The Customer Journey Based Segment-Specific Effect of Customer Experience on Customer Loyalty

(1)

The Customer Journey Based Segment-Specific Effect of

Customer Experience on Customer Loyalty

Issuing Date: 14th February 2019

Due Date: 17th June 2019

(2)

University of Groningen & University of Muenster

Master Thesis

The Customer Journey Based Segment-Specific Effect of

Customer Experience on Customer Loyalty

Faculty: Faculty of Economics and Business

First Supervisor: Prof. Dr. Peter C. Verhoef

Second Supervisor: Nicole Moch, M.Sc. Dr. Arnd Vomberg

Issuing Date: 14th February 2019

Due Date: 17th June 2019

Submitted by: Rina Dombrink Bedumerstraat 118 9716BP Groningen +491621946068

(3)

I

Preface

Dear reader,

This master thesis constitutes the final project of my double degree. In the past two years I had the opportunity to deepen my knowledge in Marketing Management and Finance at the University of Muenster and learn about the field of Marketing Intelligence at the University of Groningen.

I would like to take this opportunity to show my appreciation for some people who helped me achieve my goals in academia. In the course of this master thesis, I would like to thank my supervisors, Prof. Dr. Peter C. Verhoef and Nicole Moch, for their continuous support and advice. Next, I would like to thank my dear accountability buddies Alexandra Jelich and Julia Sitz, whom I have to especially thank for their valuable feedback and continuous support. I further thank Sophie Ladwein for her advice.

I would like to express my warmest thanks to my parents, Silvia and Joe, and my brother Nik for constantly supporting me and always believing in me.

Finally, all I can say is: Enjoy reading my thesis! Kind regards,

(4)

II

Management Summary

Along with other changes in the marketplace, digitalisation increases the number of touchpoints a customer encounters on his or her way to purchase. To retain or build a competitive advantage and foster customer loyalty, creating a strong customer experience is an important factor. Due to the heterogeneity of customers’ needs and preferences, segmenting customers helps to specifically target them. Therefore, I cluster customers based on their customer journey in the first part of this study. Besides investigating the overall effect of the five different customer experience dimensions (cognitive, emotional, sensorial, social, and behavioural) on customer loyalty, I further examine these influence for each segment. For this purpose, I analysed the survey data of 202 respondents in the context of the German grocery retail market.

The analysis of the complete customer journey indicates three customer segments. Participants of two segments can be characterised as rather pragmatic with a short customer journey. One segment, namely the experienced pragmatic shopper segment, is focused on purchasing and consuming the product. The second segment, called the receptive pragmatic shopper segment, has the same characteristics as the first segment but also talks about the product. However, this communication is limited to offline communication. The last segment, the engaged multi-channel shopper segment, uses a lot more touchpoints, offline and online. Nevertheless, all segments predominantly purchase the products offline.

(5)

III Table of Contents Preface Management Summary Table of Contents List of Tables List of Graphs List of Abbreviations List of Appendices 1. Introduction

2. Conceptual and Theoretical Background 2.1. Retail and Digitalisation

2.2. Customer Experience 2.2.1. Customer Journey 2.2.2. Customer Segmentation 2.3. Customer Loyalty 3. Overview of Studies 3.1. Customer Segments

3.2. Customer Experience’s Influence on Customer Loyalty 4. Conceptual Framework

5. Research Methodology 5.1. Study Design

5.2. Discussion of the Qualitative Pre-Study about Touchpoint Usage 5.3. Sample Composition

5.4. Analyses Methodology

5.5. Quality Criteria of Quantitative Study 6. Analyses and Results

6.1. Data Assessment 6.2. Factor Analyses 6.3. Segmentation

6.4. Regression Analyses

6.4.1. Complete Sample Regression Analysis 6.4.2. Segment-Wise Regression Analysis 7. Conclusion

7.1. Summary and Discussion 7.2. Managerial Implications

(6)

IV

List of Tables

Table 1: Segmentation Studies.

Table 2: Touchpoints Derived from the Existing Literature. Table 3: Description of Sample.

Table 4: Descriptive Statistics.

Table 5: BIC Comparison for Cluster Solution 1. Table 6: BIC Comparison for Cluster Solution 2. Table 7: Cluster Validation.

Table 8: Segment Description. Table 9: Mean Values per Cluster. Table 10: Model Premise.

Table 11: Coefficient Effect Size Comparison.

(7)

V

List of Graphs

Figure 1: Conceptual Model. 17

Figure 2: Regression Model: Complete Sample. 38

(8)

VI List of Abbreviations Approx. BIC DV IV MCM WoM approximately

Bayesian Information Criterion dependent variable

(9)

VII

List of Appendices

Appendix A. Latent Variable Measurement of Dv and IVs. Appendix B. Further Variable Measurement.

Appendix C. Assumptions Testing Regression Analysis. Appendix D. Discussion Outline Pre-Study (German). Appendix E. Discussion Outline Pre-Study (English). Appendix F. Survey Quantitative Study (German). Appendix G. Survey Quantitative Study (English).

(10)

1. Introduction

Due to an ongoing development of digitalisation, companies are faced with a continuous change of their corporate environment. The usage of internet as a means to search for products is increasing. Herein, Marketing has an important function within the company. Firms can no longer solely rely on their product to generate a competitive advantage, but instead need to create a strong customer experience, which is necessary to generate customer loyalty (Verhoef 2018). The customer experience consists of five distinct dimensions, namely cognitive, emotional, sensorial, social, and behavioural (Lemon and Verhoef 2016).

