’ personal experiences of web-based platforms and there are no precise classification criteria. Based on the data previously mentioned, the analyzed resorts were: 1. Val d’Isère, France; 2. Tignes, France; 3. St. Anton, Austria; 4. Courchevel, France; 5. Zermatt, Switzerland; 6. Méribel, France; 7. Lech, Austria; 8. Courmayeur, Italy; 9. Verbier, Switzerland; 10. Cortina, Italy.

To choose the accommodation period, another website [63,64] was used. Consequently, in order to gather information from Booking.com, a reservation from 20 February to 26 February for two adults was simulated—which is the most frequent situation. The reservation was simulated in 15 October 2022. The authors consider this date as relevant because the hotel location offer was constantly being modified and different prices, services provided and availability degrees were recorded.

The Octoparse v.8.5.5 application was used to retrieve accommodation data from each location. It allowed extracting the following data: resort name, city, distance from resort in meters, distance from ski lift in meters, booking score, number of reviews, room type, feature of free cancellation, price in euros, type (private host/hotel), destination (ski-to-door access/travel sustainable property).

We have used the following algorithm to automatically extract all data regarding locations from analyzed resorts. This algorithm (Figure 1) may be used in any Octoparse application, such as for automatically data extraction from Booking.com.

For a more accurate image, we did not use sampling nor did we choose any representative number of hotel units, but we have analyzed all the units.

The algorithm steps are:

Step 1. 

Loop Booking.com hotels

Loop mode: List of URLs—Manual input

Loop items: list of Booking.com resorts

Step 2. 

Go to Webpage

Load URLs in the loop

Step 3. 

Loop page

Loop mode: Single Element

Repeats 1000 times.

Step 4. 

Scroll page

Loop mode: scroll page

Scroll: to the bottom of the page

Step 5. 

Loop hotel

Loop mode: Variable List

Step 6. 

Extract General Data: hotel name, city, distance from resort in meters, distance from ski lift in meters, booking score, number of reviews, room type

Extract data in the loop

Step 7. 

Click hotel

Open in a new tab

Step 8. 

Extract Detailed Data: feature of free cancellation, price in Euros, type (private host/hotel), destination (ski-to-door access/travel sustainable property)

Step 9. 

Back to Previous Page

End loop (Step 5)when there’s no more content to load.

End loop (Step 4)when there’s no more content to load.

Step 10. 

Click on a “Load More” button

Load with Ajax

End loop (Step 3)when there’s no more content to load.

End loop (Step 1)when there’s no more content to load.

Initially, data from 845 records—hotel location—10 resorts were collected. Because the multi-attribute decision models used needed some mandatory data, incomplete location data were eliminated, thus using 553 records. After eliminating doubled hotel location—the same location being the search results for 2 or even 3 close resorts—there remained 368 records to be used for the analysis.

3.2. Data Analysis

3.2.1. Multi-Attribute Decision Model

The multi-attribute decision model was employed to rank the hotel locations according to the data collected. The multi-attribute decision model employs a limited (discrete) number of alternatives (variants). They are used to choose an optimal alternative (variant) from a finite number of alternatives compared to each other based on different criteria (attributes) or rank all the variants based on the criteria considered. Each alternative is characterized in relation to each quantitative attribute (numerical) and qualitative attribute (non-numerical). Each attribute has a certain scope—maximum or minimum. The attributes may have different importance for the decider.

A multi-attribute decision problem has the following elements:

$A=\left\{A_1,A_2,\dots ,A_m\right\}$—the set of variants (decision alternatives);

$C=\left\{C_1,C_2,\dots ,C_n\right\}$—the set of criteria (attributes of characteristics);

$A=\left\{a_{ij}\right\}, i=1,2,\dots ,m, j=1,2,\dots ,n$—the set of consequences ($a_{ij}$ is the value of $A_i$ variant $C_j$ criterion).

When all the criteria are equally important, the weight of each criterion will be evaluated. In decisional processes, the criterion may vary in terms of weights for a decider. The importance of the criteria may be different in time for the same problem of the same decider. Apart from the subjective weights given to the criteria by the decider, an objective weight of the criteria may be calculated considering the values of the alternatives for each criterion. This objective weight is determined by the information contained by alternative values.

