Using Closed-loop Feedback to improve NPS

Rijksuniversiteit Groningen

Using Closed-loop

Feedback

to

improve NPS

A study how customers’ willingness to recommend can be increased due to Closed-loop Feedback interference

By Gerrit Huisman 1533991 16-4-2013

- 2 -

Using Closed-loop Feedback to improve NPS

A study how customers’ willingness to recommend can be increased due to Closed-loop Feedback interference

Hand-in date: 16-04-2013

University: Rijksuniversiteit Groningen

Department of Marketing, Faculty of Economics and Business Program: Master thesis Marketing Management

First supervisor: prof. dr. P.C. (Peter C.) Verhoef Second supervisor: dr. J. (Jenny) van Doorn

External supervisor: dr. L. (Linda) Teunter

Name: Gerrit Huisman

- 3 -

A

CKNOWLEDGEMENT

This thesis is the final step in the achievement of a degree in Marketing for Marketing Management at the University of Groningen. Next to that, it is the end of my student-life that I spent in Groningen. It has been a very good time and I am glad I could develop myself on both personal level and educational level.

I want to thank my thesis supervisor Professor Peter Verhoef for his valuable feedback and suggestions. During the process, he has been of constant help. Furthermore, I want to thank the University of Groningen for giving me the opportunity to take part in their program.

Finally, I want to thank my heit and mem for their fully support during my study. And to my brother and best friend, Arjen, thanks for all your support.

- 4 -

E

XECUTIVE SUMMARY

Many companies have implemented the Net Promoter Score in recent years and base strategies on this outcomes. A new-developed tool related to the Net Promoter Score is Closed-loop Feedback. This tool aims to increase customers’ willingness to recommend after the company and customer interact with each other. In this thesis it is hypothesized that Closed-loop Feedback indeed increases the NPS. Furthermore, this thesis shows the impact of several drivers, such as relationship length, customer value and moments of truth, on CLF and NPS. Data were collected through an Internet survey of a Dutch energy company.. In addition, especially if customers have a long-term relationship with the company or with a high customer value their willingness to recommend increase after CLF. With regard to moments of truth, especially complaint handling is important and can increase customer’s willingness to recommend.

- 5 -

T

ABLE OF

C

ONTENTS

Acknowledgement ... - 3 - Executive summary ... - 4 - 1 Introduction ... - 7 - 2 Theoretical Background... - 11 - 2.1 Closed-loop Feedback ... - 11 -

2.2 Net Promoter Score ... - 14 -

2.3 What are interactions and ‘moments of touch’? ... - 15 -

2.4 What are consequences of interactions and why is it important? ... - 16 -

2.4 Optimize contact moment and maximize customer experience ... - 18 -

2.5 Customer value... - 20 -

Chapter 3 Methodology ... - 22 -

3.1 Research design ... - 22 -

3.2 Data collection ... - 22 -

3.3 Data exclusion ... - 22 -

3.4 Dependent and independent variables ... - 23 -

3.4.1 Dependent variable ... - 23 -

3.4.2 Independent variables ... - 23 -

3.5 Analysis Method ... - 24 -

4. Results ... - 26 -

4.1 Descriptive statistics ... - 26 -

4.1.1 Average NPS increase per independent variable ... - 28 -

4.1.2. Normality ... - 29 -

4.1.3 The effect of CLF on NPS ... - 30 -

4.1.4. Correlations ... - 30 -

4.2 Regression analysis ... - 31 -

4.2.1 Hypotheses ... - 35 -

4.3 Main Findings Summarized ... - 37 -

Chapter 5 Discussion and Conclusion ... - 38 -

5.1 Managerial implications ... - 39 -

Chapter 6 Limitations and Further Research ... - 41 -

- 6 -

- 7 -

1

Introduction

“COMING TOGETHER IS A BEGINNING.KEEPING TOGETHER IS PROGRESS. WORKING TOGETHER IS SUCCESS.”

– HENRY FORD

This quote of Henry Ford expresses the importance of working together. In today’s businessworld companies are focused to create a relationship with a customer. Continuousl investments in this relationship are important for companies to be successful. In the past decade companies understood this message and investedwe more in customer relationship management tools. More and more companies succeeded in retaining their customers by just asking them for simple feedback. They often used a simple but effective question developed and tested by Reichheld (2003): “How likely is that you will recommend [company X] to a friend, relative, or colleague?” This question which results in a Net Promoter Score (NPS) is an effective prediction of customers’ behavior and loyalty to the company. Despite the fact that many companies use this metric to get insights into their customers’ behavior, in science criticism towards this metric has been revealed. The study of Keiningham et al. (2008a) doubts if a single metric can measure all aspectsloyalty. Moreover, Wiesel et al. (2012) share their concern about the increasing popularity of single-question customer metrics. As they acknowledge the predicting power of NPS to individual customer’s attitudes, in their belief companies should consider a more nuanced multi-dimensional approach. By hook or by crook, it’s never been more important to keep the customers you already have, because it’s much cheaper than acquiring new ones.

- 8 -

and possibly fine-tune product and service offerings. It’s up to the companies to find out how to make these improvements happen at each process, product and employee level. Basically the CLF approach consists of four steps. First, customer feedback data should be collected. Second, learn lessons of the feedback and make changes if necessary. Third, communicate these changes to both the customers and employees and ask for their feedback again. Fourth, refine changes if this support the business. This is called event-driven feedback and the NPS is based on contact with the customer (Markey et al., 2009). CLF can track customers individually. In this tool the contact between the company and customer is evaluated, and customers are asked about their experiences. This happens on a regular basis which means customers are asked to fill in the survey and give feedback several times. Because customers give feedback at different moments companies can investigate if service and product adjustments have the desired effect. Every time a customer fills in a survey the NPS question is asked. Aim of CLF is to track if customers will recommend the company more after CLF intervention than before the CLF intervention. One of the characteristics of CLF is the velocity of the provided results. Mostly within one day results are provided and therefore the interaction with the customer identifies very quickly new market opportunities and insights in customers’ needs1.

According to Beaujean et al. (2006) companies should excel in customer service when interacting with customers during ‘moments of truth’. Those few interactions (e.g. a lost credit card or investment advice) when customers are highly emotional involved, are opportunities to strengthen the relationship with the customer. Companies can earn trust and loyalty during these moments if customer service is able to handle it exceptionally. This can be done by engaging with providing attention, acquiring and understanding exactly what they want, putting that into action and aftermath / follow up. So the customer feels connected and well looked after during the service experience and in hope, return to the company the next time.

