Introduction
The constant technological evolution of which we are a part has demonstrated the importance of studying and analyzing data to obtain information that supports day-to-day decisions. For businesses, social media, referral sites, and opinion-gathering tools have played a key role in this information gathering process. Now, customer opinions can be analyzed through text mining, allowing them to know and measure the level of customer satisfaction or dissatisfaction with a product or service.
In the same way, this information from texts allows one to know the general experience that one has with companies, recognizing possible points of improvement and anticipating future situations that may compromise the proper functioning of the businesses. There is no place for business decisions based on loose guesswork in a highly competitive market. Different data must be gathered and managed to help develop future strategies. Only in this way can products and services be personalized, and the opinion of customers will influence this because it is an extremely important resource that should be used to improve their experience and, at the same time, will serve to adjust the actions of the company.
The tips will help you pinpoint exactly where the budget should be allocated to get the best ROI. At the same time, this will help create a better user experience beyond the tangible, since the customer not only buys a product, but also thinks about obtaining comfort, security, status, affinity, style, etc. Therefore, if companies focus on delivering the best experience in every moment of truth, their customers will remain loyal to the brand (Smilers, n.d.). There is no doubt that the best way to deliver a memorable experience is by asking customers what they like and how the company can improve. Likewise, it is text analysis that helps measure customer satisfaction and is directly related to many benefits, such as higher market share, lower costs, or higher revenue.
It is important to know that there is a close relationship between customer satisfaction and business performance. Therefore, ensuring that customers are happy with the product or service offered is essential, and the best way to achieve this is by knowing their opinion. Using questions based on ratings, you can estimate the level of satisfaction and, consequently, predict the performance of your company. Text analytics is a reliable source of information and today, thanks to social networks, the customer now relies less on commercials and advertisements in general; instead, the opinions and comments of a customer who has already used the product and service are a much more reliable source of information (Alaei & Stantic, 2019).
It is for this reason that it is important to analyze all those comments, opinions, and reviews, which allow us to identify trends and behaviors. In recent years there have been many changes, some of them due to information and communication technologies and others to environmental factors that, jointly or in parallel, have set new trends and consumption patterns such as new lifestyles, habits of consumption, places of purchase and globalization, among others (Barrullas, 2016).
Today, it is essential to know the behavior of customers when acquiring products or services, the purchase decision process and the internal or external variables that shape their behavior. We are facing a society where information and knowledge influence consumer behavior, so it is important to focus on knowing the interaction variables. The interest of a customer no longer lies in the mere act of purchase but goes further. You should never ignore or underestimate the voice of customers, so you must learn to collect data using different tools and resources, because their opinion at all levels (corporate and in all departments of the company) will help develop products to improve the service and to manage their satisfaction.
General Objective
Develop a sentiment analysis using text mining to measure customer satisfaction for an American hotel chain.
Specific Objectives
The first concerns the collection of feedback from customers. The second refers to ordering and cleaning the database obtained to avoid errors in the analysis. The third refers to using SAS Enterprise Miner® software to perform sentiment analysis. And the fourth refers to the analysis of the results obtained, their description and discussion.
Theoretical Framework
What do Hotels do with Guest Reviews when the Room is delivered, and how has that Review been Measured?
Knowing your customer is just as important anywhere in the world as it is at home. This is a relevant principle to consider in the tourism industry. Each culture has its logic, and within that logic are real, sensible reasons for the way travelers do things. Tourism is currently one of the most important economic and cultural activities that a country or a region can count on. We understand tourism as all those activities that have to do with knowing or enjoying regions or spaces in which one does not live permanently. Tourism can present many variants since there are different types of tourism: cultural tourism, adventure, entertainment, relaxation. In the same way, there are also different people who carry out different types of tourism: tourism for young people, families, the elderly, couples, friends, etc. According to the World Tourism Organization, the volume of this industry is equal to or even greater than that of exports of oil, food products, or automobiles. Tourism has become one of the main actors in international trade and represents at the same time one of the main sources of income for many developing countries. This growth goes hand in hand with an increase in diversification and competition between destinations.
In the age of the internet, the activity of a company depends directly on the opinions of customers, so the comments and reviews of users represent a type of evaluation of the activity of a company. Potential customers look at reviews to rate a brand. This means that customer opinions influence a company’s sales (Soleymani et al., 2017).