Though digital is one of the key factors for companies to generate a competitive advantage, it also imposes several challenges on the companies (Leeflang et al. 2014). Herein, two of the main challenges are the efficient usage of the rapidly increasing available data, and social media (Leeflang et al. 2014). The shopping behaviour changes from a single channel to a multi-channel and further to an omni-multi-channel structure (Verhoef, Kannan, and Inman 2015). These challenges lead to an increasing number of touchpoints in the customer journey, which in turn makes the customer journey more complex. The marketing funnel developed in 2002 by the Marketing Centre Muenster (MCM) and McKinsey, consists of the following steps: awareness, consideration, preference, action, and loyalty. However, it has changed due to the increasing complexity and is no longer linear. Instead, customers can skip certain steps in the traditional journey, enter at any point in time, or stay at a certain stage (Boncheck and France 2014). Therefore, it is crucial for marketers to know and understand their current customers as well as their desired target group in order to be able to target them specifically at the particular touchpoints. Leeflang et al. (2014) also identified the potential of the firm to derive real customer insights as a potential challenge for marketers in the digital environment. These customer insights are needed to truly understand the customer. Nevertheless, the available data is increasing rapidly which firms can use for their segmentation.

(11)

should be internally homogeneous and externally heterogeneous. Thus, a segmentation can help companies to meet their customers’ individual needs.

The above mentioned customer journey can be divided into the previous experience, current customer experience, and future experience, whereby these experiences can be further subdivided into a pre-purchase stage, a purchase stage, and a post-purchase stage (Lemon and Verhoef 2016). Herhausen et al. (2018) provide a customer segmentation based on the touchpoints in the pre-purchase and purchase stage but recommend a further investigation of the post-purchase stage. Therefore, the aim of the first part of the study is to derive a customer segmentation based on the complete customer journey using a probabilistic cluster analysis. To be able to implement a marketing strategy that is specified for the different segments, it is crucial to know more about the segments. The creation of the customer experience may differ across segments and there is no sufficient holistic examination of it yet. Therefore, I aim to use the derived clusters and investigate the segment-specific influence of the five customer experience dimensions on customer loyalty. Thus, the main research questions are:

1. Which customer segments can be derived when considering the complete customer journey?

2. How does customer experience influence customer loyalty for these customer segments?

(12)

2. Conceptual and Theoretical Background 2.1 Retail and Digitalisation

The retail environment has changed over the past decades due to the ongoing digitalisation (Verhoef, Kannan, and Inman 2015) and is one of the most important changes in society nowadays affecting many elements of both the business environment and the everyday lives (Hagberg, Sundstrom, and Egels-Zanden 2015). In the past decade the variety, scope, and sophistication of retail channels have been growing (Dholakia et al. 2010). Multi-channel retailing has evolved to counter the effects of the development and the therewith related change in consumer behaviour (Verhoef, Kannan, and Inman 2015). While previously implementing a multi-channel strategy mainly focused on the question whether to add another channel (e.g., Geyskens, Gielens, and Dekimpe 2002), the ongoing developments concerning digitalisation in marketing and retailing lead to a new phase in multi-channel retailing (Leeflang et al. 2014). The challenges associated with this development include the growth in mobile device usage as well as social media usage and the integration of these new channels (Hagberg, Sundstrom, and Egels-Zanden 2015; Verhoef et al. 2015; Wang et al. 2015).

(13)

home delivery or click-and-collect (Colla and Lapoule 2012). The online shops also extend the offerings of the retailer as the retailer can now offer the long-tail online. Furthermore, new businesses evolve within the retail segment, such as Amazon expanding to Amazon Fresh or Facebook as a selling platform.

2.2. Customer Experience

In the increasingly competitive retail environment (see chapter 2.1 Retail and Digitalisation), creating a customer experience becomes more and more important (Pine and Gilmore 1999; Verhoef et al. 2009; Verhoef 2018). Customer experience is often defined as a subjective and internal response of the customer to an interaction with the company, brand, product, or service (Gentile, Spiller, and Noci 2007; Lemke, Clark, and Wilson 2011; Meyer and Schwager 2007; Pine and Gilmore 1999; Verhoef et al. 2009; Zomerdijk and Voss 2010). Pine and Gilmore (1999) stress the subjective nature of the experience construct since they describe experiences as inherently personal. Verhoef et al. (2009) further consider customer experience as a multi-dimensional construct and emphasise a holistic view on customer experience. Lemon and Verhoef (2016) incorporate the holistic view in their definition of the customer experience involving “the customer’s cognitive, emotional, behavioural, sensorial, and social responses to the firm’s offerings during the customer’s entire purchase journey” (p.71). In line with this definition, Homburg, Jozic, and Kuehnl (2017) define customer experience as “the evolvement of a person’s sensorial, affective, cognitive, relational, and behavioural responses to a firm or brand by living through a journey of touchpoints along prepurchase, purchase, and postpurchase situations and continually judging this journey against response thresholds of co-occurring experiences in a person’s related environment” (p.384).

The greater complexity of the customer journey due to an increasing number of touchpoints in the digital setting caused the customer experience management to gain more importance (Verhoef 2018). Digitalisation also enables new competitors to enter and capture the retail market such as Amazon, which underlines the importance for retailers to create a strong customer experience to differentiate themselves from the competition and eventually create customer loyalty (Verhoef 2018).

(14)

Thus, instead of being in full control over the customer experience, companies can design prerequisites and stimuli which may create the desired customer experience (Gupta and Vajic 2000). This underlines the importance to also include competitor-owned touchpoints as well as other influences that are outside of the retailer’s control (Verhoef et al. 2009). These external touchpoints include advertisement of competitors or customer Word-of-Mouth. I will further distinguish the different types of touchpoints in chapter 2.2.1 Customer Journey.