Contacts differ in time (short, medium, long), in information richness (low, medium, high), direction of flow (customer to server, server to customer, bilateral), and value of

1

- 9 -

the exchange (low, medium, high) (Kellog, 2000). Venkatesan and Kumar (2004) add another three contact modes and its antecedents, namely rich (e.g. face-to-face, trading event meetings), standardized (e.g. direct mail, telephone) and Web-based contacts. Consumers who use online communication want efficiency (Venkatesan and Kumar, 2004). However, contacting consumers too much can actually harm your business results as well. As a consequence, customers can be put off from purchasing products (Kumar et al., 2006) or this communication can be detrimental to the relationship (Fournier et al., 1997). There are many moments for an organization to interact with the customer. But there are only a few moments which hold great potential to delight customers. Those moments are so-called ‘moments of touch’ (Markey et al., 2009). Consumer do care about which contact method is used in a certain context (low/high value; low/high time sensitivity). Mail or a service call is most appreciated by customers (Smith and Eroglu, 2009).

In this study CLF and NPS are linked to each other, and will examined if customers recommendation change after CLF interference. CLF is used as a bottom-up approach, which means individual customers give their feedback via a questionnaire. In this questionnaire customers can indicate if they are interested in a company’s call to have a in-depth conversation. After this interaction, companies ask this customer to fill in a questionnaire again. Although publications are found with regard to CLF, in scientific research the influence of CLF on NPS is, to our knowledge, not investigated yet. Hence, this study identifies the next research question:

“ W H AT I S T H E I M P A C T O F CL O S E D-L O O P FE E D B AC K O N T H E NE T PR O M O T E R SC O R E, AN D W H I C H M O M E N T S O F T R U T H S AN D C U S T O M E R S A R E M O S T P R O N E FO R CL O S E D-LO O P FE E D B A C K?”

For this research a Dutch energy company provided customer data. Constructs such as previous NPS, customer value, gender, relationship length, moments of truths and product types will be used to identify the impact of CLF on NPS.

- 10 -

management and marketers to make a next step. In the end it can boost the business both financially and non-financially.

- 11 -

2

T

HEORETICAL

B

ACKGROUND

In this chapter a literature review is presented about Closed-loop Feedback and Net Promoter Score. Furthermore, attention will be addressed to moments of truth, consumer experience management, and several customer characteristics with regard to customer behavior. Hypotheses, based on current theoretical knowledge, will be presented.

2.1 C

LOSED

-

LOOP

F

EEDBACK

Customer satisfaction surveys have become a common source of performance feedback for employees and organizations. For instance, customers assess either the individual service provider (e.g. salesperson) or a group providing a service (e.g. bank). Besides, customers are asked to deliver feedback about the environment where they have been served (Hekman et al., 2010). Berglund and Ludwig (2009) use the definition of Daniels (1994) to define performance feedback as ‘transmitted information about past performance that gives the performer the opportunity to alter their future behavior’. Customers’ feedback can be categorized into two groups, namely active feedback and passive feedback; also called solicited or unsolicited feedback (Wirtz and Lee, 2004). Active feedback, for example, is collected by satisfaction surveys. Passive feedback on the other hand, relies heavily on a customer’s own willingness to report complaints, compliments or suggestions towards the organization. The focus of this study will be on active feedback.

- 12 -

Not only negative responses can lead to improvement in service quality, but also positive response can improve service quality and company’s profits (Huang et al., 2003). Customers make special effort to spread their compliment. Therefore, compliments offer interesting opportunities to initiate and establish long-term relationships. Huang et al. (2003) argues that positive responses not only boost sales, but that they stimulate employee’s morale and service attitude as well.

In general, compliments are an excellent indicator of customers’ satisfaction (Kraft and Martin, 2001). Kraft and Martin (2001) found eight motives for customers’ complimenting behavior. These motives are (1) delight or great satisfaction, (2) dissonance reduction, (3) reciprocity / social norms, (4) improve relationship with a service person, (5) high involvement with product or service, (6) voting behavior to continue special services or products, (7) to buffer complaints & increase effectiveness, and (8) Flattery: to get a tangible reward. Compliments provide specific and highly useful marketing information (Erickson and Eckrich, 2001). Furthermore, these compliments show which service- or product characteristics satisfy and please customers.

Research of Hekman et al. (2010) summarizes that the evaluation of any person, object, or idea is partly based on evaluations of other persons, objects, or ideas with which the target object is linked. In addition, an evaluation of a product, service, or a brand is partly linked to the evaluation of the persons who are connected with the product, service, or a brand.

- 13 -

CLF is a continuous process of collecting customers’ feedback (see figure 1). A well-constructed customer feedback program becomes a channel for acquiring business insights about customers and what is important to them (Neckopulos, 2010). Moreover, Neckopulos claims four key preconditions to create and implement a successful customer feedback program. First, attain a clear business goal that has to be pursued. Second, separate customer noise from customer feedback to make sure the analysis is meaningful. Third, results of customer feedback are serving as basis for improvements in the service process. Fourth, embed the customer feedback loop in the organization culture. Hart et al. (1990) concluded that service can be recovered by telling a customer about an improvement when a problem has occurred. After customers have complained it is important for the organization that they close the loop in order to give customers a positive impression. In addition, Van Vaerenbergh et al. (2012) state that managers who want to maximize the return on their complaint-handling effect should communicate process recoveries to customers. They believe that providing feedback can have significant impacts on customer outcomes as it might increase customer loyalty and create positive word of mouth. Service providers can (re)gain customers’ positive evaluation and trust through effective service recovery efforts (Du et al., 2010).

Every transaction is an opportunity to create a new promoter (Markey et al., 2009). Especially, so-called ‘moments of truth’ are unique moments to delight customers. Service employees start one-to-one conversations with a customer to understand in

NPS Question

Phone call, serve individual customer

Call back if NPS is 0-8;

Analysis Make action plan

Implement action plan

- 14 -

detail what customers’ value and how the company can serve better. Springer et al. (2011) believe that the creation of promoters has to be the main goal of CLF. They argue that with combined forces a firm should delight every customer, eliminate defects in products and interactions, and/or reduce the effort customers spend in dealing with the company. As they state, those arguments are only a part of the package, but should not be the main aim.

2.2 N

ET

P

ROMOTER

S

CORE

In 2003 the Net Promoter Score, a single indicator of customer loyalty, was introduced by Fred Reichheld. He designed NPS to measure the effect of Word-of-mouth (WOM) on sales. A customer is asked how likely he or she will recommend a product or service to a friend or colleague on a 0-10 scale. Scores of nine of ten represent promoters, which are the customers with the highest rate of repurchase and referral. Customers who listed a seven or eight are the so-called passives. Detractors is the group of unhappy customers who rated a score of zero through six. The Net Promoter Score is the percentage of promoters minus the percentage of detractors.

In the view of Reichheld (2003) the biggest advantage of NPS is the simplicity and the ease of measurement. Only one or two questions are the key to measure predicted behavior that can drive growth. Therefore, complex and long surveys are not necessary any longer. In addition, Reichheld (2003) found a correlation between NPS and revenue growth, and concluded that NPS is the best predictor for growth. The author shows that promoters are more likely to repurchase or recommend a brand than detractors. These promoters are interesting customers for a company since companies simply cannot buy recommendations. Detractors can, undoubtedly, hurt reputations and sales even solely based on an isolated encounter with a single employee.