Today, thanks to various tools such as the internet, smartphones, social networks, among many more it is possible to know in a matter of minutes if the services offered are excellent, good, fair, or bad, even if the place is on the other side of the world. It is easy to know if the business is a store, a restaurant, a hotel, etc. It is enough to “google it” to obtain thousands of reviews from people who have visited them, narrating their experiences. When choosing a lodging, the first thing you want to know is if the place is a good choice. In most cases, certain factors are more important than others, depending on the guest, such as, the cleanliness of the room, if the bed is comfortable, if there is hot water in the bathrooms, the attention of the staff, if there is good Wi-Fi network, the area in which it is located, among several other factors.
A few years ago, hotels analyzed the experience of their guests only through the questionnaires that they answered at the end of their stay. It is very different today, as social media came to change this process the popularity of online reviews and the importance of people’s opinion on these platforms is changing the way hotels handle their feedback.
According to Weed (2013), hotels want to know the opinion of their customers about their stay. They no longer consider that it is enough to leave a questionnaire in the room and wait for a response, because of sites like TripAdvisor, specialized in rating, publish customer satisfaction, or dissatisfaction on the web. Therefore, hotels must further consider customer satisfaction as the most important part of their business. The data that hotels collect affects both the way they treat their guests in general and the way they interact with individual travelers. Some questionnaires are now shorter, allowing guests to complete them quickly and send them to mobile devices to be completed by customers traveling in taxis or waiting at the airport.
New methodologies have made use of emails to question clients about their stay; however, they can seem annoying as an additional email is usually sent if they don’t reply the first time. In this way, hotels control what is said about them on social media and travel websites.
SiteMinder (2019) has published eight recommendations for handling customer reviews that hotels can rely on to prevent these reviews from being a detriment rather than a benefit. These points are the following:
-
Capture guest complaints before they are made public. For example, having a feedback system in place will give you access to real-time feedback, allowing you to respond to issues as they arise.
-
Use social listening tools or a hotel reputation management system with real-time alerts to keep up with what guests are saying on review sites, blogs, and social media.
-
In responding to a complaint, you must acknowledge the bad experience, and follow up with the customer as necessary.
-
Respond to problems as quickly as possible. The longer they remain unattended, the more they could potentially be out of control.
-
Act on promises to clients and move conversations off the internet when necessary to provide a higher level of personalized service.
-
Limit the number of different staff who respond to reviews.
-
Share customer feedback across the company. Company staff are the front-line brand ambassadors, but only if they understand the challenges and are incentivized to create the best possible customer experience.
-
Never assume a defensive posture.
Text Analysis Applications and Effectiveness in Results
The technological development that has been achieved in the last decade has generated an accelerated production of digital information that can be processed, to solve problems and challenges in any situation. Exploring and exploiting the contents of digital textual documents is the work of text mining, with the goal of obtaining relevant information and, in this case, taking future actions to always be improving.
Also, in the last decade there has been a growing interest in the study of customer opinion regarding the services or products provided to offer a better business strategy focused on the experience generated to customers. Opinions are important to almost all human activities because they influence our behavior. When a person decides, he usually considers the opinion of third parties, to ensure that the decision is adequate or correct. The following table shows examples of how industries in some sectors have taken advantage of the customer opinion study (see Table 14.1).
Table 14.1:Examples of customer opinion study.