When considering the customer experience, it is important that the firm takes the customer’s point of view in the creation process. Therefore, mapping the customer journey can help to determine the crucial touchpoints in the customer-retailer interaction. The customer journey consists of three main stages, namely the pre-purchase stage, the purchase stage, and the post-purchase stage (Lemon and Verhoef 2016). These stages will be discussed in more detail in chapter 2.2.1 Customer Journey. However, to capture the holistic construct of the customer experience as proposed by Verhoef et al. (2009), it is necessary to incorporate all three stages. The importance of doing so is further supported by Lemon and Verhoef (2016).

2.2.1 Customer Journey

As identified in the previous chapter, understanding customers’ behaviour is a crucial determinant of success in today’s marketplace. One way to foster this understanding is mapping the customer journey using the touchpoints a customer faces. The touchpoint based customer journey helps to understand the process evolution over time from the customer’s point of view (Zomerdijk and Voss 2010). This helps to identify moments of truth. These moments of truth are the most important touchpoints for the purchase decision (Herhausen et al. 2018). Thus, companies should try to elevate at these touchpoints and ensure that the customer does not get dissatisfied.

This already emphasises the need to define the concept of a touchpoint. A touchpoint describes any direct or indirect contact a customer faces with the company or brand (Baxendale, Macdonald, and Wilson 2015). This definition broadens the definition by Neslin et al. (2006) by emphasising the possibility of a two-way communication and the existence of touchpoints which are out of the firm’s control. Duncan and Moriarty (2006) provide a similar definition stating that touchpoints are occasions – verbal and non-verbal – a customer encounters and knowingly links to the firm or brand.

(15)

one-way communication. While some studies determine the customer journey to be part of the search process, ending at the purchase stage (Anderl et al. 2016; Baxendale, Macdonald, and Wilson 2015), Lemon and Verhoef (2016) already broaden this view by including the complete process a customer faces from the pre-purchase stage through the purchase stage to the post-purchase stage. Analogously, according to Neslin et al. (2006), the multichannel buying process of customers comprises three stages, namely the information search, the purchase, and the after-sales service. The importance of the post-purchase stage depends also on the industry and thus, Herhausen et al. (2018) do not include it in their study of the retail industry. However, their definition provides a broader definition of touchpoints, considering not only retailer-owned, offline touchpoints. They define the customer journey as “customers’ search and purchase usage of all online and offline touchpoints from various sources, including retailer-owned, competitor-owned, and additional touchpoints” (Herhausen et al. 2018, p.7). Within the three stages, touchpoints can be brand-owned, partner-owned, customer-owned, or social/external/independent (Lemon and Verhoef 2016) as emphasised in the definition of Herhausen et al. (2018). The importance of the touchpoint categories varies across the stages (Lemon and Verhoef 2016). Brand-owned touchpoints are designed and managed by the firm meaning that the firm has control over these touchpoints, whereas partner-owned touchpoints are designed, managed, and controlled by the firm and one or more partners. Customer-owned touchpoints in contrast are under the control of the customer and the last category of social/external/independent touchpoints describes all external influences the customer encounters during the customer journey (Lemon and Verhoef 2016).

(16)

2.2.2 Customer Segmentation

Segmentation is a crucial part of the analysis of customer behaviour. This is also of practical relevance as a segment-based multichannel strategy requires targeting the customers at an individual segment level. This in turn stipulates the segmentation to include the relevant segmentation variables (Neslin et al. 2006). Further, the segments should be internally homogeneous and externally heterogeneous.

Customers differ in their behaviour and thus use different channels or touchpoints more heavily than others. This difference in attraction to the different touchpoints may be explained by differences in sociodemographic, psychographic, and other characteristics (Verhoef and Donkers 2005). Transferring Quiggin’s (1982) anticipated utility theory to customer journey based segmentation implies that the customer evaluates the costs and benefits at each touchpoint to maximise his or her utility (Herhausen et al. 2018; Konus, Verhoef, and Neslin 2008). The anticipated utility in turn is influenced by, for instance, the shopping goal of the customer (De Keyser, Schepers, and Konus 2015). The goal a customer pursues while shopping can be explained by sociodemographic, psychographic, and situational factors (Herhausen et al. 2018). Thus, including these factors as covariates in the customer segmentation provides valuable insights.

In practice, marketers use customer segments to develop personas which are stereotypes of segment customers which should ease the memory and visibility of the segments for marketing executives by representing a typical journey, motivation, attitude, and experience of a customer in the specific segment (Verhoef 2018). However, this simplification has the drawback that it assumes homogeneous segments and due to segments not being fully homogeneous a single persona does not represent the full segment (Verhoef 2018).

2.3 Customer Loyalty

(17)

(18)

3. Overview of Studies 3.1 Customer Segments

A frequently determined segment is the offline prone or store-focussed segment (Bhatnagar and Ghose 2004; De Keyser, Schepers, and Konus 2015; Herhausen et al. 2018; Keen et al. 2004; Konus, Verhoef, and Neslin 2008; Thomas and Sullivan 2005). The equivalent of the online prone or catalogue segment is also represented in a number of relevant segmentation studies (Bhatnagar and Ghose 2004; De Keyser, Schepers, and Konus 2015; Herhausen et al. 2018; Keen et al. 2004; Thomas and Sullivan 2005). While Konus, Verhoef, and Neslin (2008) determine a physical store segment, they do not find the online-prone segment, but rather a multichannel enthusiast and uninvolved shoppers. Herhausen et al. (2018) in contrast determine two different online segments, one being a pragmatic online segment and one being the extensive online segment. However, they include the mobile device usage in their segmentation study which is of increasing importance in the retail sector (Shankar et al. 2010). Mobile devices offer new possibilities like barcode scanning or location-based services (Ström, Vendel, and Bredican 2014), which can serve as new touchpoints in the customer journey. Further, considering the ongoing development of digitalisation, the study was published ten years later. Sands et al. (2016) also find two internet-focussed segments, where one segment is not prone to mobile device usage however, and one is a multichannel enthusiast. While De Keyser, Schepers, and Konus (2015) obtain one research shopper segment, Sands et al. (2016) differentiate between three research online purchase offline segments, one not being mobile and/or social media prone, one being the opposite – a social media enthusiast – and one again being a multichannel enthusiast. Herhausen et al. (2016) also find an online-to-offline segment. Besides the customer segments that show a clear preference for the channels they use some segmentation studies obtain segments with no general preferences (Bhatnagar and Ghose 2004; Keen et al. 2004; Konus, Verhoef, and Neslin 2008).