3.2.2. Normalization of the Consequences Matrix

The values of quantitative characteristics (the distance from the hotel location to the ski slope, the price of a hotel location room and the rating of the hotel location) from the consequences matrix are generally expressed using different measurement units. For instance, in this study, the distance is expressed in meters, the price in euros and the rating as an average of all tourists’ grades. They have to be homogenous to make the comparison possible. Another important aspect is the type of characteristics, since some of them are maximum and others are minimum.

The process of homogenization is called normalization, and it is conducted by creating a correspondence between the set of criteria values with the set [0, 1] and ensuring a unique sense of the values for different criteria. Thus, the maximum values are also the best. For qualitative attributes, the function of membership to the fuzzy subset defined by that attribute may be considered.

The normalization operation means defining a function that transforms the A matrix of consequences into the R-normalized matrix with the elements between 0 and 1:

$$A\to R, r_{ij}\in \left[0,1\right], i=1,2,\dots ,m, j=1,2,\dots ,n.$$

There are many ways to normalize a matrix. Normalization through linear interpolation consists of two transformations that are applied to obtain an R-normalized matrix:

For the criteria expressing the maximum:

$$r_{ij}=\frac{a_{ij}-a_j^{min}}{a_j^{max}-a_j^{min}}, i=1,2,\dots , m, j=1,2,\dots ,n$$

(1)

For the criteria expressing the minimum:

$$r_{ij}=\frac{a_j^{max}-a_{ij}}{a_j^{max}-a_j^{min}}, i=1,2,\dots , m, j=1,2,\dots ,n$$

(2)

where $a_j^{min}=\underset{i}{min}{a}_{ij},$ and $a_j^{max}=\underset{i}{max}{a}_{ij}$

Reference [56] proposed a simple scale to normalize a matrix, using the relations (3) and (4):

For the criteria expressing the maximum:

$$r_{ij}=\frac{a_{ij}}{a_j^{max}}, i=1,2,\dots , m, j=1,2,\dots ,n$$

(3)

For the criteria expressing the minimum:

$$r_{ij}=\frac{a_j^{min}}{a_{ij}}, i=1,2,\dots , m, j=1,2,\dots ,n$$

(4)

The authors of [65] present a multicriteria decision model based on the additive weighted method, through which the alternatives evaluation is made by a using preference selection index (PSI). The model consists of the completion of the following steps

Step 1. The preference variation value (PV) is calculated for each studied characteristic.

$$P{V}_j={\displaystyle \sum }_{i=1}^m{\left[r_{ij}-\overline{R_j}\right]}^2, j=1,2,\dots ,n$$

(5)

where $\overline{ R_j}=\frac{{{\displaystyle \sum }}_{i=1}^m{r}_{ij}}{m}, j=1,2,\dots ,n$

Step 2. The deviation in preference value $\left({\theta }_j\right)$ is calculated for each characteristic.

$${\theta }_j=\left|1-P{V}_j\right|, j=1,2,\dots ,n$$

(6)

Step 3. The overall preference value $\left({\omega }_j\right)$ is calculated by each characteristic, meaning the actual values of objective weights.

$${\omega }_j=\frac{{\theta }_j}{{{\displaystyle \sum }}_1^n{\theta }_j}, j=1,2,\dots ,n$$

(7)

Obviously ${\displaystyle \sum }_j^n{\omega }_j=1$.

As an observation, the subjective weights determined by the tourist who decide may also interfere in the decision process. These subjective weights are noted $\left({\phi }_1,{\phi }_2,\dots , {\phi }_n\right)$, where ${\displaystyle \sum }_j^n{\phi }_j=\mathrm{1.}$ For instance, a tourist may consider the following two characteristics of outmost importance—the distance from the ski slope and the location price—grading, consequently, the two relevant characteristics. Thus, the tourist weights these characteristics with 0.5 and 0.3; thus, the sum of weights for each criterion will be 0.2.