| Sector | Description |
|---|---|
| Hotel | Case 1: Hyatt Hotels uses text boxes for open questions, numerical scales and check boxes to tabulate and analyze the final opinions of customers in the surveys they take at the end of their stay, where they classify how positive or negative the comments are depending on the words they use, the exclamation points, etc. They also extract comments from Facebook, Twitter, and other social networks, since the first thing other people who want information about the hotel will do is go to the web and the comments. By analyzing reviews, staff can assign them a room away from an elevator based on feedback from a previous questionnaire. Case 2: An urban hotel that heard noise complaints added earplugs to its guest amenity kits. Case 3.: A hotel restaurant that heard complaints about delays in deliveries, decided to increase the number of employees working that area. |
| Airlines | Based on customer survey feedback on overnight flights, Israel Airlines (El Al) combined food and beverage service so that passengers could finish meals more quickly and sleep earlier. |
| Cruise ships | The most common customer complaint at Crystal Cruises centered on costly and confusing pricing for internet access, so pricing was simplified. |
| Computers | Dell Technologies, through open-ended questions, analyzes customer feedback on product and service satisfaction, customer service, and marketing. This company collects, almost in real time, messages, and comments from customers to later carry out a biannual analysis of social networks to detect eventual changes in satisfaction, analyzing more than 30,000 customers on average. In this way, the organization can quickly connect with a customer and offer solutions to the problems detected, avoiding the generation of a “wave of dissatisfaction” that is more difficult to control and correct. |
| Footwear | The Grasshoppers company incorporated the assessment of texts written by customers into its analytical strategies to provide a better shopping experience and identify areas of opportunity. For this, it was decided to launch a mobile application where customers could register, search for products, buy products and give their opinion on them. The launch of the mobile application aims to record the information used by customers, as well as to integrate a series of innovative tools to position Grasshoppers in the new business technology environment. |
Source: Own elaboration with information from Weed (2013), Trujillo (2017), and Panico (2018).
The previous examples show the particular interest that is beginning to be had in certain companies to take advantage of the information generated in business. In the hotel industry, the comments made by customers are much more noticeable because the services and products offered are highly competitive, so each hotel must distinguish itself. Customer satisfaction has become a key factor in measuring the competitiveness and success of a company; thus, the analysis of raw data from individual consumers can lead to completely new ways of understanding consumer behavior and formulating marketing strategies. Unlike a uniform method, big data takes advantage of various types of analytical tools to use big data to predict trends and obtain commercial value (Ang, Seng, Zungeru, & Ijemaru, 2017).
Methodology
Transformations and Assumptions Carried out to Develop the Text Analysis in the Hotel Chain
In recent years, companies such as Google, Facebook, Amazon, and Netflix have had exponential growth due to the strategic use of technological and digital tools, such as web pages, software, social networks, but above all the use of data. A good data analysis allows the company to know the situation of a company, helps to make correct decisions based on the information found in the data, allows to know which processes are less efficient than others and an infinity of needs that are not noticed at first glance.
The objective of this project was to carry out a sentiment analysis to measure customer satisfaction of an American hotel chain. Comments from the TripAdvisor.com website were used, which were originally extracted by the Information Systems and Databases Laboratory of the University of Illinois at Urbana-Champaign and later presented as a case study (Van der Aalst, Iriondo & Van Zelst, 2018). These opinions were provided by guests after their visit. To fulfill this purpose, all comments contained in the text files, which were provided to us in a notepad format, were compiled into a single database in Microsoft Excel®.
Methodological Development
In total 14,421 comments were collected, which were categorized according to the rating provided by customers on a scale of 0 to 5, with the interval from 4 to 5 being positive comments, the interval from 2 to 3 corresponding to neutral comments and finally from 0 to 1 were considered negative comments. To finalize the data cleaning process, comments in languages other than English or with input errors that made them impossible to understand were eliminated.
Finally, the reviews were ordered according to the date and the place they came from, this was to ensure the quality of the information and avoid errors when generating a value proposition with the information derived from the data. The comments collected from 2014 to 2017 were analyzed, comprising a total of 9,332 comments made from hotels in three cities, San Francisco, Seattle, and Los Angeles. Since the database was clean and orderly, the SAS Enterprise Miner® software was used to perform a sentiment analysis. Nodes were used that allowed ordering the comments, eliminating words without relevant contribution to the analysis, highlighting the topics that were mentioned most frequently by the guests, categorizing, and identifying the existing relationships between the words used by the clients according to the type of comment (positive, neutral, or negative) and have a complete identification of the causes of these. This process was carried out individually with the three categories of comments.
The main elements of the process are described below:
-
Import file: Allow loading the clean database and begin the analysis.
-
Analysis of the text (parsing of the text): Fulfill the function of eliminating words that do not add value to the analysis and allows separating them by category, be it verbs, nouns, adverb, etc.; in the same way, it allows to ignore auxiliaries, conjunctions, interjections, prepositions, pronouns, numbers, punctuation marks, etc. Add a list of stop words in English for more accurate information.