(19)

(20)

Table 1. Segmentation Studies.

Study Phase Segment

Herhausen et al. (2018)

Pre-purchase, purchase

Store-focussed segment, pragmatic online segment, extensive online segment, multiple touchpoint segment, online-to-offline segment De Keyser, Schepers, and Konus (2015) Pre-purchase, purchase, post-purchase

research shoppers, web-focused, store-focused, call center-prone Konus, Verhoef, and Neslin (2008) Pre-purchase, purchase

uninvolved shoppers, multichannel enthusiasts, store- focused consumers

Bhatnagar and Ghose (2004)

Purchase segment 1(no general preference), Segment 2 (physical store), Segment 3 (online)

Keen et al. (2004) Purchase Generalists (no general preference), Formatters (physical store), price sensitives (online or catalog), experiencers (no preference)

Thomas and Sullivan (2005)

Purchase Segment 1 (catalog), Segment 2 (physical store) Sands et al. (2016) Pre-purchase,

purchase, post-purchase

Segment 1 (ROPO – anti-mobile/social media), Segment 2 (ROPO – multichannel enthusiast), Segment 3 (ROPO – social media enthusiast), Segment 4 (Internet-focused – anti-mobile), Segment 5 (Internet-focused – multichannel enthusiast)

Source: Author’s own illustration.

(21)

Table 2. Touchpoints Derived from the Existing Literature.

Stage of customer journey Touchpoints Source

Pre-purchase Physical store

Online store (app) Catalogue

Competitor physical store Competitor online store (app) Competitor catalogue Brand website

Social media

News portal/newspaper Offline WoM

Direct type-in Branded search Generic search Display (banner) Retargeting Affiliate Email Referral

Anderl et al. (2016)

Events

PR

Batra and Keller (2016)

Peer observation Baxendale, Mcdonald and Wilson (2015)

Radio De Haan, Wiesel, and Pauwels (2016)

Product trial and display Foroudi et al. (2018)

TV and cinema ad Billboards

Blogs

Ieva and Ziliani (2017)

Forums Pauwels, Aksehirli and Lackman (2016)

Purchase Physical store

Online store /app

Post-purchase Loyalty program

Newsletters/personalised Emails

Davis and Dunn (2002) customer service De Keyser, Schepers

and Konus (2015) Social networks

Customer satisfaction surveys

Ieva and Ziliani (2017)

Consumption/usage Lemon and Verhoef (2016)

Offline WOM

Participation in brand communities Blogging

Voluntarily suggesting improvements

Van Doorn et al. (2010)

(22)

3.2 Customer Experience’s Influence on Customer Loyalty

As discussed in chapter 2.2 Customer Experience, the construct of customer experience comprises the five dimensions cognitive, emotional, behavioural, sensorial, and social (Lemon and Verhoef 2016). However, when measuring customer experience, the studies rarely measure all dimensions. Verhoef (2018) discusses the extent to which several studies incorporate the dimensions. Most studies include the cognitive dimensions but rather focus on the intellectual dimension without considering the mere cognitive evaluation of the customer-firm interaction (Verhoef 2018).

Brakus, Schmitt, and Zarantonello (2009) develop a brand experience scale involving the dimensions sensory, affective, intellectual and behavioural and later on measure the effect from the brand experience construct on loyalty while considering partial mediation effects from satisfaction and brand personality. However, I focus on the results from direct effect of the brand experience construct on loyalty, which Brakus, Schmitt, and Zarantonello (2009) find to be positive and significant. Klaus and Maklan (2013) use the customer experience quality scale, introduced in 2011, which encompasses product experience, outcome focus, moments-of-truth, and peace-of-mind. This measure focuses on the cognitive customer experience dimension and solely the peace-of-mind construct incorporates an emotional customer experience component (Verhoef 2018). Regardless of that, Klaus and Maklan (2013) find a significant, positive influence from customer experience on loyalty intentions. Bustamente and Rubio (2017) apply the theoretical approach of Verhoef et al. (2009) in the physical retail store setting. Instead of measuring the influence of the different customer experience dimensions on store loyalty, they use one in-store customer experience construct which can be explained by four components, namely cognitive, affective, social and physical experience. They find that the influence of customer experience on loyalty is strong, positive, and significant. Therefore, customer experience is regarded as an important driver for maintaining the relationship with a certain retailer.

(23)

loyalty in these industries, which might be due to the functional nature of these services. The behavioural customer experience dimension has a negative influence on customer loyalty in this study setting. They provide the possible explanation that in these sectors the actions of the customer neither satisfy nor engage him or her. Last, Brun et al. (2017) find a positive influence from the social customer experience dimension on customer loyalty.

