Empirical Study for Factors of Customer Satisfaction Based on B2C Online Shopping

Headnote

Abstract: In order to have a comprehensive and deepening study on the "peopleoriented" college student management work, a neat, scientific and complete theoretical framework must be set up. In order to establish this kind of scientific and comprehensive theoretical framework, the core concept of relevant problems must be fully recognized. Defining the "people-oriented" college student management work, going into the nature of student management work and correctly grasping factors that influence people-oriented college student management work are necessary premises to study people-oriented college student management work. After determining the basic frame work of the scientific and comprehensive people-oriented college student management work, we will begin to see the system problem of college student management work and its manifestations. According to the progress of the society and the development of the times, fully understanding about the existing problems in the college student management work and its causes, discussion on the subject of people-oriented college student management work and fully demonstration of college student management work with people-oriented as the subject are according with era development and realistic requirements. Constructive suggestions are proposed on ways and methods of achieving people-oriented college student management work. Have a profound understanding about student management work and the meaning of student management work to social and individual development. Elaborate the objectives of student management work and provide a scientific and reasonable theoretical way for achieving people-oriented student management work.

Keywords: B2C, Online shopping, Customer satisfaction.

1.Introduction

Electronic commerce specifically B2C (Business to Consumer) has been a popular medium for users to obtain information about products and services, conduct buying and selling transactions (Korner et al., 2000; Nolasco, 2016). Thus, customer satisfaction for B2C website is crucial in attracting online shoppers to visit a cyber store and learn about its products and services and in ensuring repeat purchases. Practically all Internet users are their potential customers (Sá et al., 2016). Whether or not they can convert their potential customers into real ones and retain them depends on the service they offer and the customer satisfaction consumers perceive. Customer satisfaction is a critical issue in the success of any business system, traditi onal or cyber store.

Lots of researchers have presented their ways of the modelling of customer satisfaction. Among them, some studies proposed factors (expectations and disconfirmation) and consequences (intention to purchase) of customer satisfaction. However, these studies were based on traditional channels. Whether or not these results are appropriate for the cyber store of electronic commerce still remains to be seen.

Data mining techniques to customer satisfaction have been become more and more important. Data on customer satisfaction have been discussed in many studies, such as find the path travel mode, the website flow, the product is the most visited, the hot areas of a web page, the customer profile based on these browsing data.

The main purpose of understanding customer satisfaction is to predict the future value of an individual. It is achieved by building a model based on user's demographic characteristics, life-style, and previous behaviour. The model produces information that will focus customer retention and recruitment programs on building and keeping the most profitable customer base. It is called customer profiling. A customer profile is a tool that helps target marketers better understand the characteristics of their customer base.

The main goals and contributions of this paper are intended to explore the factors of customer satisfaction and to examine the relationships between customer satisfaction and the factors of the cyber store. Firstly, we begin by attempting to apply the factors we have adopted from the existing literature on customer satisfaction to the cyber store during our interviews with them. Their ideas as to what really matter to their customers give us valuable data from which our factors of customer satisfaction, through our analysis, are derived. Then a model is established with a series of hypotheses. All of these factors, the model and hypotheses are further examined and analyzed through data mining techniques.

This work is organized as follows. Section 2 describes related work. In Section 3 we explain our research model with hypotheses and data mining techniques. In Section 4 data analysis is given. In Section 5 we conclude this paper.

2.Related work

Despite the increased use of the online shopping as a new shopping channel, many researchers, however, have reported that the number of online shoppers and total sales through the online shopping are still marginal as compared with those in traditional retailing. For example, there are many online bookstores available on the Internet; however, not many users are using the online facilities to purchase books. This might be due to reasons such as poor website design, slow loadings of web pages, unattractive interfaces, or other related matters hindering users to adopt the electronic technology for purchasing purposes (Lohse et al., 1998).

According to Hoffman (Hoffman et al., 1995), website facilities do significantly influence attitudes towards online purchasing intention as users choose to visit a website if the websites fulfils their criteria. Some good criteria that are used in evaluating websites are clarity, accessibility, coverage and reliability.