-
Text filter: Apply filters to reduce the number of terms or documents, this allows you to check spelling, add dictionaries and with the results you can perform a deeper analysis, allowing you to see the link between concepts or even search all documents for a single word, filtering the texts only where it appears (see Figure 14.1).
Source: Own elaboration with SAS Enterprise Miner® software, Spanish Version
Figure 14.1: Text filter node properties.
-
Text cluster: Allow grouping by similarity between words, in the same way by trial and error it can be adjusted, so that the topics become more and more specific. As shown in Figure 14.2, it was decided to leave eight clusters with 15 descriptive terms each, in which you can see what the guests talk about the most.
Source: Own elaboration with SAS Enterprise Miner® software, Spanish Version
Figure 14.2: Clusters resulting from the analysis.
-
Text theme: Create themes in a group of documents. Depending on the database used, by trial and error it can be adjusted so that the topics are more specific each time providing a more precise analysis.
Finally, the interactive filter viewer was used, which allowed identifying the relationships of each word and determining the following information for each city:
Los Angeles
The positive comments about the Los Angeles hotel were mainly derived from the good condition of the rooms. Guests mentioned as positive points the comfortable bed found in most of the rooms, the large closet, the marble bathroom and, in general, the decoration of the entire room. They also highlighted the amazing room service provided by the always friendly and helpful staff, the fabulous food from the hotel’s well-rated restaurant which guests note has an incredible pool view. Finally, they highlighted the spacious and comfortable hotel, the free parking, and the location, since the hotel is in the heart of Hollywood, obtaining a spectacular view (see Figure 14.3).
Source: Own elaboration with SAS Enterprise Miner® software, Spanish Version
Figure 14.3: Relationships of the positive terms in the Los Angeles hotel.
As neutral comments, the small size of the pool, the high prices for the SPA service, and the lack of a Jacuzzi and a quiet area where you can study or meditate were mentioned. The negative comments came from unpleasant experiences due to lack of water pressure in the showers, little hot water, and bed bugs. Likewise, customers reported having problems with reservations and the check-in process, as well as bad experiences with the smell and cigarette smoke in the corridors and rooms, even though smoking is prohibited within the hotel’s facilities. Likewise, guests emphasize that too much noise is filtered into the rooms from the corridors and neighboring rooms (see Figure 14.4).
Source: Own elaboration with SAS Enterprise Miner® software, Spanish Version
Figure 14.4: Relationships of negative terms in the Los Angeles hotel.
Seattle
As seen in Figure 14.5, the hotel located in the city of Seattle was praised for the cleanliness of the rooms, their comfort and spaciousness, as well as the design and function of the showers. Guests applaud the hotel’s excellent breakfast, referring to it as varied and incredibly delicious. The location was constantly highlighted, being an important factor due to the number of restaurants and activities to be carried out near the hotel, in addition to emphasizing the colorful area of the city in which it is located. Finally, the guests appreciated the friendly and helpful treatment of the staff, the beautiful decoration and the great atmosphere that exists within the hotel.
Source: Own elaboration with SAS Enterprise Miner® software, Spanish Version
Figure 14.5: Most frequent positive hotel comments in Seattle.
Neutral comments mentioned the lack of cleaning carried out in the common areas and that, even though the hotel is somewhat old, it is kept in sufficient condition to continue operating. As a negative comment they mentioned the constant noise coming from the nearby highway, the garbage trucks, the police sirens and, above all, the large number of homeless people who are on the outskirts of the hotel that, despite being harmless, can end up being annoying to guests and affect the hotel image. Extra costs for internet and gym services were also a recurring theme. The restaurant was also listed as very small and cluttered, as most of the guests want to take advantage of the previously mentioned excellent breakfast (see Figure 14.6).
Source: Own elaboration with SAS Enterprise Miner® software, Spanish Version
Figure 14.6: Relationships of the negative terms in the Seattle hotel.