Herhausen et al. (2018) investigate the segment-specific influence of customer experience on customer loyalty. In their study, customer experience is measured by the three constructs product satisfaction, journey satisfaction, as well as customer inspiration. An increasing satisfaction is found to be in line with an increase in loyalty in previous studies (Formell 1992; Oliver 1999). While many studies include the variable “customer satisfaction” to explain some part of customer loyalty (e.g., Yieh, Chiao, and Chiu 2007), Herhausen et al. (2018) differentiate between product and journey satisfaction as customers differ in their evaluation of the experience. While process-oriented customers emphasise the journey or process of the shopping experience, outcome-oriented customers focus more on the result (Thompson et al. 2009). Thus, the customer segments differ with regard to the importance of the variables product satisfaction and journey satisfaction depending on the weight the customers in the respective segment put on either the process or the outcome (Herhausen et al. 2018). The satisfaction of customers may be one of the cognitive components of customer experience (Lemon and Verhoef 2016). The last variable included by Herhausen et al. (2018) is the customer inspiration, which can be linked to the customer experience (Verhoef 2018) and constitutes a cognitive dimension of customer experience as well. Though these three customer experience constructs solely consider the cognitive dimension of customer experience, they do extend the way the dimension was measured previously by considering the cognitive evaluation of the interaction the customer has with the firm. Herhausen et al. (2018) discover that product and journey satisfaction, as well as customer inspiration positively affect customer loyalty. Next to this finding, the researchers further investigate the differences when considering the segments elaborated on in chapter 3.1 Customer Segments, and find differences between the segments.

(24)

(25)

4. Conceptual Framework

The conceptual framework has the purpose to ease the understanding of the quantitative study which is based on a model derived from the existing literature discussed in chapter 3 Overview of Studies. The model consists of latent constructs which are not directly observable. Thus, manifest variables are used for the measurement of the latent constructs.

The first part of the study consists of the customer segmentation based on journey characteristics. The journey characteristics comprise the touchpoints across the complete customer journey incorporating the pre-purchase, purchase, and post-purchase stage. Initially, I intended to also use active covariates in the cluster analysis. However, due to a better cluster solution, I solely used the touchpoints as this is also the main interest (see chapter 6.4 Segmentation). Then, the second part of the study focusses on the customer segment-specific influence of customer experience on customer loyalty.

To integrate the customer experience, five latent constructs are used based on the definition of Lemon and Verhoef (2016), namely cognitive customer experience, affective customer experience, sensory customer experience, behavioural customer experience, and social customer experience. These constructs will serve as the independent variables (IVs) influencing customer loyalty. The scales to measure cognitive and behavioural customer experience are adopted from Bustamente and Rubio (2017). Emotional and sensorial customer experience are measured using the same scales as Brakus, Schmitt, and Zarantonello (2009). Last, social customer experience is measured using the scale used in the study of Brun et al. (2017). For a detailed display of the scales used see appendix A and B.

(26)

Figure 1. Conceptual Model.

a_{Due to a better cluster solution, the active covariates are not included in the final clustering (see chapter 6.4}

Segmentation).

b _{Based on the exploratory factor analysis, the two loyalty constructs form one factor and are thus used as one DV.}

In the following I derive hypotheses for the conceptual model displayed above (see figure 1). Prior research finds that people seek pleasure (Freud 1920/50) as well sensory stimuli (McAllister and Pessemier 1982). This leads Brakus, Schmitt, and Zarantonello (2009) to point out that experience provides value and that an experience is the result of stimuli leading to a pleasurable outcome. This in turn causes the customer to develop the wish to repeat the experience. All of this indicates that customer experience generally has a positive influence on customer loyalty. However, it is of interest to investigate how the different dimensions influence customer loyalty. People get bored when their cognition is not stimulated, what people try to avoid (Cacioppo and Petty 1982). Cognitive customer experience should engage customers in thinking and trigger their creativity, which provides a means to avoid boredom even though people differ in their need for cognition (Cacioppo and Petty 1982). Herein, I expect a positive effect from cognitive customer experience on customer loyalty. Thus, hypothesis 1 states:

H1: Cognitive customer experience has a positive effect on customer loyalty.

(27)

(Thomson, MacInnis and Park 2005). Further, the emotional bond of a customer to the brand influences the relationship stability (Fournier 1998). Emotional customer experience can trigger the emotions of a customer keeping him or her connected to the brand. Thus, I derive hypothesis 2:

H2: Emotional customer experience has a positive effect on customer loyalty.

Since people seek sensorial stimulation (McAllister and Pessemier 1982), experiencing this sensorial stimulation while purchasing or consuming the product would result in satisfaction of this desire. The customer might recall this feeling of satisfaction next time when he seeks the stimulation and thus purchase the product again. Therefore, I deduce hypothesis 3:

H3: Sensorial customer experience has a positive effect on customer loyalty.

Behavioural customer experience involves the customer’s physical response which can be displayed in form of comfort of discomfort (De Looze et al. 2003; Kuijt-Evers et al. 2004). Comfort constitutes a desirable state which customers want to repeat. Hence, using the positive connotation being targeted on the comfort, I derive hypothesis 4:

H4: Behavioural customer experience has a positive effect on customer loyalty.

From the point of evolution, a human’s survival depends on social relationships (Baumeister 2005; Dunbar 1993, 1997). Thus, people have a sense of belonging and an urge to foster social relationship to avoid social exclusion. The social customer experience can serve as a means for these social interactions which would be a pleasurable outcome for the customer and thus he has an incentive to repeat the feeling and keep buying the product. This leads to hypothesis 5:

(28)

5. Research Methodology 5.1 Study Design

To derive meaningful insights and develop managerial implications, I use primary data generated by a quantitative survey. To acquire profound insights into the touchpoint usage by customers across all three stages, ten initial qualitative interviews were conducted prior to the main study1. The pre-study was conducted between 10th April, 2019 and 14th April, 2019. The sample of ten participants consists of four men and six women between the age of 21 and 61 years. Five participants are university students, the remaining five participants are working in the fields of business and medicine. The interviews are taped and then transcribed and content analysed (see appendix H). This should ensure that the touchpoints used in the survey are not only based on literature but actually meaningful to the customers.

