Identification of Applicable YouTubers for Hotels: A Case Study of Integrated Hybrid MCDM Model

Abstract

The coronavirus disease 2019 (COVID-19) pandemic has caused a serious business recession in various walks of life, particularly in the full-service hotel industry. YouTube has one billion active users and is undoubtedly a social media platform that companies use to build relationships with customers and create value for brands. Marketers should be aware of YouTubers’ significant influence on complex decision-making processes. Given the above reasons, identifying a YouTuber attracts the concerns of various industries; thus, this important issue is focused on and offered the study’s rationality. This study proposes an integrated hybrid MCDM model to organize the four key techniques of FDM, DEMATEL, ANP, and TOPSIS to identify YouTubers for hotels. Consequently, 12 key criteria and four core dimensions were identified to improve the decision of optimal YouTubers for promoting sustainable development and increasing the efficiency of decision-making. From the limited literature review, the proposed hybrid model was not observed regarding YouTuber identification of hotels; thus, this study provides a superior application contribution to address this important and interesting topic for academicians and practitioners.

1. Introduction

1.1. Research Background and Research Problem

Coronavirus disease 2019 (COVID-19) has severely affected the economies of all countries worldwide, including Taiwan. Although Taiwan successfully contained the pandemic to a great extent in 2019–2021, domestic tourism was still significantly affected. Because of the COVID-19 outbreak, most people choose to avoid crowded places, which has a significant negative impact on businesses, particularly hotels. For the tourism industry, the flow of people is equivalent to the flow of money. The significant reduction in the number of overseas tourists forces many hotels to lay off employees and causes a serious financial problem. It is very serious whether hotels can continue to survive this special accident and can positively maintain company operations when facing the serious problem of a great decline in operating revenue due to the COVID-19 pandemic. Thus, we focus on the hotel service industry as the research object and material. Achieving sustainable development for the hotel service industry is an important trigger for increasing online hotel operations, particularly for strengthening the promotion of social media. The use of social media to overcome these problems is an important and interesting issue. Social media is growing rapidly and has become indispensable to modern life. Brand companies are attempting to use online communities effectively for message delivery and communication activities [1]. Additionally, social media is considered a vital marketing tool that influences consumer behavior and stimulates companies to seek new opportunities to promote their brands [2]. The number of social media users is growing rapidly. Consequently, social media marketing appears to be an effective approach to interacting with consumers [3]. However, social media originally focused on social networking and connections. People use the media to share their lives, follow others, and use keywords to search for information. YouTube, Facebook, and Instagram are the social media platforms most commonly used in the real world; however, there are many differences between them. YouTube is an audiovisual platform that provides dynamic videos and music, while Facebook and Instagram are mainly based on posts [4]. YouTube provides a forum for users worldwide to connect with each other, exchange information, and stimulate creativity. YouTube is the preferred content publishing platform for video creators and advertisers. Many companies upload promotional videos on YouTube. Audiences can leave messages below videos, and companies can understand the public’s perception and acceptance of their audiovisual messages via comments [5]. Based on the above reasons, YouTube is considered the core of social media.

1.2. Research Motivation

With more than one billion users, YouTube is one of the largest video-sharing websites worldwide, and it plays a major role in popular culture [6]. Currently, YouTube is the largest audiovisual sharing website. It has also become a social networking platform for viewers to interact with other audiovisual creators and users speaking different languages to communicate on various topics [7]. To advertise on social media (e.g., Instagram, YouTube, and Facebook), YouTube’s method of recording short videos and focusing on the key points is the most effective marketing method to motivate viewers to take action [4]. YouTube is an appropriate medium to promote a brand [8]. The platform’s large number of engaged users allows YouTube to play an important role in corporate digital marketing [9]. Social networking websites can bring people closer. As one of the world’s largest audiovisual sharing platforms, YouTube has become an indispensable part of daily life for internet users. Many brands and business owners focus their digital marketing efforts on YouTube to achieve a faster and wider internet marketing reach [10].

Social media impacts consumers’ decision-making processes when selecting a hotel by influencing their searches, decisions, and hotel reservations [11]. Therefore, the hotel industry can use social media to communicate with consumers to understand their needs [12]. The hospitality industry has increased the use of new technologies and social media tools for marketing over the past 10 years [13]. In recent years, social media technologies have revolutionized marketing and advertising. Some platforms (e.g., YouTube) have become dependable and affordable marketing implements. Hotel managers should consider social media as helpful in improving business performance [14]. Social media has modified the interactions between organizations and customers. Many organizations, such as hotels, have started to employ social media as a marketing strategy. Thus, YouTube is the motivation for this study, given the above description.

1.3. Research Significance and Rationale

YouTube has the advantages of building communities and generating publicity. With the rising popularity of YouTube, a new type of internet amateur called a “YouTuber” has emerged. YouTubers know how to use YouTube’s strengths: they make their own videos, set up and operate their own YouTube channels, attract viewers through creative content, and accumulate active followers who subscribe to their channels. Consequently, YouTubers have become “influencers” in the community [15]. YouTubers also offer significant economic benefits. YouTubers have become influential owing to their large number of followers. Consequently, companies cooperate with well-known YouTubers to endorse their products, which is a social media marketing method. YouTubers convey values and beliefs and promote branded products on their own social platforms. They create videos based on their own characteristics and styles and use lively and realistic videos to build rapport with consumers, thereby influencing their followers’ and viewers’ desire to purchase a product. This is a new and effective way to endorse products [16]. Various companies want to market products and services more widely and prefer to work with YouTubers rather than artists on television [17]. In addition, YouTuber is an appropriate way to promote a brand. It is also a marketing communication tool that affects the purchasing journey [8]. Organizations should work hard to stimulate awareness and disseminate reliable product information through YouTubers. In addition, organizations should view their collaboration with YouTubers to market their goods and services as a continuous marketing strategy [18]. YouTube is currently the most popular video hosting website in Taiwan. It has also become a new marketing platform. Notably, advertisements recommended by YouTubers attract fewer negative ratings from viewers [19]. Companies should support their brands through well-known YouTubers to create favorable responses from Generation Y [3]. As a result, YouTubers play an important role in hotel marketing, although earlier studies have not focused on ways to offer hotel managers the right YouTubers for their business. In particular, the identification of optimal YouTubers is very difficult because decisions are made under the restrictions of multiple criteria and effective factors addressed. Thus, identifying YouTubers is an issue of multiple criteria decision making (MCDM), and this is the research significance of this study.

Decision-making has long been studied by many scholars and experts. While most studies on this topic have focused only on single criterion problems, in today’s rapidly changing environment, the problems faced by decision-makers are complicated and volatile. Usually, they cannot solve these problems using a single criterion, but should consider multiple criteria belonging to the same decision-making problem, thereby making the most appropriate decision accordingly. Consequently, MCDM has become a method often used by decision-makers. MCDM refers to an analysis method that evaluates solutions using mathematical programming to find the best solution under circumstances in which resources are limited and multiple criteria must be considered. MCDM can help decision-makers rank solutions and determine their ideal solutions [20]. The most suitable solution for MCDM can be determined only after determining the proper selection criteria and the screening process. Interestingly, the identification of a YouTuber is an MCDM problem, and it is important to maintain the sustainable development of hotel operations during the COVID-19 pandemic. This interesting issue provides the rationale for the present study.