San Francisco
At the San Francisco hotel, most guests highlighted the exceptional attention of the lobby staff, the excellent and central location near most of the tourist spots and a variety of restaurants around, but above all, the city and its fantastic streets with innumerable places to visit and with many attractions. Generally, neutral reviews mention that the hotel was only suitable for short stays, preferably one night. Most guests who provided neutral feedback mentioned that they were visiting the hotel for the first time. It is noteworthy that many guests mentioned the rooms as terribly small, with many failures to maintain the shower and with old and uncomfortable furniture. Despite being central, the location was classified as bad because it is dirty and full of people in the street, which is noteworthy since guests mention that the hotel is in a rather expensive area (see Figure 14.7).
Source: Own elaboration with SAS Enterprise Miner® software, Spanish Version
Figure 14.7: Relationships of negative terms in the San Francisco hotel.
Specific Recommendations for Each Case
Since a specific analysis was carried out for each of the cities, the following recommendations are proposed to each of the hotels to improve these negative comments.
Los Angeles
The main recommendation for the Los Angeles hotel is to verify that the maintenance and cleaning process is being carried out correctly, the constant spraying for pests. Being in constant vigilance to prevent guests from not abiding by the rules and smoking inside the hotel and, verifying the correct functioning of the reservation systems and the ability of the people in charge of checking-in.
Seattle
In the Seattle hotel, the recommendations would be to eliminate the cost for Wi-Fi and gym services; create a more business hotel concept where guests only must spend one or a maximum of two nights in it; due to the continuous noise outside the rooms, provide free ear plugs; be in constant vigilance of cleanliness in common areas; expand the restaurant due to the success of the breakfasts it includes, but as guests mentioned, space is limited.
San Francisco
The recommendation that is given to the hotel located in San Francisco is a constant maintenance in the showers because it is a complaint with a lot of weight on the part of the guests; remodel the hotel property for one that is warmer and cooler; hire a cleaning company for public spaces to keep the area where the hotel is located clean, because for the cost the guests are paying, they deserve to have a clean space.
Conclusions and General Recommendations to Improve the Measurement of Customer Opinion Through Text Analysis
As a general recommendation, it is suggested that for future occasions in all hotels, a follow-up is carried out with each of the clients mainly with those who leave negative comments, letting them know that their opinion was and is very important for the company and that the necessary actions will be taken to solve the problems that may have arisen during your stay.
On the other hand, it is important to have all the updated guest data, so it is recommended to include sections in the satisfaction survey that request the guest’s name, age, and email, always letting them know that the company considers the laws of data protection. These surveys must be carried out electronically at the time the client performs the check-out process if it is intended to reduce the time of data capture and cleaning. An electronic form connected to a database that requests all this information would be sufficient, if the customer is identified with a unique code. This database could be connected to the reservation database to find out more information about your stay and thus offer even more personalized attention, so it is also suggested to properly identify which hotel each comment comes from to facilitate decision-making in the future.
Through this analysis, many happy clients can be identified, but also many were dissatisfied with their stay; but what if the client can present their complaints before checking out? Hotels are recommended to install a tool that allows guests to always express their opinion. When connecting to the Wi-Fi, and with their reservation number, the client will be able to connect to the network. If they accept, provide their data to a form that will request the software automatically, allowing the client’s profile to be identified and subsequently segment them to creating marketing campaigns and improving strategies when it comes to building customer loyalty, always complying with the general data protection regulations. In this way, if the client has a problem when checking-in, with this tool the guest’s problem could be solved in a matter of minutes, which would avoid bad reviews when checking-out.
Additionally, in any case it is important to keep in mind the Global Code of Ethics for Tourism, which is a set of principles designed to guide governments, tourism companies, communities and tourists in general, to maximize the benefits of the sector while minimizing possible negative consequences for the environment, cultural heritage and societies. The 10 principles of the code cover the economic, social, cultural, and environmental aspects of the travel and tourism sector worldwide (World Tourism Organization, 2017):
-
Article 1: Contribution of tourism to mutual understanding and respect between men and societies.
-
Article 2: Tourism, an instrument of personal and collective development.
-
Article 3: Tourism, a factor of sustainable development.
-
Article 4: Tourism, factor of use and enrichment of the cultural heritage of humanity.
-
Article 5: Tourism, beneficial activity for the countries and communities of destination.
-
Article 6: Obligations of tourism development agents.
-
Article 7: Right to tourism.
-
Article 8: Freedom of tourist travel.