- 15 -

line with this study Hanson (2011) agrees that NPS can be a good metric, but states that companies should consider other metrics as well. Furthermore, this study points out that there can be different scenarios but the same NPS score. Therefore, the NPS score can be misleading in some cases. Wiesel et al. (2012)2 conclude NPS is clearly a good predictor of the individual customer’s attitudes, but ask companies to be careful for what they use their metrics for. In addition, a more multi-dimensional approach to predict customer’s behavior is more reliable (Wiesel et al. (2012).

An outstanding customer experience creates promoters, and promoters are more valuable to a company than other customers. A series of positive interactions build towards higher willingness to recommend, creating a promoter. One negative interaction at a moment of truth can turn an advocate into an antagonist if it is not quickly recognized and addressed (Springer et al., 2011). Consistency in operations is therefore crucial, because customers become promoters when all key interactions leave them with a positive impression relative to what they have experienced elsewhere. Promoters are what every business should strive to achieve, because they contribute much more to the bottom line than simply the sum of their purchases (Bildfell, 2011).

The aim of CLF is to create more promoters and to decrease the amount of detractors. Therefore the first hypothesis as follows.

H1: CU S T O M E R S W H O U N D E R G O CL O S E D-L O O P FE E D B AC K H A V E A H I G H E R NPS A F T E R T H E CLF T H A N B E F O R E T H E CLF

2.3 W

HAT ARE INTERACTIONS AND

‘

MOMENTS OF TOUCH

’?

In general interaction is a mutual or reciprocal action where two or more parties have an effect upon the other (Grönroos, 2011). The main aspect of interaction is connectivity. In other words, the parties involved are in some contact with each other (Grönroos, 2011). When a customer has corresponded with a firm, either on a positive or a negative tone, it provides the firm with an unique chance to have a conversation with the customer on personal level. Regardless the nature of a customer’s communication, a reaction by the company is always expected (Shields, 2006).

2

- 16 -

Interactions can be associated with advertisements, promotional material, service provision and purchase decisions. In these interactions it is not necessary that customers are physically involved. For example, an individual can interact with a company by simply reading a brochure or by seeing an advertisement on television. Moreover, in the financial service sector, a bank can make numerous automated transactions without any ongoing physical interaction with its customers. These examples indicate that involvement can vary according to frequency and duration (Howcroft et al., 2007) and interaction with a brand without noticing it consciously (Springer et al., 2011).

Contacts are the way a customer interacts with the business (Spencer-Matthews and Lawley, 2005). In high and medium consumer service settings, regular customers often form a quasi-friendship with individual service personnel (Bove and Johnson, 2002). In terms of Springer et al. (2011) this represents an interaction that included the offer and its main touch points.

Interaction management influence performance according to research of Ramani and Kumar (2008). The authors define interaction orientation as the firm’s ability to interact with its individual customers and to take advantage of information obtained from them through successive interactions to achieve profitable customer relationships. Furthermore, in this study it is examined that this strategy leads to superior performance outcomes. They identified customer concept, interaction response capacity, customer empowerment, and customer value management as the four components of interaction orientation.

In research of Kellog (2000) it is shown that information exchange contacts differ from each other on several aspects, namely in (1) time, (2) information richness, (3) direction of flow, and (4) value of the exchange. Venkatesan and Kumar (2004) add another three contact modes and its antecedents, namely rich (e.g. face-to-face, trading event meetings), standardized (e.g. direct mail, telephone) and Web-based contacts.

2.4 W

HAT ARE CONSEQUENCES OF INTERACTIONS AND WHY IS IT IMPORTANT

?

- 17 -

customers during ‘moments of truth’. Such moments give an opportunity to strengthen the relationship with the customer because at these moments customers are highly (emotional) involved. Examples of ’moments of truth’ are, a customer who has lost his or her credit card or a customer who is willing to gain advice about an investment opportunity. Companies can earn trust and loyalty during these moments if customer service is able to handle it exceptionally. Consumers who use online communication want efficiencies are high relational and transact frequently (Venkatesan and Kumar, 2004). However, contacting consumers too much can actually harm your business results. As a consequence, customers might put off from purchasing products (Kumar et al., 2006) and this communication can even be detrimental to the relationship (Fournier et al., 1997). Consumer do care about which contact method is used in a certain context (low/high value; low/high time sensitivity). Which contact method suits the situation best, is tested on four aspects, namely satisfaction, socialization, empathy and privacy sensitivity (Smith and Eroglu, 2009).

Interactions with customers help firms with expanding their knowledge about customer preferences and needs (Srinivasan et al., 2002). As customers take the initiative to interact the information is still rich with customers’ concerns about product and firm, but it is less expensive than when the firm takes initiative (Bowman and Narayandras, 2001). Grönroos (2000) claims that interaction with customers can change customers’ perceptions of quality. During the interaction both firm and customer influence the creation of value. In line with Grönroos (2000) research of Ojasalo (2001) also makes clear that during interactions with customers firms can adjust customers’ expectations. More specific, if customers share their complaints during an interaction, a firm can change the perception of the service experience (Tronvoll, 2012)

- 18 -

to move the relationship to a deeper level of commitment for both parties. The quicker complaints are resolved, the more likely customers are to be satisfied (Bitner, 2006). All interactions play a role in creating promoters, however, it depends on industry which interactions are the best. For example, in the automotive industry product interactions are important, while in the insurance industry interactions with sales and service points matter most. Firms should learn which interactions are most important to each segment of their target customers, in order to create promoters.

In line with the theory the following hypothesis is formulated:

H2: TH E I N C R E A S E I N NPS D U E T O CLF I S H I G H E R F O R C U S T O M E R S W H O H A V E C O M P L A I N T S A B O U T T H E P R O D U C T/S E R V I C E T H A N F O R C U S T O M E R S W H O H A V E P R O D U C T F AI L U R E S

2.4 O

PTIMIZE CONTACT MOMENT AND MAXIMIZE CUSTOMER EXPERIENCE

Throughout the years, marketing practice transformed into creating compelling relationship experiences. As stated in research of Verhoef et al. (2009) customer experience is nowadays a key strategic objective for firms.

In research of Meyer & Schwager (2007) is customer experience defined as the internal and subjective response customers have to any direct or indirect contact with a company. For example, the product or service itself as well as interactions with company’s representatives or some third party are included in these touch point. Frow & Payne (2007) describe customer experience as the user’s interpretation of his or her total interaction with the brand. The study of Lemke et al. (2010) shows that some touch points may not be relevant, and equally, other categories of customer experience do not involve touch points at all. They state that the customer experience quality is perceptual and intimately related to the customer’s goals. Verhoef et al. (2009) add that customer experience is holistic in nature and that it involves the customer’s cognitive, affective, emotional, social and physical responses to the retailer. In addition, they argue that the retailer can control some of the elements but not all of them.