1.4. Technical Research Used

Hsu and Lin [19] combined the modified Delphi method and the analytic hierarchy process (AHP) to select the best YouTubers, whereas Tüzemen [21] combined the Delphi method, AHP and TOPSIS techniques. However, Singh and Sarkar [22] stated that although the traditional Delphi method can be applied to determine criteria, the ambiguity and uncertainty of expert opinions still exist. Abdul-Hamid et al. [23] believed that the fuzzy Delphi method (FDM) transforms the perception of human language into a measurable scale through the fuzzy set theory, thereby obtaining objective and reasonable results. FDM requires only a single survey to reach a final decision. This method has become popular and has been proven to help managers solve real-life problems effectively. Bui et al. [24] also noted that fuzzy set theory is utilized to convert experts’ qualitative opinions into quantitative information, while the Delphi method is applied to filter out unnecessary criteria and rank criteria according to their importance. As a result, FDM can be used to obtain criteria with high validity and reliability from qualitative information. In addition, Dawood et al.’s [25] study shows that the criteria acquired from FDM have a high degree of validity and reliability. Based on the above literature review, past studies have pointed out the advantages of FDM against the disadvantages of the Delphi method, and it is particularly noted that FDM can help managers solve real-life problems more effectively; thus, FDM is emphasized and used in this study as a useful tool.

Second, AHP can systemize and hierarchize a huge and complex problem and make a comprehensive evaluation according to quantitative judgment to provide sufficient information for decision-makers to select an appropriate solution and reduce the risk of decision-making. A hierarchical structure can help decision-makers understand things deeply. To select an appropriate solution, decision-makers must evaluate alternative solutions in accordance with certain criteria and prioritize them accordingly [26]. Ho and Chen [27] also indicated that by using a hierarchical structure, AHP represents the complex relationships between the evaluation criteria and conducts a pairwise comparison (PC) between the evaluation criteria. As a result, this reduces the complexity of decision-making, as well as errors arising from subjective judgment. AHP makes a comprehensive evaluation after quantitative judgment, thus enabling decision-makers to analyze a problem and solve the problem based on objective data. Nevertheless, Teng [28] noted that AHP is an MCDM method that mainly decomposes decision-making problems into vertical hierarchical relationships and then evaluates them through quantitative judgments. AHP has a basic assumption that the elements in each level must be independent of elements at another level, and there is an independent relationship between the elements and solutions. Based on this assumption, the structure of a decision-making problem can only have a hierarchical relationship, which is an unreasonable restriction. Furthermore, in more complex decision-making problems, the original problem structure is deformed, which affects the quality of decision-making. In fact, numerous decision-making problems cannot be constructed with purely hierarchical relationships [29] because there may be interdependence and interactions between higher- and lower-level factors.

Third, the analytic network process (ANP), developed by Saaty [29], involves the use of a supermatrix to demonstrate the association between factors. Specifically, a supermatrix was used to resolve problems concerning interdependence between factors. The supermatrix is composed of several submatrices, each of which contains mutual relationships between factors in the same cluster and compares them pairwise with factors in other clusters. The values in each submatrix denote eigenvectors calculated using a PC and serve as weights. Subsequently, all the submatrices are combined to form a supermatrix. When the factors in the supermatrix are interdependent, repeatedly multiplying the supermatrix by itself yields a fixed converged value, with the extrema remaining unchanged. Kurian et al. [30] noted that ANP is strongly recommended for solving MCDM problems because this method considers the interaction between decision-making elements from different levels. In addition, Liu et al. [31] believed that ANP had been proven to be more effective than AHP in calculating the importance of different elements because it can explain the internal dependence and relationships between these elements more clearly. However, Schulze-González et al. [32] stated that although ANP can provide better results, it requires many questionnaire items and, therefore, requires more time for decision-makers and experts. This is mainly owing to the use of pairwise matrices, which require many comparisons.

Finally, TOPSIS is widely used to solve MCDM problems [33]. In addition, because of its ease of use, TOPSIS has become a practical tool for evaluating and ranking solutions, which makes it one of the most commonly applied approaches in different fields, such as human resources [34]. To reduce the excessive number of PC matrices of ANP and lower the difficulty for decision-makers to answer the questionnaire in practice, TOPSIS was applied in this study to rank the solutions.

Today, society is in the age of an online community. YouTube is the social media that most people use every day, and companies see the potential of the YouTube market. They cooperate with YouTubers to leave viewers a strong impression and engage in purchase or sharing behaviors. It is common for companies to use YouTubers to promote their products. YouTubers’ influence is mainly used to reach their larger markets. Therefore, this study proposes an integrated hybrid MCDM model with four key steps to highlight hotels in identifying applicable YouTubers to ensure sustainable development.

1.5. Research Purpose

From the limited literature review, the proposed integrated hybrid model has not been seen in the issue of YouTuber identification of hotels. Thus, this study makes a superior application contribution for academicians and practitioners to address this important and interesting topic. The following four key techniques were used for some applications in this study. The objectives of the study are addressed in the fourth section.

(1)

We propose an integrated hybrid model for the four key techniques of FDM, decision making trial and evaluation laboratory (DEMATEL), ANP, and TOPSIS, to effectively simulate the real case of YouTubers.

(2)

Interviews with hotel managers responsible for social media marketing in Taiwan. Review the literature on the selection of YouTubers and collect selection criteria. Choose the selection criteria for YouTubers according to FDM. Then, based on previous studies and interviews with hotel managers, the criteria are classified into a hierarchical structure that can be used to identify suitable YouTubers.

(3)

Integrate the hierarchical structure into a case hotel and combine meaningful applications of the four techniques to identify YouTubers.

(4)

Determine the optimal YouTubers for sustainable hotel development.

The remainder of this paper is organized as follows. First, we define the applications of YouTube and YouTubers from a review of academic activities and then describe the research methods for the background information. The next section presents the empirical case of a real application for YouTuber identification. Through the constructed integrated hybrid MCDM model, this study helps hotel managers find the best YouTuber. The final section presents the conclusions, research contributions in two corresponding views, and recommendations for future research.

2. Literature Review

To enhance the background of the issue addressed for the application solutions of YouTube and YouTubers, this section reviews past studies of both YouTube- and YouTuber-related applications.