The sample of the pre-test consists of 18 participants and resulted in no major changes in the questionnaire. However, combined with the results from the qualitative pre-study, a few touchpoints from the customer journey were removed. In the pre-purchase stage, the touchpoints product test, newspaper and news portal, as well as brand website are removed after not being selected by the participants of the pre-test and also not being important to the participants of the pre-study (chapter 5.2 Discussion of the Qualitative Pre-Study about Touchpoint Usage). In the post-purchase stage solely the touchpoint brand website is removed. Further, a short introduction was added to prevent any misunderstandings.

The participants of the main study are acquired using a so-called ad hoc mail survey. This type of survey is characterised by a single usage for one specific project and no prior contact to the participants (McDaniel and Gates 2012). Due to the non-feasibility of a complete sample survey, I use a partial sample survey. Therefore, a nonprobability sample, more precisely a convenience sample, is constructed (McDaniel and Gates 2012). Thus, the online survey is distributed via email, and social media, including Facebook, Instagram, and WhatsApp. The survey is in German to enable a larger sample size. The questionnaire is constructed using validated scales from prior studies. Some of these scales include reverse coded items to counter a potential response bias (see appendix F and G).

The study focusses on fast moving consumer goods. Fornell (1992) investigates customer satisfaction in Sweden including nondurable goods which are grouped into six subgroups, one

1_{The initial qualitative interviews were conducted in cooperation with the master thesis of Sophie Ladwein}

(29)

of them being candy and coffee. Tarkianen and Sundqvist (2009) find that coffee is generally a high-involvement product. Involvement is defined as “the general level of interest in the object or the centrality of the object to the person’s ego structure” (Day 1970, p.45). According to this, involvement possesses an individual component encompassing unique attitudes and values of a person (Bloch and Richins 1983). This implies that the involvement may not be high for every participant. Nevertheless, purchase decisions about these two product categories are probably made more conscious compared to basic foods or dairy products and are thus useful to foster the memory of the participants with regard to their latest purchase and customer journey. Besides these two categories, alcoholic beverages are included. Previous studies found that there are both, highly and lowly involved customers of red wine, as an example of this product category (Quester and Smart 1998). The participants are asked to recall a purchase of a product from one of these three product categories whereby the purchase should be within the past three months to avoid a potential recall bias (Herhausen et al. 2018). The data collection took place between the 5th May 2019 and the 17th May 2019.

5.2 Discussion of the Qualitative Pre-Study about Touchpoint Usage

Analogous to the quantitative study, the participants of the pre-study are asked to remember a purchase of a product from the categories coffee, sweets or snacks. The product category alcoholic beverages has only been included after the conduction of the pre-study and is thus not part of it. Five participants chose coffee, two sweets, and three other snacks. The discussion outline starts with open questions about the touchpoints in the pre-purchase, purchase, and post-purchase stage which the participants recalls him-/ or herself. Thereafter, the participant is shown a list of identified touchpoints based on the existing literature (chapter 3.2 Customer Experience’s Influence on Customer Loyalty) to investigate which touchpoints the participants recognise. The last question is again an open question about touchpoints the participant might recall after being stimulated by the list of touchpoints. The discussion outline is structured in this way to avoid influencing the participant in the first questions (see appendix D and E).

Pre-Purchase. Table 2 displays the touchpoints included in several studies. Out of the ten

touchpoints included in the study by Herhausen et al. (2018), the participants actually reported using eight of these touchpoints. Solely the online store of a competitor and the brand website are not used.

(30)

branded search and display, which in this case refers to a banner on a website, are used to gather information about the product prior to the purchase. One participant reports that she has

“directly searched for the product, I think I typed in “Markus Kaffee Aldi Nord or Aldi …” and looked for the pictures […]”2. Instead of using a price comparison website, one participant reports comparing prices offline in the physical store. As in the study of Batra and Keller (2016), one participant remembers having had contact with the product on an event and in the context of PR, which has already been some time ago though. Peer observation, a pre-purchase touchpoint used by Baxendale, Mcdonald, and Wilson (2015), is confirmed by most participants of the pre-study. While investigating the products in the physical store, they report consciously realising that other customers show interest in the same product as one customer states “I looked at the product and shortly after that I consciously noticed that two other

customers were looking at the packaging of this Ehrmann-Pudding as well […]”3. Further, the participants recognised advertisements on the radio (e.g., DeHaan, Wiesel, and Pauwels 2016), on the TV and in the cinema (e.g., Ieva and Ziliani 2017), as well as on billboards (e.g., Ieva and Ziliani 2017). Foroudi et al. (2018) include the touchpoint of product trial and displays. Even though the participants could not recall any product trials, they could confirm that they have seen displays in store. “In store, there were two displays, one right next to the entrance

and one near the coffee section.”4 While, the touchpoints blogs (Ieva and Ziliani 2017) and forums (Pauwels, Aksehirli, and Lackman 2016) are not used by the participants, one mentions being in contact with other products of the same company, either through the consumption of the product or advertisement of this product, as another touchpoint. She states that she thinks that she“[…] consume(s) several products of Ferrero in general […]”5 and continues “[…] I think Kinder also belongs to Ferrero, the Kinderschokolade, and since Easter is around the corner, one can see advertisements for the special Easter packaging frequently […]”6.