- 19 -

- 20 -

In several relationship marketing studies relationship age or relationship length has been linked to the several stages in the relationship lifecycle. According to Eggert et al. (2006) if relationship age is used as an indicator of relationship development it is implied that the customers with the same relationship age are in the same relationship stage, despite the differences in growth rates.

In a study of Grayson & Ambler (1999) it is argued that longer relationships are quantitavely different from shorter ones. According to Verhoef et al. (2002) studies in social psychology show that the length of the relationship is positively related to confidence in one’s evaluations. Furthermore, satisfied customers act as referrals who recommend the business and long-term customers are less dependent to companies’ employees for advice and help since they are more familiar with the company (Berger & Nasr, 1998). In contrast with novices, long-term customers base their opinions upon their own experiences and knowledge about the product. Berger & Nasr (1998) found a difference on which service perception novice and long-term customers judge service quality. The effect of expertise and outcome on service quality is stronger for long-term customers, and tangibles have more effect on novices. Because CLF is more focused on expertise and outcome, this study assumes that long-term customers are more prone for CLF. Therefore, we state the following hypothesis:

H3: LO N G-T E R M R E L A T I O N S H I P S A R E M O R E S U S C E P T I B L E FO R CLF C O M P AR E D T O S H O R T-T E R M R E L A T I O N S H I P S

2.5 C

USTOMER VALUE

- 21 -

detractors (Reichheld, 2006). In general, the customer value of promoters is higher compared to detractors:, they stay longer with the company, and their purchases increase more rapidly. Nevertheless, a CLF interference is designed for passives and detractors and does not further interfere with promoters. However, in general business it does not have to mean that the high-valued customers are the promoters. Among passives and detractors there are high-valued customers as well. These customers are important for the firm because of their value, and therefore from a firm perspective it is preferred that these customers recommend the company. Research of Verhoef (working paper) shows that low-value customers might experience more reactance towards the firm than high-value customers, because high-valued customers are more forgiving than low-valued customers. Consequently, the CLF intervention can be especially useful to interact with the high-valued customers in order to increase their willingness to recommend the company more. It is therefore expected that the effect of CLF on NPS is higher for high-valued customers than for low-valued customers.

- 22 -

C

HAPTER

3

M

ETHODOLOGY

This chapter discusses the research design. After that, the data collection will be described. The chapter ends with a description of several statistic tests which will be used to analyze the data. Results of these tests will be presented in Chapter 4.

3.1 R

ESEARCH DESIGN

The nature of this research is to test specific hypotheses and examine the relationship between CLF and NPS. According to Malhotra (2007, p.79-86) this is descriptive research, which is based on a single cross-sectional design, because only one sample of respondents is obtained. However, to evaluate whether there is an improvement in NPS due to CLF, NPS needs to be measured by longitudinal design, thus two points in time.

3.2 D

ATA COLLECTION

To investigate the influence of CLF on NPS it is necessary to collect market data. This data has been collected by a marketing research company, who is continuously measuring customer satisfaction for a large Dutch energy company. The data is a collection of different surveys together, which consists of surveys in regard to complaints about energy services, including billing as well as digital services

The data of this regular research, has been obtained from January 28th, 2011 until October 31th, 2012. In the first round of the questionnaire the total sample consists of 6903 respondents who have answered the Net Promoter Score question based upon research of Reichheld (2003). In the follow up questionnaire only respondents have been targeted who scored a 0-8 on the NPS in the first round. Therefore, in the second round data of 595 respondents has been collected. These respondents did undergo the Closed-loop feedback procedure and have filled in the second survey.

3.3 D

ATA EXCLUSION

- 23 -

3.4 D

EPENDENT AND INDEPENDENT VARIABLES

This study is based upon one dependent variable and multiple independent variables. Next, both this dependent and these independent variables are presented.

3.4.1 DEPENDENT VARI ABLE

The dependent variable in this study is the change of the Net Promoter Score over time measured in this study. This variable is the total points shifted by subtracting the first NPS from the second NPS . The variable is calculated as follows:

Where:

3.4.2 INDEPENDENT VARI ABLES

In this study several independent variables are used to determine the change in the Net Promoter Score. The independent variables are: Net Promoter Score t-1, gender, relationship length, customer value in euro’s, moments of truth and the type of product (Water, Boiler, Gas, Cable, Cable Internet, Electricity, Digital Television). An overview of the variables are in table 1 below.

Independent variable Based on: Scale Net Promoter Score t-1 How likely is that you would recommend

[company X] to a friend, relative or colleague? (Reichheld, 2003)

0 not at all likely, 10 extremely likely

Relationship length Start date of first product purchased (provided by company)

Ratio

Customer value in euro’s The customer value per individual on annual basis (provided by company)

Ratio

Moment of truth What is your reason to contact the company? Nominal

Type of product Which of the following products have you purchased?

Nominal

Gender What is your sex? Nominal

TABLE 1 INDEPENDENT VARIABLES BASED ON ONLINE SURVEY

- 24 -

decided to take the date of the first purchase as the starting date of the total relationship.

Customer value is the total monetary value in Euro’s of all products a customer has purchased on an annual basis. This means customer value is total sum in Euro’s of all product subscriptions together. The value per product differs between customers because customers can purchase different packages of the products (Woodruff, 1997).

3.5 A

NALYSIS

M

ETHOD

The analysis will start with general descriptive statistics of the data and then the distributions of the variables will be examined by testing the Skewness (Malhotra, 2007, p.462) and Kurtosis (Malhotra, 2007, p.462). To get some general insights into the data, scatter plots will be used to show patterns in the data, and to show some first relations between independent variables on the dependent variable. To check whether independent variables have relations among each other a correlation test will be conducted. In this study a threshold level of the p-value is equal to or below .05 is set to test the significant level of the variables and to test the hypotheses. According to Malhotra (2007 p. 466), one should control type I errors (rejecting the hypothesis when it is in fact true) by using a tolerable significance level (α). In this research, a significance level of 0.01 (99%) and 0.05 (95%) is used, with some exceptions where the significance level is 0.10 (90%), these levels are indicated in the coefficients tables.

The hypothesis with regard to the impact of the first NPS on the second NPS will be tested by a paired-samples T-test. Because both groups are related to each other a paired-samples T-test is an appropriate test to measure whether there is a difference in the mean (Malhotra, 2007, p.487). By comparing both groups before and after CLF the effect of CLF can be analyzed. A higher mean indicates the group scores better. Furthermore, the probability associated with the z statistic is less than .05, indicating that the difference is indeed significant. If the result of the test is significant, it means that both groups differ and therefore CLF influences the final NPS.

Before conducting a regression analysis first the correlation among the independent variables will be measured. A correlation matrix is helpful to summarize the strength of association between two variables (Malhotra, 2007, p.534).