The era of Web 2.0 focuses on content sharing and creation. The rising trend of networked audio and video content has seen not only conventional audio and video suppliers uploading content to the internet but also many internet users creating their own audio and video content. This boom in audio and video sharing has also driven the popularity of video-sharing websites and blogs. Owing to the emergence of social media platforms that provide audio and video sharing functions, online audio and video content have become a force that cannot be ignored [35]. The growth of audio and video streaming platforms has dramatically changed the way people use audio and video materials and how they use them for learning and entertainment. Among various audio and video streaming platforms, YouTube is the most representative. In addition to allowing users to upload videos, it provides simple query functions and allows users to create exclusive channels. YouTube is also a social media platform on which users can establish social communication through functions such as expressing their likes and dislikes, commenting on videos, and subscribing to channels [36]. YouTube is an online audio- and video-sharing platform that was developed in the United States. It provides users worldwide with a means to upload, watch, share, and comment on videos. It is also an audiovisual platform for communication, information exchange, and creativity [37]. An increasing number of marketing executives are turning to YouTube to spread their brand messages, and these messages influence customers’ purchasing behaviors.

The popularity of YouTube has led to the emergence of a new profession called YouTubers. YouTubers operate their own YouTube channels as professionals, upload videos to express themselves, increase their popularity, and generate income [38]. Moreover, YouTubers use editing software to create videos and publish or upload their own videos on the internet. YouTubers draw viewers through creative and in-depth audio and video content and attract viewers who subscribe to their channels [37]. The 21st century is an era of mobile communications and social media. Internet technology allows users to own their media, so many previously unknown people have become influential YouTubers. YouTubers’ videos attract millions of views and have become an influential group in the era of the internet [39].

User-generated content (UGC) is the most obvious feature of YouTube. Any user can create and upload content, which has allowed many amateurs to become influential YouTubers. Videos produced by popular YouTubers attract a certain number of views, which increases the number of advertising exposures. In addition, because of their popularity and high number of views, business owners may approach YouTubers to endorse and promote their products, thus allowing YouTubers to generate income other than just through video advertisements. Given the low barriers to entry, freedom to create, and opportunities to earn both fame and fortune, an increasing number of people are investing in the YouTuber industry, which is booming worldwide [10]. However, the identification of the optimal YouTuber is important. To address the deficiencies in the methodological results of past research, this study proposes an integrated hybrid MCDM model for YouTube identification defined in the COVID-19 outbreak.

3. Research Methods

This section describes the research methods of this study in the following four steps. Given the reasons mentioned above, this study proposes an integrated hybrid MCDM model to join the four core techniques of FDM, DEMATEL, ANP, and TOPSIS to benefit hotels in identifying suitable YouTubers for sustainable promotion during the COVID-19 pandemic. (1) FDM was used to select the criteria for the identification of YouTubers. The criteria were then classified into dimensions to form a hierarchical structure based on previous studies and interviews with hotel managers. (2) Next, a mutual influence relationship between the dimensions was established based on DEMATEL. (3) ANP was used to generate the weight of each dimension and selection criterion according to this mutual influence relationship. (4) Finally, to reduce the high number of PC matrices and address decision-makers’ trouble in answering the questionnaire, TOPSIS includes the weights of the selection criteria integrated by ANP to help the hotel in identifying the optimal YouTuber and increases the efficiency of their decision-making. The main research flowchart of the proposed hybrid model is illustrated in Figure 1 for ease of reading.

They are described in the following four subsections for supporting the two views of technical background and managerial applications used in the order of four key methods—FDM, DEMATEL, ANP, and TOPSIS, respectively—and their empirical case is explained in the next part.

3.1. Technique 1—FDM and Its Application Context

In MCDM methods, data are accumulated through discussions among experts. Hence, the Delphi method has been applied extensively [40]. The Delphi method relies on experts’ professional experience, intuition, and value judgment. Therefore, the Delphi method requires an expert group of 10–15 people to solve a specific problem. The Delphi method uses the diverse viewpoints of experts from different fields to understand the opinions of other experts in successive questionnaires and to revise their own views. Consensus is achieved based on this brainstorming feedback. Experts’ commitments influence the completion of the Delphi method. Therefore, before selecting experts, it is essential to make a list of experts, discuss research issues with them, and acquire the commitment of participating members of the expert group. The following points must be discussed with the members of the expert group when implementing the Delphi method [28]: (1) the procedure, content, and time investment of the Delphi method; (2) the purpose of the Delphi method: why the research is being undertaken, how it will be conducted, and the benefits of the research; and (3) the benefits of participation—in addition to learning the Delphi method, experts will also gain a lot of new knowledge, and may even be paid.

A fixed number of experts were selected for the Delphi method, and they could not quit at will. These experts do not conduct direct face-to-face discussions, but rely entirely on a carefully designed series of questionnaires [28]. However, owing to criticisms and limitations of the Delphi method, scholars continue to revise the way in which the Delphi method operates to make it more efficient and reliable. After 1985, the Delphi method and fuzzy theory were combined to form FDM, which attempted to resolve the difficulties encountered in the traditional Delphi method. It also has the advantage of being able to cope with vague language and retain expert opinions [41].

FDM can be applied to generate expert consensus and establish critical criteria [42,43]. Therefore, FDM is a hybrid method that combines a qualitative method (Delphi method) and a mathematical tool (fuzzy analysis) to consider MCDM problems. Using tools that combine the qualitative (Delphi method) and quantitative (fuzzy analysis) methods can reduce uncertainty when collecting data and opinions [44]. While the Delphi method filters out invalid criteria, FDM determines valid criteria by converting language preferences into explicit numerical values [45]. FDM is capable of considering the ambiguity and uncertainty of data and has been extensively employed in many fields.

FDM has another useful function and already has successfully applied to delete the unimportant criteria for personnel selection issues. For example, Chang [46] used FDM to screen the selection criteria for professional e-sports team gamers. Wu et al. [47] utilized FDM to delete the unimportant selection criteria for variety show hosts. Lim et al. [48] utilized FDM to delete the unimportant criteria for selecting live streamers. Therefore, this study used FDM as the research method, which can be completed by inviting experts to fill in a single questionnaire. On the one hand, it solves the problem of respondents’ vague language and improves their willingness to answer the questionnaire. However, it decreases errors in experts’ answers to improve the reliability of the research.

According to Hwang et al. [49], after designing an FDM questionnaire, experts are employed to evaluate the importance of all the questions. After answering the questionnaire, the triangular fuzzy number (TFN) for every selection item with the minimum, geometric mean (GM), and maximum values can be obtained. Criteria with importance greater than 80% are generally regarded as important. In the questionnaire, which is based on a 9-point Likert scale (1 = strongly unimportant, 2 = unimportant, 3 = moderately unimportant, 4 = slightly unimportant, 5 = neutral, 6 = slightly important, 7 = moderately important, 8 = important, and 9 = strongly important), criteria with geometric means (GMs) greater than 7.2 are regarded as vital. In this study, criteria greater than 7.2 are thus reserved as YouTuber selection criteria.

Given the reasons of the helpful functions mentioned above from past studies (particularly in Hwang et al. [49]), concepts of FDM are applied in this study.

3.2. Technique 2—DEMATEL and Its Application Context

The five calculation steps of DEMATEL are as follows [32,46,47,50,51]. (1) Establishment of the direct-relation matrix (DM): a list of elements can be created using different approaches, such as brainstorming, literature review, and expert opinions. The values of the degree of mutual influence between the elements were obtained by experts to establish a DM. (2) Establishment of the initial DM. (3) Establishment of normalized DM. (4) Construction of the total relation matrix. (5) Setting the threshold: the threshold depends on the experts’ opinions and can be used to filter out negligible correlations with lower impact in the total relation matrix.