2_{“Ich habe schon direkt nach dem Produkt, ich glaube ich habe „Markus Kaffee Aldi Nord oder Aldi irgendwas“}

eingegeben und dann bei den Bildern geguckt [...].“ (Pre-Study: Interview 4)

3_{“Also ich habe mir das Produkt ja angeguckt und kurz darauf habe ich das auch bewusst so wahrgenommen,}

dass auch zwei andere Leute halt sich die Verpackung angeguckt haben von diesem Ehrmann-Pudding im Prinzip.“ (Pre-Study; Interview 5)

4_{“Im Laden waren zwei Aufsteller, gleich am Eingang und in der Nähe der Abteilung, der Kaffeeabteilung.“}

(Pre-Study: Interview 1)

5_{“[...] glaube aber dass im Alltag generell viele Fererroprodukte, also dass ich die generell konsumiere.“}

(Pre-Study: Interview 2)

6_{“[...] ich meine Kinder gehört ja auch zu Ferrero, die Kinderschokolade eben und da ja jetzt Ostern vor der Tür}

(31)

Purchase. Both touchpoints in the purchase stage used in the study by Herhausen et al. (2018)

– the physical store and the online store/ app – are used by at least one participant. Further, one participant reports talking about the product at the cashier. “I mentioned at the cashier at Edeka

that the product has been on promotion for the third time in a row and that I was happy about it.”7

Post-Purchase. Davis and Dunn (2002) investigate the post-purchase touchpoints by

integrating loyalty programs and newsletters/ personalised emails. While the latter are not perceived by the participants, loyalty programs such as Payback, which is a multi-vendor loyalty program in Germany, are mentioned. Neither did the participants report contacting the customer service (e.g., De Keyser, Schepers, and Konus 2015) nor participating in customer satisfaction surveys (e.g., Ieva and Ziliani 2016). However, the participants communicate with others about the product after the purchase. In this context they encounter two different touchpoints already used in prior studies – social networks (e.g., Ieva and Ziliani 2016) and Offline Word-Of-Mouth (e.g., Van Doorn et al. 2010). Besides, also online Word-of-Mouth is used, again by communicating via messengers such as WhatApp. Thus, one participant narrates that he “[…] texted via WhatsApp that I liked the product […]”8_{. This also led to another}

touchpoint he mentions, which are sponsored advertisements after the online Word-of-Mouth. Hence, he reports that “I got a recommendation for the product after the product purchase on

a website. […] as I took a photo and send that to a friend via WhatsApp, […] but I do not know if that is related […]”9. Lemon and Vehoef (2016) further include the consumption/ usage as a post-purchase touchpoint, which holds true for most participants if the purpose of the purchase was not to stock up. Besides the already discussed offline Word-of-Mouth, the participants did not report being in contact with any of the other touchpoints used in the study by Van Doorn et al. (2010).

Conclusion. Based on the results from the pre-study, the touchpoints in the pre-purchase and

post-purchase stage are adopted for the quantitative main study in the following way. While Anderl et al. (2016) differentiate between direct type-in, branded search, and generic search, I combine these touchpoints to one, namely web-search. From the remaining proposed

7_{“Ich habe bei Edeka an der Kasse erwähnt, dass es jetzt zum dritten Mal hintereinander im Angebot war, dass}

mich das gefreut hat.“ (Pre-Study: Interview 6)

8_{“[...] habe über WhatsApp dann halt geschrieben, dass ich das Produkt gut finde [...]“ (Pre-Study: Interview 5)} 9_{“Nach dem Produktkauf wurde mir das mal vorgeschlagen auf einer Webseite. [...] weil ich habe davon ein Foto}

(32)

touchpoints by Anderl et al. (2016), solely display is included in the main study. The two touchpoints blogs and forums (Ieva and Ziliani 2017) are not included separately but are rather summarised under the touchpoint social media. The touchpoint TV and cinema ad (Ieva and Ziliani 2017) is extended by further video advertisings, which can be displayed in various locations. Further, the touchpoint product trial and display (Foroudi et al. 2018) is separated into two touchpoints in the main study (product trial and product display) since the participants narrated seeing product displays but no product trials. Besides offline Word-of-Mouth (Herhausen et al. 2018), online Word-of-Mouth is added because of the communication via messengers such as WhatsApp.

In the purchase stage, the two touchpoints physical store and online store (Herhausen et al. 2018) are both included in the main study.

Regarding the post-purchase stage, the touchpoint customer service (DeKeyser, Schepers and Konus 2015) will be changed to return. The participants did not report using the touchpoint customer service. However, in case of a defective product, a return is a possible touchpoint. While social networks are included, customer satisfaction surveys (Ieva and Ziliani 2017) are not used by the participants and are not included. Out of the touchpoints in the Van Doorn et al. (2010) study, solely offline Word-of-Mouth is included as participation in brand communities, blogging, and voluntarily suggesting improvements are not used as touchpoints by the participants. Besides these touchpoints, based on the answers of the participants, online Word-of-Mouth, peer observation, and display banner on a website are added as touchpoints in the post-purchase stage.

5.3 Sample Composition

(33)

Table 3: Description of Sample. Sample Frequency (in %/ in total) Sample Frequency (in %/ in total) Gender Incomea - female 66.3 - below 1000€ 24.3 - male 33.7 - 1000€ to 2000€ 22.3 - 2001€ to 3000€ 18.8 Age groups - 3001€ to 4000€ 12.9 - 15 to 18 years 1.0 - more than 4000€ 10.4 - 19 to 24 years 23.3 - 25 to 34 years 40.6 Educationb;;c - 35 to 44 years 7.9 - “Haupt-/ Volksschule” 1.5 - 45 to 54 years 7.4 - “Realschule/ Lehrabschluss” 11.4 - 55 to 64 years 14.4 - “Abitur/ Fachabitur” 29.7 - 65 years and older 5.4 - “Hoch-/ Fachhochschulabschluss” 56.4

a _{Cumulative percent is 88.6%, because 11.4% of the participants choose to not state their income.} b_{Cumulative percent is 99.0%, because 1.0% of the participants stated to have a different education.}

c_{Please note that the survey was distributed among German participants. Thus, the categories display the different}

graduation levels in ascending order which are available in the German education system. Source: Author’s own illustration.