- 25 -

- 26 -

4. R

ESULTS

In this chapter results of the analysis will be presented. First, a descriptive insight will be given of the respondent’s characteristics and the means of the Net Promoter Scores. Second, results will presented about which variables influence the change in the Net Promote Score when the Closed-loop Feedback process has been conducted.

4.1 D

ESCRIPTIVE STATISTICS

Table 2 presents the average NPS before and after CLF. The increase in the average indicates that respondents are willing to recommend the company more to others after they have been in contact with the company via this surveys then before. The question is why customers are willing to recommend the company more after CLF interference. These outcomes are presented later in this chapter. First, table 2 summarizes the outcomes of the descriptive analysis briefly. Remarkable is that most respondents experienced a product failure (82%) and that only a few had a complaint (6%) . For about 11 % of the respondents the contact reason is unknown.

Variable Outcomes

Gender Female 16.5%

Male 83.5%

Reason for contact To complain 6.4%

Product failure 82.4%

Number of product per customer Mean score 5.4

Customer value in Euro’s per year Mean score € 569.17

Length of relationship Mean score 5.17

Net Promoter Score Before CLF 5.32

After CLF 6.93

TABLE 2OVERVIEW DESCRIPTIVE STATISTICS

Despite the effort companies put into CLF not all customers are willing to recommend the company more to others after CLF. About 82% of the detractors (n=252) are willing to recommend the company more, 11% does not change their recommendation after CLF, and 8% are intended to recommend the company less.

- 27 -

passives the percentage of promoters is still higher compared to detractors. Results of shifts in NPS groups are presented in Table 3 below, which indicates two things. If companies want to increase the number of promoters, they should focus on passives. However, if they want to increase the willingness to recommend they should focus on detractors.

NPS after CLF

Detractors Passives Promoters

N PS Bef o re CLF t-1 Passives (n=203) 17 (8%) 133 (66%) 53 (26%) Detractors (n=252) 111 (42%) 115 (46%) 26 (10%)

- 28 -

4.1.1 AVERAGE NPS I NCREASE PER INDEPENDENT VARI ABLE

To get a deeper understanding about the relationship between the average increase in NPS and the individual independent variables another analysis is performed. In table 4 the results are presented. For all variables holds that after the CLF interference the final NPS has increased with at least 1.32 point. An one-way ANOVA has been conducted to check whether there is a relationship between the independent variable and the change in NPS. Only the product ‘Internetcable’ has a significant outcome.

Independent variable Average increase in NPS after

CLF

One-way ANOVA

Reason to call is a complaint (n=28) M = 1.46 (SD = .387) .719

Reason to call is a product failure (n=375) M = 1.58 (SD = .126) .692

Gender (n=453) – Female (n = 75) – Male (n=378) M = 1.57 (SD = .278) M = 1.63 (SD = .124) .348 Products – Electricity (n=411) – Gas (n=404) – Water (n=432) – Cable (n=432) – Boiler_CV (n=57) – Internetcable (n=406) – Digital television (n=297) M = 1.65 (SD = .119) M = 1.67 (SD = .120) M = 1.64 (SD = .116) M = 1.61 (SD = .117) M = 1.32 (SD = .331) M = 1.58 (SD = .118) M = 1.57 (SD = .141) .943 .899 .487 .644 .506 .005* .615 *Significant at 0.01 level

- 29 -

Next, the average increase in NPS per variable for ratio-sales variables has been calculated. With regard to the ratio-scales variables, customer value and relationship length, there is chosen to obtain for these variables three groups of customers. Based on their index scores, a group of the lower 25% quartile, a group of the middle 50%, and a group of the upper 25% quartile are formed.

Short / Low (lower 25% quartile) Average / Medium (middle 50%) Long / High (upper 25% quartile)

Relationship length (n=439) < 3.29 years (n=112) Average NPS: +1.84 Mean: 5,72 years (n=217) Average NPS: +1.53 > 6. 73 years (n=110) Average NPS: +1.74 Customer value (n=453) < €508,50 (n=113) Average NPS: +1.75 Mean: €603,00 (n=114) Average NPS: +1.33 > €684,50 (n=226) Average NPS: +1.71

TABLE 5AVERAGE NPS INCREASE FOR RELATIO NSHIP LENGTH AND CUS TOMER VALUE

As shown in table 5, the average increase of NPS is at least 1 point for both independent variables. The largest increases are found for the lower 25% quartiles. Especially the group of customers who stayed shorter than 3.29 years at the company are prone for CLF as their willingness to recommend the company increases the most. For both variables counts that the upper 25% quartiles are the second group of customers who seems to be most prone for CLF. The middle group is in both cases slightly behind. Although, still the increase of NPS is at least 1.33 for both variables in these groups.

4.1.2. NORMALITY

- 30 -

4.1.3 THE EFFECT O F CLF O N NPS

To examine whether there is an effect of CLF on NPS a paired-samples t-test is conducted, because both NPS groups are related to each other. As already shown in the descriptive analysis the average of the NPS without CLF is lower compared to the NPS after CLF. The t-test shows a significant (.000) relationship between both groups (see table 6). Shifting the focus to the proposed hypotheses, we first examine the influence of CLF on NPS. As the mean differs between the pairs by 1,611, therefore this study shows that CLF has a positive influence on the final NPS. Consequently, this supports hypothesis 1.

The next step is to get insights in how the independent variables effect the dependent variable. Therefore, first several scatter plots are made to gain knowledge about the influence of the independent variables on the change in NPS. The scatter plots show only a relationship of the first NPS towards the change in NPS. Obviously people who were most likely to recommend the company in the first measurement can score less higher than people who were not willing to recommend the company in the first measurement. Other variables seem to be not affected by CLF, however a better insight of this variables will be given when the regression analysis is conducted.

4.1.4. CO RRELATIONS

The correlation matrix gives some insights in how the independent variables relates among each other. The table presented in Appendix 1. A correlation (.656) is found between the first NPS and change in NPS. Furthermore, the length of relationship correlates with several other variables. There seems to be a connection between this variable with the first NPS, almost all products, customer value and with both moments of truth. In addition, independent variables such as; number of products, customer value

Paired Samples Test

Paired Differences t df Sig.

(2-tailed) Mean Std. Deviation Std. Error Mean 95% Confidence Interval of the Difference Lower Upper Pair 1 nps_opvolg - nps_clf 1,611 2,416 ,113 1,388 1,834 14,223 454 ,000

- 31 -

in euro’s, and customer value per product are significantly related and will probably cause some multicollinearity in the regression analysis. Also, a correlation among both moments of truth is detected. Collinearity is observed among several products. Since respondents in the dataset often purchased several products this is collinearity is expected. Due to the collinearity between customer value in euro’s and the number of products, in this study is chosen to include the type of products instead of the number of products. Therefore, in the regression analysis tests should be executed to check for multicollinearity.