DEMATEL can identify interdependence between elements and help develop diagrams to reflect interrelationships between elements, which can be used to explore and solve complex and entangled problems [52]. DEMATEL can effectively explain the complex structure of causal relationships by examining the degree of influence between elements and uses matrix operations to obtain the causal relationship and strength of the influence between elements [51]. DEMATEL is a research method used to identify the causal relationship between multiple elements and solve existing or potential problems. Therefore, the government and relevant organizations can use the empirical results of this research method to formulate policies. DEMATEL was originally used to solve national and social problems and was subsequently widely used to study causal relationships in various fields [53]. Finally, DEMATEL is an effective approach for exploring the causal relationships between elements [42]. Therefore, this study used DEMATEL to establish the interaction relationships among dimensions.

3.3. Technique 3—ANP and Its Application Context

Proposed by Saaty, AHP is a qualitative and quantitative MCDM method applied to prioritization and resource planning. AHP aims to systemize a complex problem, hierarchize each evaluation dimension of a problem, and determine different evaluation layers for performing PC matrices [54]. AHP is primarily applied to MCDM problems under uncertainty. In the process of problem-solving, AHP assumes that the elements of each level are independent of each other and systematize complex problems for evaluation. However, real-life problems often involve interdependence and feedback. As the problem increases, the relationships become more complex. If the assumption of independence is used, the problem may be oversimplified, resulting in biased evaluation results. To avoid this shortcoming, Saaty [29] developed ANP that considered interdependence and feedback connections. In some structures of decision-making problems, it is difficult to express subordination and high–low relationships between levels. This type of problem structure has feedback connections; therefore, one level may dominate some levels and be dominated by other levels. ANP can be administered to solve complex decision-making problems [28]. The decision-making procedure of ANP is implemented and highlighted in the following key points [28,29]. (1) Definition of the decision problem: the elements that could affect the decision-making problem are included based on the nature and system of the problem. At this stage, a planning team was established to collect relevant information and define the scope of the decision-making problem. (2) Composition of the decision-making group: experts in related fields are recruited to form decision-making groups based on the fields involved and the degree of complexity of the decision-making problem. (3) Construction of the problem structure: the planning group organizes and summarizes the relevant information of the decision-making problem and provides the information to the members of the decision-making group for reference. The decision-making group then used brainstorming to determine the system elements that affect the decision-making problem, including goals, dimensions, criteria, and solutions. In the problem structure, loop arcs, two-way arrows, and one-way arrows are used to connect the levels to indicate their subordination and feedback connections. (4) Questionnaire design and survey: the relative importance of each element was judged by experts from the decision-making group according to the hierarchical structure, with higher-level elements as evaluation benchmarks. Generally, the survey can be carried out by designing a questionnaire that clearly describes each PC question to assist experts’ judgments. (5) Integration of expert preferences: this study utilizes the aggregation of individual judgments (AIJ) to integrate the opinions of decision-makers. (6) Building the PC matrices and calculating the weight of each element: many PC matrices can be generated according to the experts’ integrated judgment preferences. The relative importance of the lower-level elements dominated by higher-level elements can be judged on a scale of 1 to 9 (the lower-level elements are compared in pairs). The judged preferences of decision-makers or experts must be transitive and therefore must go through a consistency test procedure. Once the integration of PC matrices is completed, the eigenvalues and eigenvectors of the PC matrices can be obtained, and the weights of the elements can be obtained accordingly. (7) Supermatrix operation: a supermatrix is adopted in ANP to compute the relative weights of the elements to deal with dependence between elements in the problem structure. The supermatrix consists of many submatrices, which are the PC matrices obtained in the previous step. If there is no correlation between the elements, then the PC value of the submatrices is 0. The unweighted supermatrix must be converted through a specific procedure because the column values of the matrix may not conform to the column-stochastic principle (for example, the sum of column values does not equal 1). The weighted supermatrix can be obtained by assigning a relative importance weight to the unweighted supermatrix. Through the limiting process, the relative weights of the elements are obtained. (8) Calculation of the relative weights of the solutions: under each criterion, the relative importance of each solution is compared, a PC matrix is established, the preferences are integrated, and the maximum eigenvalue and the corresponding eigenvector are calculated to obtain the relative weight of each solution. (9) Rank of solutions: finally, the rank of solutions is determined by integrating the relative weight of elements (from the limiting supermatrix) and solutions.

In mathematical problems where elements are related, AHP cannot determine the weights of elements; however, ANP provides decision-makers with a more generalized model that can consider the interdependence between higher- and lower-level elements [55]. Abidi et al. [56] noted that compared to other MCDM tools, ANP is useful because it is relatively simple to apply and the results are reliable. Additionally, ANP can be applied to cope with difficult decision-making problems. It can also deal with the interdependence and feedback between elements in the network, and it is suitable for dealing with similar complex situations [57]. This study obtains the weights of the dimensions and criteria through ANP according to the mutual influence relationships between the dimensions.

3.4. Technique 4—TOPSIS and Its Application Context

TOPSIS is an MCDM approach that selects an alternative solution based on the shortest distance to the ideal solution (IS) and the longest distance from the anti-ideal solution (AS). IS refers to a solution with maximum benefit but minimum cost, whereas AS refers to a solution with minimum benefit and maximum cost. In short, an IS comprises the best evaluation results for all evaluation criteria, and an AS comprises the worst evaluation results for all evaluation criteria [58,59]. Liang et al. [54] also stated that TOPSIS provides a simple and clear framework and is easily accepted by decision-makers in the process of evaluation. Its basic principles are as follows. (1) Evaluation of a solution under each criterion, identifying the IS that has optimal values under all criteria and the AS that has the worst values under all criteria. (2) Calculate the distance between each solution and the IS and AS based on the principle of “closest to the IS” and “farthest from the AS.” (3) Computation of the relative closeness of each solution to the IS and identification of the optimal solution accordingly.

TOPSIS is one of the most appropriate and effective means for solving MCDM problems [33]. It is beneficial for solving MCDM problems in certain and uncertain environments [60]. TOPSIS has been successfully applied to MCDM problems in various disciplines to provide a simple, effective, and computationally efficient mechanism for coping with various criteria [61]. TOPSIS is also efficient and beneficial for solving multiple criteria problems [62]. TOPSIS has been widely used because of its beneficial effects on uncomplicated computing processes and ease of operation [63]. In this study, to avoid subjectively assigning weights, TOPSIS includes the weights of the selection criteria integrated by ANP to objectively identify the best YouTuber. Based on the above statements of the four key techniques, the proposed hybrid model is implemented and provided in detail in the following section.