Thus, one can aggregate the subjective weights with the objective weights as follows

$$w_j=\frac{{\phi }_j·{\omega }_j}{{{\displaystyle \sum }}_1^n{\phi }_j·{\omega }_j}, j=1,2,\dots ,n$$

(8)

Step 4. In the end, preference selection index $\left(PS{I}_i\right)$ is calculated for each decisional alternative $A_1,A_2,\dots ,A_m$ if the objective weights are considered.

$$PS{I}_i={\displaystyle \sum }_j^n{r}_{ij}·{\omega }_j, i=1,2,\dots ,m$$

(9)

In the case of aggregated objective and subjective weights, relation (9) becomes (10):

$$PS{I}_i={\displaystyle \sum }_j^n{r}_{ij}·w_j, i=1,2,\dots ,m$$

(10)

Considering the obtained values for each alternative, the preference selection index $\left(PS{I}_i\right)$ allows the descending ranking of the calculated values of variants or selects the maximum value variant (the optimum alternative). For each alternative, a value (score) as a weighted average of characteristics (attributes) consequences with weights associated with each criterion is calculated. The method is compensatory because in the calculus of values associated with each variant, a low value of a criterion may be compensated with higher values of other criteria.

4. Results

4.1. Data Analysed

The criteria used for the analysis (for which the authors had enough data) were:

C1—distance from resort in meters

C2—distance from ski lift in meters

C3—booking score

C4—number of reviews

C5—price in euros

We have normalized the matrix using the previously presented methods: through linear interpolation and through simple scale as in the model in [65].

Due to the results in

Table 1. The results obtained through normalization through simple scaling.

	C1	C2	C3	C4	C5
Rj	0.59	0.74	0.76	0.10	0.89
PVj	40.38	24.52	7.74	7.86	5.04
${\theta }_j$	39.38	23.52	6.74	6.86	4.04
${\omega }_j$	0.49	0.29	0.08	0.09	0.05

for our data, ${\omega }_j$ in case of simple scale normalization has anormal values—for instance, the weight level for price is 0.05, meaning that the price has a very small influence—we considered the results obtained by linear interpolation in

Table 2. The results obtained after normalization through linear interpolation.

	C1	C2	C3	C4	C5
Rj	0.04	0.08	0.85	0.10	0.20
PVj	4.19	4.26	2.79	7.85	7.22
${\theta }_j$	3.19	3.26	1.79	6.85	6.22
${\omega }_j$	0.15	0.15	0.08	0.32	0.29

. We have to mention that these results are very strongly connected to the data analyzed. For instance, regarding the price, we have very large differences between the minimum price (EUR 287) and the maximum price (EUR 20.679). For other analyzed values, it is possible to obtain valid results. Thus, for this analysis, we used as weight for the five criteria, the following values in

Table 3. Weighted values for the 5 criteria used in ranking hotel locations.

	C1	C2	C3	C4	C5
${\omega }_j$	0.15	0.15	0.08	0.32	0.29

:

We consider that the results in Table 3 are relevant because they are of great importance for the number of reviews (0.32), for the price (0.29), for the distances from resorts and the ski slope (0.015), and less importance for the booking score (0.08).

Because we consider as essential the first 25 location hotel positions (top 25), in

Table 4. Top 25 hotel locations using multi-attribute analysis.