-
Article 9: Rights of workers and entrepreneurs in the tourism sector.
-
Article 10: Application of the principles of the Global Code of Ethics for Tourism.
In recent years, tourism has experienced continuous growth and profound diversification, which has made it one of the fastest growing economic sectors in the world. In addition, tourism is closely related to the development of communities, which is why it has become a key driver of economic progress. Despite the recent economic crises in the United States, Europe and emerging countries, tourism has registered virtually uninterrupted growth in the last five years. Currently, tourism contributes 10 percent of world GDP (Isik, Dogru, & Turk, 2018).
As a conclusion to the text mining process, it is important to note that this analysis, and the information from it, demonstrate the importance of collecting opinions and reviews from customers. This compilation is not only useful for the hotel industry, but for all businesses that seek to avoid or correct errors, as well as identify possible points for improvement in all their processes, allowing them to better know their consumers and offer products or services that meet all their expectations and needs.
References
Alaei, A. R., Becken, S., & Stantic, B. (2019). Sentiment Analysis in Tourism: Capitalizing on Big Data. Journal of Travel Research, 58(2): 175–191. →
Ang, L. M., Seng, K. P., Zungeru, A. M., & Ijemaru, G. K. (2017). Big Sensor Data Systems for Smart Cities. IEEE Internet of Things Journal, 4(5): 1259–1271. →
Barrullas, J. (2016, October 14). El comportamiento del consumidor y las nuevas tendencias de consumo ante las TIC. Economía y Empresa: Blog de los Estudios de Economía y Empresa. https://economia-empresa.blogs.uoc.edu/es/consumidor-y-tendencias-consumo-tic/ →
Isik, C., Dogru, T., & Turk, E. S. (2018). A Nexus of Linear and Nonlinear Relationships Between Tourism Demand, Renewable Energy Consumption, and Economic Growth: Theory and Evidence. International Journal of Tourism Research, 20(1): 38–49. →
Panico, C. (2018). La eficacia del análisis de sentimientos para la empresa: el caso de estudio Dell Technologies Inc. [Unpublished dissertation]. Universidad Complutense de Madrid. https://www.ucm.es/data/cont/docs/758-2019-01-04-TFG_Panico_Chiara_TFG.pdf →
SiteMinder. (2019, May 24). Hotel Reviews: How to Manage Online Guest Reviews at Your Property. https://www.siteminder.com/r/hotel-reviews-manage-online-property/#hotel-guest-reviews-how-to-handle-them-at-your-hotel →
Smilers (n.d.). 8 razones por las que es importante conocer la opinión de tus clientes. Smilers. Retrieved January 9, 2021, de https://smilers.co/8-razones-por-las-que-es-importante-conocer-la-opinion-de-tus-clientes/ →
Soleymani, M., Garcia, D., Jou, B., Schuller, B., Chang, S. F., & Pantic, M. (2017). A Survey of Multimodal Sentiment Analysis. Image and Vision Computing, 65: 3–14. →
Trujillo, J. C. (2017). Análisis de sentimientos en entornos de mercadeo móvil [Unpublished dissertation]. Universidad de San Buena Ventura Seccional Cali, Colombia. http://bibliotecadigital.usbcali.edu.co/bitstream/10819/5568/1/Analisis_sentimientos_entornos_mercadeo_2017.pdf →
Van der Aalst, W. M. P., Iriondo, A. B., & Van Zelst, S. J. (2018). RapidProM: Mine Your Processes and Not Just Your Data. In M. Hofmann, & R. Klinkenberg (Eds.), RapidMiner: Data Mining Use Cases and Business Analytics Applications (2nd ed.). (Chapman & Hall/CRC Data Mining and Knowledge Discovery Series). Chapman & Hall/CRC Press, London, UK. →
Weed, J. (2013, May 27). Checking In After Checkout. The New York Times. https://www.nytimes.com/2013/05/28/business/hotels-work-harder-to-collect-customer-responses.html a, b
World Tourism Organization. (2017). Global Code of Ethics for Tourism. World Tourism Organization. Madrid. Retrieved on April 1, 2018 from: http://cf.cdn.unwto.org/sites/all/files/docpdf/gcetbrochureglobalcodees.pdf →