4. An Empirical Application Case for YouTuber Identification

4.1. Implementation Results with a Real Case

Basically, this study initially determined the criteria for YouTuber identification based on interviews with managers in charge of social media marketing in Taiwanese hotels and a review of the literature on YouTuber selection. Five social media marketing managers in the Taiwanese hotels revised the first draft of the questionnaire and refined the wording of the items to make the content of the questionnaire applicable, practical, and realistic, thereby ensuring the reliability of the questionnaire. Moreover, as the questionnaire items were taken from existing studies, they have content validity. In the real process of implementation, the study interviewed hotel managers and also reviewed the literature to obtain short list of final 28 selection criteria and an item of other suggestion. For the section on answering the questionnaire, hotel managers scored the importance of the criteria based on their practical work experience. Accordingly, we sent out the paper questionnaires personally and via an online questionnaire system for the process of data collection. We added the informed consent document and its questionnaire of FDM (Appendix A). In total, 54 questionnaires were returned. To improve the representativeness of the sample, this study calculated the GM of each criterion based on the answers provided by 40 hotel managers with more than 10 years of experience and retained 12 criteria with GMs greater than 7.2. Valid returned questionnaires were analyzed scientifically and systematically. We also added the general information of 40 hotel managers in Appendix B.

Table 1. The empirical results of FDM.

Criteria	Contributors	TFN Values
The content of the video is close to life.	[19]	(8.0000, 8.4853, 9.0000)
The content of the video is easy to understand.	[19]	(8.0000, 8.5354, 9.0000)
The product is introduced objectively.	[19]	(8.0000, 8.7646, 9.0000)
The content of the video is combined with personal characteristics.	[19]	(8.0000, 8.2877, 9.0000)
Withstanding the pressure of public opinions.	[19]	(4.0000, 5.9050, 8.0000)
Cooperation with the contract content.	[19]	(5.0000, 6.7733, 9.0000)
Applying the characteristics of social media to share the video.	[19]	(5.0000, 7.0020, 9.0000)
Communication ability of the YouTuber.	[19]	(6.0000, 6.9729, 8.0000)
Interaction well with audiences.	[19]	(6.0000, 7.5476, 9.0000)
The crisis management ability of the YouTuber.	[19]	(5.0000, 6.5324, 9.0000)
Understanding of audience needs.	[19]	(5.0000, 6.3902, 8.0000)
The YouTuber is trustworthy.	[19]	(5.0000, 6.8771, 8.0000)
The market insight of the YouTuber.	[19]	(5.0000, 6.2889, 8.0000)
Script creativity.	[19]	(7.0000, 8.2192, 9.0000)
Production of new videos regularly.	[19,21]	(5.0000, 6.8322, 7.0000)
Controlling video production time effectively.	[19]	(7.0000, 8.4222, 9.0000)
The quality of the video.	[19,21]	(7.0000, 8.3908, 9.0000)
Personal image of the YouTuber.	Managers recommend.	(5.0000, 6.5905, 9.0000)
The affinity of the YouTuber.	[19]	(5.0000, 6.2619, 7.0000)
The audience appeal of the YouTuber.	Managers recommend.	(5.0000, 6.2302, 8.0000)
Cooperation with celebrities.	Managers recommend.	(6.0000, 6.7370, 8.0000)
The number of subscribers to the YouTube channel.	[21]	(6.0000, 8.3042, 9.0000)
The number of views of the YouTube channel.	Managers recommend.	(8.0000, 8.6111, 9.0000)
The ambition of the YouTuber.	Managers recommend.	(5.0000, 6.5916, 9.0000)
The continuous learning ability of the YouTuber.	Managers recommend.	(4.0000, 5.3372, 8.0000)
The persuasiveness of the video.	[19]	(6.0000, 6.4119, 9.0000)
The number of likes for the videos on the YouTube channel.	[21]	(7.0000, 8.0519, 9.0000)
Video title setting ability of the YouTuber.	Managers recommend.	(7.0000, 8.1042, 9.0000)

Note: TFN refers to the minimum, GM, and maximum values of the criteria.

presents the empirical results of FDM with the TFN values.

Finally, based on the literature [19,21] and interviews with hotel managers, the criteria were classified into dimensions. A hierarchical structure is represented as follows. (1) The personal dimension comprises three criteria: the content of the video is combined with personal characteristics, interaction well with audiences, and script creativity. (2) The content dimension comprises three criteria: the content of the video is close to life, the content of the video is easy to understand, and the product is introduced objectively. (3) The marketing dimension comprises three criteria: the number of subscribers to the YouTube channel, the number of views of the YouTube channel, and the number of likes for the videos on the YouTube channel. (4) The production dimension comprises three criteria: controlling video production time effectively, the quality of the video, and video title setting ability of the YouTuber.

This study will help hotel managers responsible for social media marketing to effectively identify the best YouTuber. The questionnaire of DEMATEL, ANP, and TOPSIS was designed based on the hierarchical structures. The questionnaire was filled out by three managers from a company to analyze three YouTubers (solutions). The three managers are general manager of marketing department, manager of marketing department, and manager of sales department, respectively. Furthermore, it is noted that the short list of the three YouTubers is given and selected by the case hotel. YouTuber 1 made the channel in 2011, and she focuses on hotel and restaurant unboxing. YouTuber 2 made the channel in 2021, and this couple aims on starting up a business, investment, child rearing, hotel, and food unboxing. As to YouTuber 3, she made the channel in 2016, and she focuses on providing guidelines on domestic tourism, outbound tourism, and hotel unboxing. Accordingly, valid returned questionnaires were analyzed using Microsoft Excel. In the DEMATEL part, the initial DM and total relationship matrix of the dimensions are displayed in

Table 2. The initial DM of the dimensions.

Dimensions	Personal	Content	Marketing	Production	Total
Personal	0.0000	4.0000	4.0000	4.0000	12.0000
Content	3.6667	0.0000	4.0000	4.0000	11.6667
Marketing	4.0000	3.3333	0.0000	3.6667	11.0000
Production	4.0000	3.3333	3.6667	0.0000	11.0000
Total	11.6667	10.6666	11.6667	11.6667

and

Table 3. The total relation matrix of the dimensions.

Dimensions	Personal	Content	Marketing	Production
Personal	4.9459	4.8649	5.1892	5.1892
Content	5.0721	4.5135	5.0811	5.0811
Marketing	4.8829	4.5405	4.6262	4.8603
Production	4.8829	4.5405	4.8603	4.6262

The threshold value is 4.5000.

, respectively, where the threshold value is 4.5000, and the mutual influence relationship between the dimensions is established. A hierarchical structure with a mutually influential relationship is shown in Figure 2.

Next, based on the mutual influence relationship between the dimensions, the overall score of the decision-making group was aggregated by GM based on ANP. The consistency ratio (CR) of the questionnaires was lower than 10%, indicating that the questionnaires were acceptable. The PC matrices and weights of the dimensions are presented in

Table 4. The PC matrix of the dimensions under the effect of Personal.