4.2 R

EGRESSION ANALYSIS

The next step is to test the variables within a regression analysis. The independent variables will be tested on the dependent variable ( ). Therefore, the next regression analysis is tested:

Where is:

X1 = NPSt-1

X2 = Relationship length X3 = Customer value in euro’s X4= Value per product in euro’s X5= Contact reason is to complaint X6= Contact reason is a product failure X7= Gender

X8= Type of product

- 32 -

change in NPS. In this model not every independent variable is significant but only four variables. As the regression analysis shows NPSt-1 has a significant influence of (p-value <.000). In addition, the relationship length (.037), a complaint (.032), and the product Cable Internet (.012) significantly influence the CLF process.

A Variance Index Factor (VIF) of 10 or higher indicates multicollinearity. In this model, two variables show a number higher than 10. As already the correlation indicated an overlap between customer value and value per product could cause multicollinearity, therefore a decision should be made which variable should be included and which one excluded in the model. Since the different type of products are included in the model, the variable value per product explains less than the variable customer value. Therefore, customer value in euro’s will be included in the new model. Moreover, the normality plot with the standardized residuals shows some non-normality.

- 33 -

After excluding the value per product in euro’s a new regression analysis is executed. The regression analysis is formulated as follow:

Where: X1 = NPSt-1

X2 = Relationship length X3 = Customer value in euro’s X4= Contact reason is to complaint X5= Contact reason is a product failure X6= Gender

X7= Type of product

ANOVA Sum of Squares df Mean Square F Sig.

Regression 1196,532 13 92,041 28,421 ,000b

Residual 1421,671 439 3,238

Total 2618,203 452

Dependent Variable: NPS verschil

Predictors: (Constant), nps_clf, Gender, Water, boiler_CV, Reason is to complain, Gas, digitale televisie,

Relationlenght in Years, kabel, Reason is product failure, Customer Value in euros, internetkabel, Elektriciteit

- 34 -

Final model Unstandardized

Coefficients B Std. Error Standardized Coefficients Beta Sig. Collinearity Statistics VIF (Constant) 4,700 ,595 ,000 Relationlenght in Years ,079 ,039 ,087 ,041 1,438 Gender ,055 ,232 ,009 ,813 1,043 Electricity -,523 ,665 -,063 ,432 5,199 Gas ,674 ,620 ,087 ,277 5,183 Water ,264 ,428 ,023 ,537 1,134 Cable -,131 ,515 -,011 ,799 1,639 Boiler -,225 ,258 -,031 ,384 1,023 Cable Internet -1,088 ,452 -,138 ,016 2,653 Digital television -,124 ,189 -,024 ,512 1,123 Customer Value in euros ,001 ,001 ,109 ,051 2,514 Reason is to complain -,950 ,433 -,095 ,029 1,521 Reason is product failure -,191 ,293 -,030 ,514 1,709 nps_clf -,624 ,033 -,677 ,000 1,034 Model Summary R: .676 R Square: .457 Adjusted R Square: .441

TABLE 8REGRESSION MODEL (FINAL)

The ANOVA test (table 7) shows that the model is significant (.000). Therefore, the included variables predict the dependent variable. In addition, with an Adjusted R Square of .441 , about 44% of the dependent variable is explained by the independent variables (table 8). The VIF values do not exceed 10, which indicates that no multicollinearity is present in the model.

- 35 -

Compared to the previous model, this model predicts the change in NPS poor. This model is not significant and therefore we refuse to use this model to predict the change in NPS. Therefore, we continue with regression model 2 (table 8) and consider this one as the final one. Several independent variables are significant as the t-value are within the critical limits. First, the previous NPS has significant(.000) influence on the difference in NPS. With a standardized Beta of -.624, it shows that the biggest increases in NPS is measured for customers who were not likely to recommend the company at all. The chance of success for CLF is bigger for those customers because more improvement is possible compared to customers who were most likely to recommend the company.

4.2.1 HYPO THESES

The first hypothesis about the positive effect of CLF on customers’ willingness to recommend is accepted (see page 28). Changing the focus to the other proposed hypotheses, this study assumed a positive relation between moments of truth and chance to improve in NPS. In this study support is found for at least one moment of truth that contributes to the change in NPS. If customers contact the company because a complaint it negatively effects the change in NPS (sig .029), which means that if the complaint is not solved customers’ willingness to recommend the company decrease. No statistical conclusions can be made with regard to a product failure as contact reason since the t-value exceeded its critical limit. Therefore, H2 is supported. The effect of CLF for customers with complaints is indeed higher than for customers with product failures.Focussing onto the effect of CLF related to relationship length, it was assumed long-term customers would be more prone to CLF. As the coefficient for relationship length is positive (B .079) it implies that customers who stay longer at the company are more affected by CLF. Because the t-value (.041) is within the limit, support is found for H3.

- 36 -

To get a deeper understanding about which way the best customers – in terms of customer value- recommend the company this study shows that they recommend the company more often. Although, the t-value is not within the critical limit of 95%, it’s significant on a level of 90% and therefore it can be claimed that the best customers recommend more. Nevertheless, with a coefficient of .001 the impact of customer value on the shift in NPS is rather small. Table 9, see below presents an overview of the shift based on customer value in NPS after the CLF process. The table shows that being a high-valued customer not automatically means that this customer will be a promoter as well. All three NPS groups include high-, medium- and low-valued customers. As the outcome of the regression analysis already suggests the higher the value of the customer is the more prone the customer is for CLF. In table 9 shows among low-valued 24%, 25% of the medium-valued customers, and 30% of the high-valued customers shift to another NPS group. These numbers show that the difference between the groups is visible but the differences are small. Accordingly, weakly support is found for H4. With regard to the research question about the relationship between the customer value and the final NPS a correlation test is conducted (see table 10). There is evidence found that indeed high-valued customers have a higher willingness to recommend the company than others (sig .046).

Customer value Total

Low Medium High

NPSt-1 NPSt NPSt-1 NPSt NPSt-1 NPSt Promoters 14 21 44 79 Passives 45 58 55 63 102 126 247 (202) Detractors 68 41 59 30 124 56 127 (251) Shift in % 24% 25% 30% *Remark:

Read table as follow: Number of customers after CLF; (number of customers before CLF)

TABLE 9OVERVIEW OF NPS GROUPS BASED ON CUST OMER VALUE

Customer Value categorized

NPS_final Pearson correlation: .094

Sig. (2-tailed): .046

- 37 -

4.3 M

AIN

F

INDINGS

S

UMMARIZED

The study’s main findings are summarized in table 11. This study found support for several hypothesis and is able to answer the research question with regard to the link between the most valuable customers and the willingness of recommendation.