The concept of customer satisfaction occupies a central position in marketing thought and practice. Bearden think that satisfaction is important to the individual consumer because it reflects a positive outcome from the outlay of scarce resources and the fulfilment of unmet needs (Bearden et al., 1983).

Churchill (Churchill et al., 1983) urged disconfirmation as an intervening variable affecting satisfaction and that the effect of disconfirmation is adequately captured by expectation and perceived performance. They used experimental procedures and processed two types of products: a durable and a nondurable good. The results suggested the effects are different for the two products. Rather, that satisfaction was determined solely by the performance of the durable good.

Tse (Tse et al., 1988) followed the results proposed by Churchill, and investigated customer satisfaction formation. Results of a laboratory experiment suggested that perceived performance exerted direct significant influence on satisfaction in addition to those influences from expected performance and subjective disconfirmation. However, expectation and subjective disconfirmation seem to be the best conceptualization in capturing customer satisfaction formation.

Besides, there is a two measure instrument of customer satisfaction. They include the ratio and difference between perceptions and expectations. Cooper (Cooper et al., 1989) has adapted the SERVQUAL instrument designed by Parasuraman (Parasuraman et al., 1989) and compared these two instruments. The investigation found that the 'ratio of perceptions' and expectations resulted in a scale with lower reliability, lower relative validity, and had dimensions that were more difficult to interpret than the scale developed using the 'differences of perceptions'. SERVQUAL include five dimensions: tangibles, reliability, responsiveness, assurance, and empathy. Further empirical research was also undertaken on other targets: telephone companies, insurance companies, and banks in order to confirm the reliability and validity of the model.

From the perspective of factors and consequences of satisfaction,

Oliver proposed a model that expresses consumer satisfaction as a function of expectation and expectancy disconfirmation. Results have confirmed this concept (Oliver, 1980). Moreover, satisfaction significantly affected customer's attitude and their intention to purchase.

Bearden (Bearden et al., 1983) also studied the same issue. His data obtained from 375 members of a consumer panel in a two-phase study of consumer experiences with automobile repairs and services were used to examine the factors and consequences of consumer satisfaction. The results support previous findings that expectations and disconfirmation are appropriate determinants of satisfaction, and suggest that complaint activity may be included in satisfaction or dissatisfaction research.

In order to investigate the moderating effects of customer satisfaction, Carsky (Carsky, 1989) examined information, prior beliefs, experience and styling preferences for automobiles on consumer satisfaction and intention to repurchase. Data were obtained from experiment and the results indicated that information would moderate satisfaction, but not intention to repurchase.

3.Methodology

From the discussions of related literatures above, we find that few researches have described and verified the relationships between customer satisfaction and its factors for cyber store. In this paper, the methodology on research consists of two major parts: research model and data mining method.

3.1. Research model with hypotheses

Case method based on the literatures begins with a pilot case and is followed by multiple cases. From this method a research model may be derived and some hypotheses proposed. The researchers, using the theories of electronic commerce and marketing, have listed a pool of items on factors of customer satisfaction under online shopping environment.

For example, we have chosen a well-known cyber store in China. This company has operated its business for more than ten years and now has a standard demo store. The store sold includes books, computer hardware, software, CDs, etc. We interviewed thirty staff members of this cyber store. They include the general manager, marketing staffs and technicians. After discussions with these staffs, some additional items that affected customer satisfaction were identified. At the end of interviews, a 12-item instrument was developed using a seven point Likert-type scale ranging from "strongly disagree" to "strongly agree." These items were to be confirmed again by the staff.

Based on the results of the case, the factors of customer satisfaction for cyber store induced five factors: logistical support, technological characteristics, information characteristics, homepage presentation, and product characteristics. Logistical support contained quick response to customers' needs, providing communication channels, quickly delivering goods for customers, and providing after service. Technological factors contained modern computer and network facilities and well-structured information systems. Information factors included reliable output information and secure transaction. Homepage presentation contained providing ease to use interface and detail information of goods. Product characteristics included variety of goods and lower prices for goods. Moreover, to measure customer satisfaction, the researchers applied the overall scale attributed to the marketing research.

From the descriptions above, this paper proposes a research model for customer satisfaction of cyber store shown in Figure 1.