Dimensions	Personal	Content	Marketing	Production	Weights
Personal	1.0000	0.6934	1.0772	0.5848	0.1948
Content	1.4422	1.0000	2.1544	0.7368	0.2947
Marketing	0.9283	0.4642	1.0000	0.5503	0.1672
Production	1.7100	1.3572	1.8171	1.0000	0.3433
CR = 0.0074

,

Table 5. The PC matrix of the dimensions under the effect of Content.

Dimensions	Personal	Content	Marketing	Production	Weights
Personal	1.0000	1.7100	2.4662	1.0000	0.3414
Content	0.5848	1.0000	1.3572	0.7368	0.2083
Marketing	0.4055	0.7368	1.0000	0.5503	0.1517
Production	1.0000	1.3572	1.8171	1.0000	0.2986
CR = 0.0032

,

Table 6. The PC matrix of the dimensions under the effect of Marketing.

Dimensions	Personal	Content	Marketing	Production	Weights
Personal	1.0000	0.4055	0.7937	1.0000	0.1819
Content	2.4662	1.0000	1.7100	1.2599	0.3666
Marketing	1.2599	0.5848	1.0000	1.0000	0.2237
Production	1.0000	0.7937	1.0000	1.0000	0.2279
CR = 0.0131

and

Table 7. The PC matrix of the dimensions under the effect of Production.

Dimensions	Personal	Content	Marketing	Production	Weights
Personal	1.0000	1.3867	2.7144	1.0000	0.3243
Content	0.7211	1.0000	1.8420	0.3684	0.1947
Marketing	0.3684	0.5429	1.0000	0.5503	0.1341
Production	1.0000	2.7144	1.8171	1.0000	0.3469
CR = 0.0350

.

The weight of each criterion in the dimensions was obtained from the PC matrices, and an unweighted supermatrix was then generated. For example, under the effect of the criterion (the content of the video is combined with personal characteristics), the PC matrix of the content dimension is displayed in

Table 8. The PC matrix of the content dimension under the effect of “The content of the video is combined with personal characteristics”.

Criteria	4	5	6	Weights
4	1.0000	4.9324	1.1006	0.5030
5	0.2027	1.0000	0.4149	0.1254
6	0.9086	2.4101	1.0000	0.3716
CR = 0.0326

.

Subsequently, the weight of each criterion was also generated.

Table 9. The unweighted supermatrix.

Criteria	1	2	3	4	5	6	7	8	9	10	11	12
1	0.5371	0.3355	0.5470	0.4642	0.2961	0.3926	0.2860	0.1477	0.1219	0.2622	0.3450	0.3040
2	0.2111	0.5121	0.3271	0.3304	0.3405	0.1657	0.5130	0.4261	0.3485	0.3464	0.4246	0.3255
3	0.2518	0.1524	0.1259	0.2054	0.3634	0.4417	0.2009	0.4261	0.5296	0.3914	0.2304	0.3705
4	0.5030	0.6218	0.5905	0.5337	0.2271	0.1559	0.1210	0.5426	0.4379	0.2422	0.1610	0.1687
5	0.1254	0.2967	0.2047	0.2284	0.4835	0.4943	0.3155	0.1719	0.3162	0.2825	0.2319	0.3366
6	0.3716	0.0815	0.2047	0.2379	0.2894	0.3498	0.5635	0.2855	0.2459	0.4753	0.6070	0.4947
7	0.5205	0.5659	0.4439	0.2642	0.2015	0.4283	0.1579	0.5325	0.4328	0.4283	0.5258	0.4190
8	0.1825	0.3011	0.1991	0.1501	0.3798	0.2109	0.3575	0.2286	0.3505	0.2109	0.2502	0.2905
9	0.2970	0.1330	0.3570	0.5857	0.4186	0.3609	0.4846	0.2388	0.2167	0.3609	0.2240	0.2905
10	0.5346	0.5024	0.3890	0.2615	0.2255	0.4023	0.3197	0.3649	0.3258	0.4295	0.5537	0.2024
11	0.2044	0.1665	0.1267	0.0871	0.1296	0.2826	0.1631	0.1754	0.1370	0.2230	0.1371	0.5361
12	0.2611	0.3311	0.4844	0.6515	0.6449	0.3151	0.5172	0.4597	0.5372	0.3474	0.3092	0.2615

presents the unweighted supermatrix. The weighted supermatrix is listed in

Table 10. The weighted supermatrix.

Criteria	1	2	3	4	5	6	7	8	9	10	11	12
1	0.1046	0.0653	0.1065	0.1585	0.1011	0.1340	0.0520	0.0269	0.0222	0.0850	0.1119	0.0986
2	0.0411	0.0997	0.0637	0.1128	0.1162	0.0566	0.0933	0.0775	0.0634	0.1123	0.1377	0.1056
3	0.0490	0.0297	0.0245	0.0701	0.1241	0.1508	0.0365	0.0775	0.0963	0.1269	0.0747	0.1201
4	0.1482	0.1833	0.1740	0.1112	0.0473	0.0325	0.0443	0.1989	0.1605	0.0472	0.0314	0.0329
5	0.0370	0.0874	0.0603	0.0476	0.1007	0.1030	0.1156	0.0630	0.1159	0.0550	0.0452	0.0655
6	0.1095	0.0240	0.0603	0.0496	0.0603	0.0729	0.2066	0.1046	0.0901	0.0925	0.1182	0.0963
7	0.0870	0.0946	0.0742	0.0401	0.0306	0.0650	0.0353	0.1191	0.0968	0.0574	0.0705	0.0562
8	0.0305	0.0503	0.0333	0.0228	0.0576	0.0320	0.0800	0.0511	0.0784	0.0283	0.0335	0.0390
9	0.0496	0.0222	0.0597	0.0889	0.0635	0.0547	0.1084	0.0534	0.0485	0.0484	0.0300	0.0390
10	0.1835	0.1725	0.1335	0.0781	0.0673	0.1201	0.0729	0.0832	0.0742	0.1490	0.1921	0.0702
11	0.0702	0.0572	0.0435	0.0260	0.0387	0.0844	0.0372	0.0400	0.0312	0.0774	0.0476	0.1860
12	0.0896	0.1137	0.1663	0.1945	0.1925	0.0941	0.1179	0.1048	0.1224	0.1205	0.1073	0.0907

and can be generated by multiplying the criteria in the unweighted supermatrix by the weights of the dimensions. Through the limiting process, a limit supermatrix is generated. The weight of each criterion can then be estimated, as shown in

Table 11. The limiting supermatrix.