Hypothesis Result

H1: Customers who undergo Closed-loop Feedback have a higher NPS after the CLF than before the CLF

Supported

H2: The increase in NPS due to CLF is higher for customers who have complaints about the product/service than for customers who have product failures

Partially supported

H3: Long-term relationships are more susceptible for CLF compared to short-term relationships

Supported

H4: Customer with high-value for the company are more susceptible for CLF than customers with low-value

Weakly supported

- 38 -

C

HAPTER

5

D

ISCUSSION AND

C

ONCLUSION

Now the results have been presented this chapter continues with a discussion and the conclusions. This thesis was written in response to the need for more understanding of the effect of Closed-loop Feedback on the Net Promoter Score. More specific, this study intended to explore how CLF can increase customers’ willingness to recommend the company to other people.

The main finding is that the impact of NPS t-1 is significant for the process of CLF. This study has proven that CLF is able to influence the final NPS positively. In addition, NPS t-1 affects other independent variables as well. The model is not significant anymore when NPS t-1 is excluded. Therefore, making the choice to only include passives and detractors, and to take these groups as starting point for the further process has its effect. After CLF the willingness to recommend to product to others has increased for both passives and detractors. This finding is to be expected and is in line with Springer et al. (2011), who state that a series of positive interactions builds towards higher willingness to recommend, and Krishna et al. (2011) believe positive reactions is one of the most important ways to make quality relation Moreover, Zeithaml et al. (1996) also found a significant relationship between customers’ perceptions of service quality and their willingness to recommend the company.

- 39 -

customers with a high value are relatively more promoters, detractors are relatively more customers which have a low value.

Subsequently, in this thesis’ theoretical predictions it was hypothesized that the CLF interaction can stimulate customers’ willingness to recommend even after a complaint or product failure. This study shows that if the customers’ contact reason is a complaint, than indeed CLF can significantly influence the final NPS. In line with research of van Vaerenbergh et al. (2012) this study shows that companies certainly can impress customers with complaint handling. Also Tronvoll (2012) states that companies can change customers’ perception during an interaction when a complaint is handled. Markey et al. (2009) argue that moments of truth can eventually lead to more promoters. Results of this study does not directly show that customer perception is changed or that a moment of truth leads to more promoters, however it proved that effective complaint handling during the CLF process can at least lead to more willingness to recommend. No support is found with regard to the influence of CLF on NPS when customers experience a product failure. For product failures counts that customers expect the company to recover the service immediately after customers remark this failure. No service recovery damages the customer experience, while good service recovery does not harm further customer recommendations. Research of Du et al. (2010) state that service providers can regain customers’ positive evaluation through effective service recovery efforts. Other literature about this service recovery paradox (McCollough, 2009) shows that customer satisfaction increase after recovery, but the effect might be limited. Fair communication builds relationships (Krishna et al., 2011). Consumer tolerance increases for the failure which results in greater trust, loyalty, and positive word-of-mouth (Liao, 2007) The relation between product recovery and willingness to recommend is not found in this thesis .

5.1 MANAGERIAL I MPLI CATIO NS

- 40 -

with CLF, but it should be an organizational wide vision. Customers get impressions of the firm in several interactions, for example while using the product, visiting the store, calling the helpdesk, or watching a television commercial. Therefore, it is important that the company communicate consistently.

- 41 -

C

HAPTER

6

L

IMITATIONS AND

F

URTHER

R

ESEARCH

As in every research also this study has its limitations. In this chapter a few will be discussed. Furthermore, during this study interesting areas for further research appeared. Attention to these will be addressed in this chapter as well.

Main concern to this study is the lack of a control group of people who have been surveyed twice but not have been in the CLF process. In this study only respondents have been involved who interact with the company because of CLF, and therefore the results might misrepresent the effect of CLF. A control group of respondents who have been surveyed over time with regard to their willingness to recommend the company but not involved in CLF could give deeper insights about the true effect of CLF. In further research this control group is of paramount importance to measure the effect of CLF on NPS. Furthermore, this study focused only on data of one company. Other industries might be interesting as well and especially the focus on more high-involved products would be interesting. The emotional connection with high-involved products is rather different compared to low-involved products and CLF could be guiding in the interaction between company and customer in this area.

Another remark for this study is the style of questioning. The current style has focused on positive outcomes and therefore customer probably shifted, due to the leading questions, to more positive answers. Interesting will be to gather information with more neutral questions. This will result in more fairer outcomes with regard to the effect of CLF on NPS.

- 42 -

R

EFERENCES

Agustin, C., and Singh, J. (2005), “Curvilinear Effects of Consumer Loyalty Determinants in Relational Exchanges,” Journal of Marketing Research, 42 (1), 96–10

Ang, L. & Buttle, F. (2012), “Complaints-handling processes and organizational benefits: An ISO 10002-based investigation”, Journal of Marketing Management, Vol.28, No.9-10, August, p.1021-1042

Beaujean, M., Davidson, J. & Madge, S. (2006), “The ‘moment of truth’ in customer service”, McKinsey Quarterly, No. 1

Berger, P.D. & Nasr, N. (1998), “Customer Lifetime Value: Marketing Models and Applications”, Journal of Interactive Marketing, Vol.12, No.1, Winter, p.17-30

Berglund, K.M. & Ludwig, T.D. (2009), “Approaching Error-Free Customer Satisfaction Through Process Change and Feedback Systems”, Journal of Organizational Behavior Management, Vol. 29, p.19-46

Bildfell, D. (2011), “A Bird in the Hand: Keeping the Customer You Worked Hard to Acquire”, Canadian Institute for Management, Fall, p.8-9

Bitner, M.J., Booms, B.H. & Stanfield Tetreault, M. (1990), “The Service Encounter: Diagnosing Favorable and Unfavorable Incidents”, Journal of Marketing, Vol. 54, January, p.71-84

Bowman, D. & Narayandras, D. (2001), “Managing Customer-Initiated Contacts with Manufacturers: The Impact of Share of Category Requirements and Word-of-Mouth Behavior”, Journal of Marketing Research, Vol.38, August, p.281-297

Bradley, G. & Sparks, B. (2012), “Explanations: if, when, and how they aid service recovery”, Journal of Services Marketing, Vol.26, No.1, p.41-51

- 43 -

Du, J., Fan, X. & Feng, T. (2010), “An experimental investigation of the role of face in service failure and recovery encounters”, Journal of Consumer Marketing, Vol.27, No.7, p.584-593

Edvardsson, B., Johnson, M. D., Gustafsson, A. & Strandvik, T. (2000), “The effects of satisfaction and loyalty on profits and growth: products versus services”, Total Quality Management, Vol. 11, No. 7, p.917-927

Erickson, G.S. & Eckrich, D.W. (2001), “Consumer Affair Responses to Unsolicited Customer Compliments”, Journal of Marketing Management, , Vol.17, p.321-340

Estelami, H. (2000), “Competitive and Procedural Determinants of Delight and Disappointment in Consumer Complaints Outcomes”, Journal of Service Research, Vol. 2, No.3, February, p.285-300

Fournier, S., Dobscha, S. & Mick, D.G. (1997), “Preventing the Premature Death of Relationship Marketing”, Harvard Business Review, Vol. 76, January-February, p.42-44 Frow, P. & Payne, A. (2007), “Special Issue Papers Towards the ‘perfect’ customer experience”, Brand Management, Vol.15, No.2, p8