Logistical support, technological characteristics, information characteristics, homepage presentation, and product characteristics, are each held to influence customer satisfaction in Figure 1. Thus, the research model encompasses 5 hypotheses regarding factors of customer satisfaction.

3.2. Data mining techniques

Considering the amount of information available for each customer, it is clearly non-trivial to detect the factors of customer satisfaction manually. Data mining and programming tools help to extract automatically prototypical customer profile from data. They are also useful to visualize non-linear interaction of variables. There are several data mining methods, and determination of which to apply can be decided by the quality of the data, the situation, and the objective.

The most known data mining methods are: Clustering, Classification and regression, Association rule discovery and sequential pattern discovery.

Clustering is a technique that puts similar entities into the same groups based on similar data characteristics and those with dissimilar entities are put in different groups. Similarity is measured according to a distance measure function. The meaning of the clusters is therefore dependent on the distance function used. Thus, clustering always requires significant involvement from a business or domain expert who needs to both propose an appropriate distance measure to judge whether the clusters are useful. Clustering example is shown in Figure 2.

Clustering supports the development of population segmentation models, such as demographic-based customer segmentation. Clustering techniques include k-means or k-nearest neighbours, a special type of neural network called Kohonen net or selforganising maps (SOM).

Classification and regression represent the largest part of problems to which data mining is applied today, creating models to predict class membership (classification) or a value (regression). Classification is used to predict what group a case belongs to.

Regression is used to predict a value of a given continuous valued variable based on the values of other variables, assuming a linear or non-linear model of dependency. Logistic regression is used for predicting a binary variable. It is a generalization of linear regression, because the binary dependent variable cannot be modelled directly by linear regression. Logistic regression is a classification tool when used to predict categorical variables such as whether an individual is likely to purchase or not, and a regression tool when used to predict continuous variables such as the probability that an individual will make a purchase. There are several classification and regression techniques including decision trees, neural networks, Naïve-Bayes and nearest neighbour. Decision trees example is shown in Figure 3.

Association and sequencing tools analyze data to discover rules that identify patterns of behavior, e.g. what products or services customers tend to purchase at the same time, or later on as follow-up purchases. While these approaches had their origins in the retail industry, they can be applied equally well to services that develop targeted marketing campaigns or determine common (or uncommon) practices. In the financial sector, association approaches can be used to analyze customers' account portfolios and identify sets of financial services that people often purchase together. They may be used, for example, to create a service "bundle" as part of a promotional sales campaign.

4.Data analysis

A comprehensive survey was distributed through mail to each person who had purchased goods from cyber store and had agreed to participate in the study. Respondents were voluntary, and people were assured that their individual responses would be treated as confidential.

Eighteen items of questionnaires (twelve items for factors, two items for evaluation of customer satisfaction, and four items for individual data) were sent to respondents. A total of 8900 completed questionnaires (7100 were men and 1800 were women) from 12000 respondents were returned to the researchers. The response rate was 74%. The average age is about twenty-five.

From the research model of Figure 1, we induce five factors as follows:

1. Logistical support: There were four indicators in this study. These indicators included: quick response to customers' needs, providing communication channel for customers, quick delivery of goods for customers, providing after service.

2. Technological characteristics: Two indicators of technological characteristics were included in this study: the importance of providing modern computer and network facilities, well-structured information systems.

3. Information characteristics: It was measured by two item scale. This scale included: reliable output information, and the importance of secure transaction.

4. Homepage presentation: Two indicators of homepage presentation were included in this study: ease to use the interface, and detailed information of goods for homepage presentation.

5. Product characteristics: This measure included: variety of goods, and lower prices for goods.

For the evaluation of customer satisfaction, there were two overall indicators: confirmation of customers' needs, and high degree of satisfaction. All items used a seven point Likert-type scale ranging from "strongly disagree" to "strongly agree." Single item questions were used to ascertain respondents' gender, age, education, and which goods the customers purchased.

Using the sample of 8900 responses, the data were examined by data mining methods. Three different data mining methods were used during the process. Decision tree was used to classify the customers into two groups. The training set consisted of both thousands of customers who satisfy the cyber store and thousands of customers who didn't. With the decision tree model the customers were flagged according to their membership in the group of likely satisfied customers.