Criteria	1	2	3	4	5	6	7	8	9	10	11	12
1	0.0948	0.0948	0.0948	0.0948	0.0948	0.0948	0.0948	0.0948	0.0948	0.0948	0.0948	0.0948
2	0.0913	0.0913	0.0913	0.0913	0.0913	0.0913	0.0913	0.0913	0.0913	0.0913	0.0913	0.0913
3	0.0844	0.0844	0.0844	0.0844	0.0844	0.0844	0.0844	0.0844	0.0844	0.0844	0.0844	0.0844
4	0.0940	0.0940	0.0940	0.0940	0.0940	0.0940	0.0940	0.0940	0.0940	0.0940	0.0940	0.0940
5	0.0718	0.0718	0.0718	0.0718	0.0718	0.0718	0.0718	0.0718	0.0718	0.0718	0.0718	0.0718
6	0.0873	0.0873	0.0873	0.0873	0.0873	0.0873	0.0873	0.0873	0.0873	0.0873	0.0873	0.0873
7	0.0661	0.0661	0.0661	0.0661	0.0661	0.0661	0.0661	0.0661	0.0661	0.0661	0.0661	0.0661
8	0.0416	0.0416	0.0416	0.0416	0.0416	0.0416	0.0416	0.0416	0.0416	0.0416	0.0416	0.0416
9	0.0542	0.0542	0.0542	0.0542	0.0542	0.0542	0.0542	0.0542	0.0542	0.0542	0.0542	0.0542
10	0.1194	0.1194	0.1194	0.1194	0.1194	0.1194	0.1194	0.1194	0.1194	0.1194	0.1194	0.1194
11	0.0698	0.0698	0.0698	0.0698	0.0698	0.0698	0.0698	0.0698	0.0698	0.0698	0.0698	0.0698
12	0.1253	0.1253	0.1253	0.1253	0.1253	0.1253	0.1253	0.1253	0.1253	0.1253	0.1253	0.1253

.

In the TOPSIS section, decision-makers allocate points to the solutions according to each criterion on a scale of 1 to 9. The GM is used to integrate the decision matrices of the three managers’ opinions to establish a normalized decision matrix, as presented in

Table 12. The normalized decision matrix.

	1	2	3	4	5	6	7	8	9	10	11	12
YouTuber 1	0.6133	0.4461	0.5383	0.5428	0.5664	0.5757	0.5647	0.5415	0.5311	0.5310	0.5724	0.5704
YouTuber 2	0.5016	0.6186	0.5065	0.5191	0.5733	0.5535	0.5430	0.5960	0.5163	0.5346	0.4681	0.4983
YouTuber 3	0.6101	0.6468	0.6735	0.6602	0.5921	0.6019	0.6215	0.5929	0.6719	0.6575	0.6733	0.6529

.

The weights obtained by ANP are multiplied by the normalized decision matrix to generate the weighted decision matrix of the solution presented in

Table 13. The weighted decision matrix.

	1	2	3	4	5	6	7	8	9	10	11	12
YouTuber 1	0.0581	0.0407	0.0454	0.0510	0.0406	0.0502	0.0374	0.0225	0.0288	0.0634	0.0400	0.0714
YouTuber 2	0.0475	0.0565	0.0427	0.0488	0.0411	0.0483	0.0359	0.0248	0.0280	0.0638	0.0327	0.0624
YouTuber 3	0.0578	0.0591	0.0568	0.0621	0.0425	0.0525	0.0411	0.0247	0.0364	0.0785	0.0470	0.0818

, thereby determining IS and AS. The separation distance of the solutions was based on the Euclidean distance (between IS and AS). The rank of the solutions was determined based on the relative closeness of each solution to the IS, as shown in

Table 14. The ranking of solutions.

	The Separation Distance to the IS	The Separation Distance to the AS	The Relative Closeness to the IS	Rank
YouTuber 1	0.0325	0.0163	0.3336	2
YouTuber 2	0.0375	0.0159	0.2982	3
YouTuber 3	0.0003	0.0418	0.9921	1

Note: IS refers to ideal solution; AS refers to anti-ideal solution.

. In this case, the third YouTuber is the best, and the second is the worst, as shown in Table 14. The company selected the top YouTuber based on the conclusions of this study.

In particular, data in MCDM problems often cannot be precisely determined and changeable. Hence, a vital step in various applications of MCDM is to execute a sensitivity analysis on the input data [64]. The sensitivity analysis, which can check the robustness and reliability of the decision, was carried out in this study. We changed the weights of criteria with the minimum and maximum values, respectively. Firstly, the weight of the criterion with the minimum value was adjusted to the maximum (i.e., 0.0416 to 0.6367); conversely, the weight of the criterion with the maximum value was adjusted to the minimum (i.e., 0.1253 to 0.0010).

Table 15. The results of sensitivity analysis for the integrated model.

Criteria	Original Weights	The Weight Value from the Minimum to the Maximum	The Weight Value from the Maximum to the Minimum
1	0.0948	0.0407	0.1061
2	0.0913	0.0372	0.1026
3	0.0844	0.0303	0.0957
4	0.0940	0.0399	0.1053
5	0.0718	0.0177	0.0831
6	0.0873	0.0332	0.0986
7	0.0661	0.0120	0.0774
8	0.0416	0.6367	0.0529
9	0.0542	0.0001	0.0655
10	0.1194	0.0653	0.1307
11	0.0698	0.0157	0.0811
12	0.1253	0.0712	0.0010
Best solution	YouTuber 3	YouTuber 3	YouTuber 3

presents the results of sensitivity analysis. Consequently, the rankings are stable, and thus it is clear that YouTuber 3 is still the best solution in this study.

4.2. Empirical Results

In the empirical results, this study achieves the following key strengths, which can support some meaningful information and knowledge associated with the related identification of optimal YouTubers to interested parties. These are highlighted as follows:

(1)

Initially, 40 hotel managers with more than 10 years of experience retained 12 criteria with GMs greater than 7.2 from the empirical case. The 12 selection criteria of YouTubers were filtered to create an effective control mechanism structure, which can be used as a capable and helpful reference for future hotel or hospitality strategic management.

(2)

DEMATEL is an effective operational interface and tool to unearth the causal relationships between elements, such as the 12 criteria in this study. This study particularly used good DEMATEL to build an interesting interaction relationship found among dimensions where the threshold value is 4.5000, and this threshold value can provide a relative measure of the watershed to obtain a satisfactory result. Thus, this study proves that DEMATEL has efficient and reliable consequences.

(3)

From a literature review, ANP has a superior effect and is positively suggested to address and solve MCDM problem; thus, this study adopted ANP to correctly acquire the weights of each dimension and criterion according to the mutual influence relationship. The study results show that ANP is a good alternative to process and determine the key factors.

(4)

More importantly, TOPSIS makes the aggregate include the weights of various core criteria integrated by ANP, which helps identify and determine the best YouTuber to increase decision-making quality. This study provided an integrated model of FDM, DEMATEL, ANP, and TOPSIS, which is a good alternative for the social media issue of YouTubers for sustainable hotel development and management.

(5)

Consequently, a managerial outcome: YouTuber 3 > YouTuber 1 > YouTuber 2 is the goal for the interested parties. From the managerial perspective, it implies that the hotel manager should select YouTuber 3 as the first spokesperson for propaganda advertisements to promote the new commercial activity, and this will have the most suitable solution for hotels. At the same time, this result can make a more sustainable practice for hotels than for others.

5. Conclusions

This section provides a summary of the data analysis results, including research highlights, management implications, research contributions, and subsequent research.