Place No	Country	Resort	Numele Locatiei Hoteliere	PSi
1	France	Tignes, France	Base Camp Lodge Hotels	0.47
2	Switzerland	Zermatt, Switzerland	Backstage Boutique SPA Hotel	0.47
3	Switzerland	Zermatt, Switzerland	Hotel Bahnhof	0.46
4	Switzerland	Zermatt, Switzerland	Alpenhotel Fleurs de Zermatt	0.40
5	Switzerland	Zermatt, Switzerland	Carina—Design&Lifestyle hotel	0.40
6	Switzerland	Zermatt, Switzerland	Hotel Alpenroyal	0.37
7	France	Méribel, France	alpes studio	0.37
8	Switzerland	Zermatt, Switzerland	Hotel Continental	0.34
9	Switzerland	Verbier, Switzerland	Hôtel Chalet Royal—Like at Home	0.33
10	Switzerland	Zermatt, Switzerland	Hotel Pollux	0.33
11	Switzerland	Zermatt, Switzerland	Alpen Resort Hotel	0.33
12	France	Tignes, France	studio Les Glieres	0.33
13	Italy	Courmayeur, Italy	iH Hotels Courmayeur Mont Blanc	0.33
14	Austria	Lech, Austria	Revier Mountain Lodge Montafon	0.32
15	Austria	Lech, Austria	TUI BLUE Montafon	0.31
16	France	Méribel, France	Chalet-Hôtel Le Belvédère	0.31
17	Switzerland	Zermatt, Switzerland	BEAUSiTE Zermatt	0.30
18	Switzerland	Verbier, Switzerland	Hôtel Magrappé—Like at Home	0.30
19	Switzerland	Zermatt, Switzerland	Hotel Jägerhof	0.30
20	Switzerland	Verbier, Switzerland	B&B Café de la Poste	0.29
21	Switzerland	Verbier, Switzerland	Hôtel Terminus	0.29
22	Italy	Courmayeur, Italy	Hotel Locanda Belvedere	0.29
23	Switzerland	Zermatt, Switzerland	Antares Hotel	0.29
24	France	Courchevel, France	Chambre avec SdB, WC et entrée indépendantes	0.29
25	France	Méribel, France	Résidence Plein Soleil	0.29

we present these locations along with their PSi. As a remark, the PSi for the first 25 hotel location varies from 0.47 to 0.30 compared to the total data, where PSi varies from 0.47 to 0.06.

An analysis of the top 25—Figure 2—hotel location ranked descending on the Psi (preference selection index) indicates that the Verbier resort in Switzerland predominates, having 15 hotel locations out of 25 from the top 25 list, followed by the Méribel resort in France with only three hotel locations out of 25 from the top 25 list.

4.2. Multidimensional Data Analysis

The data were retrieved in Tableau Public software where they were analyzed with aggregated data, and Pearson correlations were calculated. By identifying the cities near the resorts, we used those data as geographical dimension, attributing geographical coordinates to each unit. Thus, the design of the map in Figure 3, contains the division of hotel location units.

Figure 4 presents the number of hotel location units analyzed per country for the 10 resorts analyzed. One can observe that the most analyzed units are from France (118) and the fewest are from Italy (38).

Another interesting analysis was conducted regarding the average price on each resort (Figure 5); one may observe that the highest price average is for Val d’Isere (EUR 5862), and the lowest price average is for Verbier (EUR 1691).

As for the booking score, one may observe that there are large differences between the resorts analyzed, the average score being between 8.03 and 8.95, which means very good reviews (Figure 6).

In Figure 7, we conducted an analysis of countries’ average booking scores and the average prices. Thus, in Austria, the average booking score is 8.83, and in France it is 8.27. At the same time, the maximum average price is in Austria (namely EUR 2786), and the lowest average price is in Italy (namely EUR 1869).

Regarding the type of hotel location, we observed that the predominant hotel locations are hotels (270 units), compared to the units managed by private hosts (98 units) (Figure 8).

To calculate the correlations between the analyzed data, we have used Pearson correlation model. The results are presented in

Table 5. Pearson correlations between analyzed criteria values.

Correlations
Corr No reviews/Price	−0.0567
Corr Score/Price	0.1502
Corr Distance from resort/Price	−0.2755
Corr Distance from ski lift/Price	−0.2549
Corr Distance from resort/Score	0.0528
Corr Distance from ski lift/Score	0.1010
Corr Score/No of reviews	0.0516
Corr Free cancellation/Price	−0.0544

. Generally speaking, there are no correlations between the measures analyzed, with a few exceptions. Thus, we may observe a weak reversed correlation between the distances between the resort and the ski slope/ski lift and the price (−0.27, respectively −0.25), which is an expected fact because people tend to pay more to get faster access to the ski facilities.

Surprisingly, there is no correlation between free cancellation and the price, which demonstrates that the hotel location units do not ask for higher prices when offering free cancellation, which might have a satisfying occupancy rate.

Another iteresting correlation between the booking score and the price is observed in Figure 9, with values varying from one country to another. Thus, if for Italy there is a weak reversed correlation between the booking score and the price, in Austria there is a medium correlation (0.33).