Sequential pattern finding was used to find out the frequent sequence of events that preceded the customer satisfaction.

Clustering was finally used to cluster the customers into groups with similar characteristics. The clustering found 12 clusters of customers.

We ended up using classification and regression tree (CART) for prediction. After the training the tree had 29 leaf nodes, which were considered as separate customer segments. Each segment was described by the rules produced by CART. The maximum predicted rate in the leaves was 84 % and the cumulative lift chart can be seen in Figure 4.

Prediction of factors of customer satisfaction is the next phase in profiling of the online customers. Neural networks are good tools for this kind of tasks and they were trained with online activity of the customers to discover patterns in their behaviour. In addition to give a better knowledge of the existing customers, neural networks can be used to target more efficiently new visitors of the website. Finally, a sensitivity analysis of the neural network gives information about the factors affecting the customer satisfaction. In this particular case the most important factors were quick response to customers' needs, ease to use the interface and products price. Thus, the findings of the sensitivity analysis agree with the previous analyses.

5. Conclusions

The discussion of customer satisfaction in the paper is based toward traditional channels, and there has been little research to explore and examine customer satisfaction and its factors for cyber store on the Internet. We contribute to this area by building a model of customer satisfaction for cyber store and proposing five validated hypotheses. Moreover, a survey was used to examine these hypotheses and found that the factors and customer satisfaction are significantly correlated. At last, we use data mining methods to verify our model.

References

References

Bearden O., Teel J. (1983). Selected Determinants of Consumers Satisfaction and Complaint Reports, Journal ofMarketing Research, 1, 21-28.001:10.2307/3151408

Carsky, M. (1989). Moderating the Effects of Point-Of-Sale Information and Consumer Satisfaction, Journal of Consumer Satisfaction, Dissatisfaction, and Complaining Behavior, 2, 111-119.

Churchill A., Surprenant C. (1982). An Investigation into the Determinant of Customer Satisfaction, Journal ofMarketing Research, 4, 491-504. DOI: 10.2307/3151722

Cooper, R., Cooper B., Duhan F. (1989). Measurement Instrument Development Using Two Competing Concepts of Customer Satisfaction, Journal of Consumer Satisfaction, Dissatisfaction, and Complaining Behavior, 2, 28-35. DOI: 10.1590/ S1807-76922007000200002

Hoffman L., Novak P., Chatterjee P. (1995). Commercial scenarios for the Web: opportunities and challenges, Journal of Computer-Mediated Communication, 1, 23-45. DOI: 10.1111/j.1083-6101.1995.tb00165.x

Korner V., Zimmermann H. (2000). Management of customer relationships in business media, Electronic Markets, 10, 162-168. DOI: 10.1080/10196780050177062

Lohse L., Spiller P. (1998). Electronic shopping: the effect of customer interfaces on traffic and sales, Communications of the ACM, 41, 81-87. DOI: 10.1.1.46.9592

Nolasco, Karina Abigail Aldape; Almazán, Demian Abrego; Quintero, José Melchor Medina. (2016). Análisis de la percepción de uso de las redes sociales como herramienta de marketing en las MiPYMESde Tamaulipas, México, RISTI - Revista Ibérica de Sistemas e Tecnologias de Informagäo, (18), 49-65.

Oliver, L. (1980). A Cognitive Model of the Antecedents and Consequences of Satisfaction Decisions, Journal of Marketing Research, 4, 460-469. DOI: 10.2307/3150499

Parasurman, A., Zeithaml, A., Berry, L. (1988). SERVQUAL: A Multiple-Item Scale for Measuring Consumer Perceptions of Service Quality, Journal of Retailing, 1, 12-40.

Sá, F., Rocha, Á., & Cota, M. P. (2016). From the quality of traditional services to the quality of local e-Government online services: A literature review. Government Information Quarterly, 33(1), 149-160.

Tse K., Wilton C. (1988). Models of Consumer Satisfaction Formation: An Extension, Journal of Marketing Research, 2, 204-214. DOI: 10.2307/3172652