5.1. Research Highlights

The COVID-19 outbreak arrived as a tsunami and hit the economies of various countries. Even though pandemic prevention in Taiwan has been successful, it is still difficult to avoid the impact of the pandemic. Domestic tourism industries have borne the brunt of the pandemic. Tourists cannot enter Taiwan, and many hotels cannot afford fixed costs to survive. Sustainable development in the service industry is an important trigger for online hotel operations, particularly on social media. Social media advances not only online personal interaction but also creates new relationships among organizations and customers. Social media is considered an important marketing tool that influences consumer behavior and is a valuable asset for organizations. To promote products and services, marketers place many more resources in social media and concentrate on how to apply social media and communicate with people.

YouTube allows billions of people worldwide to watch, explore, and share original video content. It provides users with a platform that is convenient to use, exchange information, connect with other users, and inspire creativity. Consequently, many people share information on YouTube. YouTube is the most commonly used social media platform in Taiwan. YouTube has attracted Taiwanese users because of its rich content. Organizations have already used YouTube as a marketing tool. With the spread of internet use and the rise of the YouTube platform, the YouTuber has become a popular profession among young people. YouTubers’ creations are diverse and free. They have considerable influence on young people. Studies have demonstrated that sponsoring the creators of UGC on YouTube can be an effective way to promote a brand and strengthen respondents’ purchase intentions. YouTubers play an important role in social media marketing, which highlights the rationality of the study.

5.2. Management Implications

Some management implications with managerial contribution were obtained from the study results, which are as follows.

(1)

YouTubers play an important role in hotel marketing, and their identification is a complicated MCDM problem. This study reviewed the relevant literature on YouTuber selection and interviewed 40 managers who had served in hotels for more than 10 years and selected 12 key decision-making criteria. From a managerial perspective, narrowing multiple and complex decision criteria can help managers make decisions more efficiently and correctly.

(2)

From a commercial viewpoint, social media impacts consumers’ decision-making processes when selecting a hotel by influencing their searches, decisions, and hotel reservations. Twelve important decision-making criteria were selected and ranked by the degree of importance, including video title setting ability of the YouTuber > controlling video production time effectively > the content of the video is combined with personal characteristics > the content of the video is close to life > interaction well with audiences > the product is introduced objectively > script creativity > the content of the video is easy to understand > the quality of the video > the number of subscribers to the YouTube channel > the number of likes for the videos on the YouTube channel > the number of views of the YouTube channel. Our results (ranks of criteria) are consistent with a previous study [19]. For example, the criteria “the content of the video is combined with personal characteristics” and “the content of the video is close to life” are both in the top four.

(3)

From a risk management perspective, most managers make decisions based on the information or knowledge at their disposal at the moment the decisions need to be made. Frequently, this practice provides evidence of incomplete information [65]. In other words, managers do not often use solid processes when making decisions, which is a serious and dangerous problem. Through the decision-making model proposed for the scientific methods in this study, managers can effectively make decisions and reduce the risk of corporate decision-making. Furthermore, under the limited budget situation, the case hotel can also choose more suitable YouTubers according to the ranking results to lower the risk of promotion activity.

(4)

It is difficult to imagine the use of the hybrid MCDM model; thus, discussion is needed about practical implementation in the hospitality industry. In practice, this study suggests that hotels can collect data on YouTubers who are suitable for cooperation and classify them into a database according to their characteristics. Afterwards, as long as there are any seasonal or themed marketing activities, managers of hotels can use the decision-making model of this study to identify suitable YouTubers in the database to reduce the risk of ineffective marketing.

(5)

In terms of research topics, identifying YouTubers is important to hotels for sustainable hotel development and to improve their financial condition, but very few studies in the literature have addressed this topic. In the methodological part, to increase the quality of decision-making and accelerate the decision-making process, this study provides an integrated hybrid MCDM model to offer hotel managers the ability to identify the best YouTuber, which has managerial applications and addresses the deficiencies in the methodological results of past studies.

5.3. Research Contributions

Importantly, this study proposes an integrated hybrid MCDM model for organizing FDM, DEMATEL, ANP, and TOPSIS to benefit and identify YouTubers. A case study of a hotel proved the effectiveness of the proposed hybrid model. In summary, from the perspective of sustainable hotel development, the research contributions have the following four key directions.

(1)

Technical contribution: Traditionally, prior research has used other decision-making methods to address MCDM problems; however, this is not sufficient because of the complex background. Thus, this study proposes an integrated MCDM model to match real-life cases. The issue of personnel selection is an important topic for organizational success and belongs to MCDM issues. In the processing procedure, the mutual influence relationship between dimensions was generated based on DEMATEL, and the weights of the dimensions and criteria were estimated according to the mutual influence relationship based on ANP. To reduce the excessive number of PC matrices of ANP and the difficulty for decision-makers to answer the questionnaire in practice, TOPSIS can include the weights of various criteria integrated by ANP, which can help identify the best YouTuber and increase decision-making efficiency.

(2)

Practical contribution: The empirical results can yield advantages and benefits for interested parties. Managers typically make many decisions, some of them being operational and others strategic. Making decisions is a huge responsibility not only for the organizations themselves but also for their employees and other stakeholders. Although there is no systematic structure or selection model for identifying YouTubers for hotel managers, the proposed hybrid model can hopefully serve as a perfect and effective reference to enhance decision-making quality and efficiency.

(3)

Application contribution: Specifically, the proposed hybrid model is rare, and its use is not observed in identifying YouTubers for hotels in the limited literature review. Therefore, this study makes a significant application contribution in addressing this important and fascinating topic for interested parties, such as hotel managers.

(4)

Although the methodological contribution is not the key purpose of this study, the study has many merits of managerial contributions. In particular, for research purpose, this study proposes an integrated hybrid MCDM model to organize four key techniques of FDM, DEMATEL, ANP, and TOPSIS to identify YouTubers for hotels. From the limited literature review, the proposed hybrid model, which can increase the efficiency of the decision-making, has never been seen in the issue of YouTuber identification of hotels; thus, the study provides a superior application contribution to address this important and interesting topic both for academician and practitioner.

5.4. Subsequent Research

The entire research process, from the review of relevant literature to the collection of empirical data and the design of research analysis methods, was pursued with academic rigor. Although the proposed hybrid model performs well in the identification of YouTubers, the following sections can be extended as recommendations for subsequent studies. (1) The survey objects of this research are managers of Taiwanese hotels; thus, the findings may not be applicable to other countries with different cultures and languages. Future studies could collect the opinions of hotel managers in other countries regarding the identification of YouTubers and compare the results. (2) FDM could include the opinions of more industry managers to diminish errors. (3) Future studies may apply another approach to identify the selection criteria for YouTubers, such as importance–performance analysis (IPA). The results of the different methods are also compared. (4) This study considered 12 selection criteria. More relevant criteria should be considered in future studies to construct a complete selection model. (5) Several selection criteria are qualitative and difficult to express accurately. In addition, the criteria or solution scores given by managers are based on subjective judgments. In future studies, ANP and TOPSIS sections should include the concept of fuzzy theory to deeply study the issue of hotels. (6) Researchers could use other MCDM methods to further explore this topic.