In Figure 10, we observe that there is a medium correlation between score and price in case of hotel units and there is no correlation between score and price in case of privately managed units.

5. Discussion

5.1. Theoretical Discussion

This research bridges the existing gap in the literature by studying the criteria used by tourists when choosing online a hotel location unit for a top destination ski resort. Previous studies did not approach this topic using the methodology described and the instruments employed in this research. Other studies focus on consumer perception of price fairness, which is also influenced by many factors, such as the (minimum) room rate, free cancellation and other services, which are directly under the control of hotel management, while others, such as room characteristics, are not [65].

Even so, the paper is only a starting point, as the analysis will be continued for other online reservation platforms as well.

The hypotheses are confirmed by the research. Thus, tourists have different preferences when choosing different countries’ ski resorts; they do not always choose according to scores and prices.

Additionally, tourists pay attention to the score when choosing a hotel or a privately managed host unit, with a strong cognitive correlation between the price and the expectations.

The study empowered the researchers by scrutinizing the features of five other dimensions, and it offered more comprehensive suggestions for practitioners on how to better utilize online data as a marketing tool. Theoretical implications also refer to the development of online marketing theory; one may conclude that when trying to advertise the hotel location units as a strategy, the accuracy and quantity of real data are very valuable strategies.

There are very few papers that focus on multicriteria decision models in the tourism industry [66] and the application of the multiple-criteria decision-making—MCDM—methods is proposed because the examined problem is related to a set of alternatives that should be estimated against a set of conflicting criteria.

5.2. Managerial Implications

As for managerial implications, hotel locations with ski-to-door access should pay attention to the description of the location, especially since they are very near the slope and the tourist cherish this feature.

The research represents a starting point for any online marketer and hotel location manager striving to attract more tourists. The practical implications are that online marketing strategists have to pay attention to the accuracy of hotel descriptions and location unit descriptions to develop a sustainable business. Future research will be conducted since all online reservation platforms provide innovative marketing opportunities.

Hotel ratings directly impact occupancy rates, and thus, turnover. The online reputation management deals with ratings and the continuously updated content (images and texts about the services being offered), considering that users expect professionally designed interfaces with exhaustive information and up-to-date profiles when looking for suitable accommodation. Responding to user ratings and reviews is another part of online reputation management.

Hospitality practitioners could sustainably enhance review management by setting alerts when reviews come in and ensuring that no review goes unnoticed. Additionally, they have to optimize the website for direct booking by publishing the best guest reviews on their website, responding to online reviews to show commitment to guest satisfaction and integrating a sustainable hotel reputation management tool to increase efficiency and speed when responding to guest reviews.

6. Future Research Directions and Limitations

From a scientific point of view, the novelty of the study and the proposed methodology is reflected in the application of the multi-attribute decision model and the two methods for the normalization of matrixes, which simplify the process of combining the objective and the subjective weights, which is a relatively new approach not widely applied in the field of the location selection.

A major theoretical contribution of this paper is its comprehensiveness in examining the criteria for choosing a hotel location unit, since previous studies explored the topic using questionnaires and surveys—traditional instruments of research—and the present study is using a methodology combining electronic application and a multi-attribute decision model, and also two normalizations. This study also offered areas worthy of more research efforts from the perspectives of practitioners and researchers. Even though the results of our research are attractive, we observe some limitations.

First of all, the choice of resorts was made on the basis of the literature review and the scores from a designated ski platform, but not on the basis of mathematical methods. Future studies will include regressions and factor analysis.

Second of all, the lack of data retrieved from booking determine the exclusion of many hotel locations units. There is a remark to be made: the lack of information may not be substituted by 0 because the results would have been distorted. Future studies will be extended to data from other platforms, such as Airbnb.

Another factor influencing our results was the small number of hotel location units referred as ski-to-door units (Figure 11). Unfortunately, there were insufficient available data regarding ski-to-door access/travel sustainable property, which is a limitation of our study. Future studies will collect data from other platforms as well in order to eliminate this limitation.

Third of all, if other values are subjected to analysis, one may obtain different weights, and thus, different rankings because the dispersion of values is crucial for a multicriterial analysis. Future studies will use other mathematical models, even more precise, which will complete these findings.