Gelbrich, K. & Roschk, H. (2011), “A Meta-Analysis of Organizational Complaint Handling and Customer Response”, Journal of Service Research, Vol.14, No.1, p.24-43

Grönroos, C. (2000), “Marketing as promise management: regaining customer management for marketing”, Journal of Business & Industrial Marketing, Vol.24, No.5/6, p.351-359

Grönroos, C. (2011), “A Service perceptive on business relationships: The value creation, interaction and marketing interface”, Industrial Marketing Management, Vol.40, p.240-247

Hair, J.F., Black, W.C., Babin, B.J., Anderson, R.E. & Tatham, R.L. (2006), “Multivariate Data Analysis”, Seventh Edition, New Jersey: Pearson Prentice Hall

Hanson, R. (2011), “Life after NPS; Controversy still surrounds use of Net Promoter Score”, Marketing Research, Summer, p.8-11

- 44 -

Hekman, D.R., Aquino, K., Owens, B.P., Mitchell, T.R., Schilpzand, P. & Leavitt, K. (2010), “An examination whether and how racial and gender biases influence customer satisfaction”, Academy of Mangement Journal, Vol. 53, No.2, p.238-264

Homburg, C. & Fürst, A. (2005), “How Organizational Complaint Handling Drives Customer Loyalty: An Analysis of the Mechanistic and the Organic Approach”, Journal of Marketing, Vol.69, July, p.95-114

Howcroft, B., Hamilton, R., & Hewer, P. (2007), “Customer involvement and interaction in retail banking: an examination of risk and confidence in the purchase of financial products”, Journal of Services Marketing, Vol.21, No. 7, p.481-491

Huang, J., Ho, S. & Lee, B.C.Y. (2003), “Who are the adorable customers – men or women? Serivice quality improvement via customers’ positive responses” , TQM & Business Excellence, Vol.14, No.7, September, p.789-800

Keiningham, T.L., Aksoy, L., Cooil, B., Wallin Andreassen, T. & Williams, L. (2008a), “A holistic examination of Net Promoter”, Database Marketing & Customer Strategy Management, Vol.15, No.2, p.79-90

Keiningham, T.L., Aksoy, L., Cooil, B. & Wallin Andreassen, T. (2008b), “Linking Customer Loyalty to Growth”, MIT Sloan Management Review, Vol.49, No.4, Summer, p.50-57 Kellog, D.L. (2000), “A Customer Contact Measurement Model: an extension”, International Journal of Service Industry Management, Vol. 11, No. 1, p.26-44

Kraft, F.B. & Martin, C.L. (2001), “Customer compliments as more than complementary feedback” , Journal of Customer Satisfaction, Dissatisfaction and Complaining Behavior, Vol. 14, p.1-13

Krishna, A., Dangayach, G.S. & Jain, R. (2011), “A conceptual framework for the Service Paradox Recovery”, The Marketing Review, Vol.11, No1, p.41-56

Kumar, V., Venkatesan, R. & Reinartz, W. (2006), “Knowing What to Sell, When, and to Whom”, Harvard Business Review, March, p.131-136

- 45 -

Malhotra, N.K. (2007), “Marketing Research, An Applied Orientation”, fifth edition, Prentice Hall

Markey, R., Reichheld, F.F. & Dullweber, A. (2009), “Closing the Customer Feedback Loop”, Harvard Business Review, December, p.43-47

McDollough, M.A. (2009), “The recovery paradox: the effect of recovery performance and service failure severity on post-recovery customer satisfaction”, Academy of Marketing Studies Journal, Vol. 13, No. 1, January, p.89-104

Meyer, C. & Schwager, A. (2007), “Understanding Customer Experience”, Harvard Business Review, February, p.116-126

Morgan, N.A. & Rego, L.L. (2006), “The Value of Different Customer Satisfaction and Loyalty Metrics in Predicting Business Performance”, Marketing Science, Vol. 25, No. 5, September-October, p. 426-439

Lemke, F., Clark, M. & Wilson, H. (2010), “Customer experience quality: an exploration in business and consumers contexts using repertory grid technique”, Journal of the Academic Marketing Science, Vol.39, p.846-869

Liao, H. (2007), “Do it right this time: the role of employee service recovery performance in customer-perceived justice and customer loyalty after service failures”, Journal of Applied Psychology, Vol. 92, No.2, p.475-489

Neckopulos, J. (2010), “Creating a Feedback Loop”, ABA Bank Marketing, December, p.26-30

Ojasalo, J. (2001), “Managing customer expectations in professional services”, Managing Service Quality, Vol. 11, No.3, p.200-212

Ramani, G. & Kumar, V. (2008), “Interaction Orientation and Firm Performance”, Journal of Marketing, Vol. 72, January, p.27-45

Reichheld, F.F. (2003), “The one number you need to grow”, Harvard Business Review, December, p.46-54

- 46 -

Reichheld, F.F. & Rogers, P. (2005), “Motivating through Metrics”, Harvard Business Review, May, p.20, 24

Sampson, S.E. (1998), “Gathering customer feedback via the Internet: instruments and prospects”, Industrial Management & Data Systems, Vol.98, No.2, p.71-82

Shields, P.G. (2006), “Customer Correspondence: Corporate Responses and Customer Reactions”, Marketing Management Journal, Vol.16, No.2, p.155-170

Smith, R.J. & Eroglu, C. (2009), “Assessing consumer attitudes towards off-site customer service contact methods”, The International Journal of Logistics Management, Vol. 22, No. 2, p.261-277

Spenner, P. & Freeman, K. (2012), “To Keep Your Customer, Keep It Simple”, Harvard Business Review, May, p.108-114

Springer, T., Azzarello, D., & Melton, J. (2011), “What it takes to win with customer experience”, Bain & Company publication

Srinivasan, S.S., Anderson, R. & Ponnavolu, K. (2002), “Customer loyalty in e-commerce: an exploration of its antecedents and consequences”, Journal of Retailing, Vol.78, No.1, p.41-50

Swanson, S.R. & Kelley, S.W. (2001), “Service recovery attributions and word-of-mouth intensions”, European Journal of Marketing, Vol.35, No.1, p.194-211

Tax, S.T., Brown, S.W. & Chandrashekaran, M. (1998), “Customer Evaluations of Service Complaint Experiences: Implications for Relationship Marketing”, Journal of Marketing, Vol. 62, April, p.60-76

Terblanche, N.S. (2009), “Customer experiences, interactions, relationships and corporate reputation: a conceptual approach”, Journal of General Management, Vol.35, No.1, Autumn, p.5-17

Tronvoll, B. (2012), “A Dynamic Model of Customer Complaining behaviour from the Perspective of Service-Dominant logic”, European Journal of Marketing, Vol.46, No.2, p.284-305