Due to the study investigating the German market, the sociodemographic characteristics have to be compared to the German population to assess the representativeness of the sample. First, the proportion of women is conspicuously higher in our sample compared to the actual German population, where men and women are almost equally represented (female: 50.7%, male: 49.4%) (Statistisches Bundesamt 2018). Further, the age distribution from the sample differs from the German population. According to the Statistische Bundesamt (2018) the majority of people is aged between 20 and 64 years (60.3%), which is still less than in this sample. This illustrates that the sample does not match the German population with regard to the age distribution. Hence, the representativeness of the sample is limited, which must be taken into account when investigating the results and developing implications.

(34)

Table 4. Descriptive Statistics.

Minimum Maximum Mean Median Standard Deviation Customer Loyalty 1.50 7.00 4.9546 5.0000 1.2350 Cognitive CX 1.00 7.00 3.1570 3.1429 1.2784 Emotional CX 1.00 7.00 4.0033 4.0000 1.1723 Sensorial CX 1.33 7.00 4.4422 4.6667 1.1610 Social CX 1.00 7.00 3.7203 4.0000 1.4638 Behavioural CX 1.00 7.00 4.7327 5.0000 1.2375 Source: Author’s own illustration.

To better understand the purchasing behaviour of the participants I shortly elaborate on the point of purchase. The vast majority of participants purchased their product in a supermarket (85.6%), followed by the specialty store (7.4%) and the online shop (5.0%). Two participants purchased their product in a canteen (1.0%) and one stated to have purchased at a kiosk (0.5%). Out of those participants having shopped online, four used their computer or laptop, two used a tablet, and one a smartphone. Thus, mobile device usage for purchasing the product is not pronounced.

5.4 Analyses Methodology

(35)

Information Criterion (BIC) or the Akaike Information Criterion (AIC) (IBM 2019). The BIC uses a larger parameter penalty. Rust et al. (1955) found that the BIC is the single best selection criterion. Thus, I use this information criterion.

The second part of the study consists of a regression analysis to investigate the influence of the customer experience dimensions on customer loyalty. To test these effects also for each segment, I separate the data based on the cluster membership variable obtained from the cluster analysis and conduct the regression analysis for the complete data set and then unit-by-unit for the segments. In the course of the regression analysis I examine the seven assumptions for ordinary least square regression analysis. Prior to conducting the regression analysis, I conduct an exploratory factor analysis and reliability test using Chronbach’s Alpha.

5.5 Quality Criteria of Quantitative Study

The quality criteria of quantitative research involve objectivity, reliability, and validity. Objectivity means that the results are independent of the researcher. There are three different measurement objectivities: objectivity of application, objectivity of analysis and objectivity of interpretation.

First, objectivity of application requires that the participants do not get influenced by the researcher. This is given due to the minimal social interaction as participants are solely asked to participate via social media. Next, objectivity of analysis can be achieved using standardised items and scales. I use scales from prior studies and thus this request is fulfilled. Objectivity of interpretation is given when the researcher has a minimal wiggle room for interpretation. Due to the statistical analyses of the data, an unambiguous interpretation of the results is given. Reliability describes the degree to which the study is not influenced by random errors. The lower the degree of random errors the more reliable is the study (McDaniel and Gates 2011). To ensure the internal consistency reliability, multiple items are used to measure one construct and the Cronbach’s Alpha is used as a measure of internal consistency (Mooi, Sarstedt, and Mooi-Reci 2018).

(36)

(37)

6. Analyses and Results 6.1 Data Assessment

In a first step, some incorrect answers were corrected which affected the variable education where participants chose the category other and named a university degree or apprenticeship. Thus, these observations were changed into the correct category. Next, there are no missing values in the data set.

The rule-of-thumb for outlier states that for a sample size of 80 or more observations, the threshold value for the standard deviation is +/- 4.0. The standard deviations in this data set range from 0.894 to 1.855. Therefore, the data set does not contain outliers.

Next to this, I elaborate on the correlations in the following. Some of the touchpoints correlate. The pre-purchase touchpoint advertisement video correlates with the pre-purchase touchpoint online display, which might be due to the algorithm analysing the data and displaying the advertisement adjusted to the individual. Further, pre-purchase online Word-of-Mouth correlates with the online display in the post-purchase stage. Again, this might be due to the algorithm analysing the texts. Next, the pre-purchase online Word-of-Mouth correlates with the post-purchase online Word-of-Mouth, which might indicate that the customers who use online communication tools to talk about the product have a general tendency to do so. Last, the pre-purchase touchpoint online shop correlates with the post-pre-purchase touchpoint newsletter/ personalised email. This indicates that participants either subscribed for the newsletter when visiting the website or the website visit was tracked. Further correlations do not seem to indicate a causality.

6.2 Factor Analyses

As described in chapter 4 Conceptual Framework, the latent constructs for the DVs and IVs are measured using manifest variables. Thus, I test whether the variables for each construct form one factor, which reliably reflects the construct. Therefore, I first investigate the Cronbach’s alpha. However, even if this value is sufficiently high, I cannot assume the unidimensionality of the measure which is why I perform exploratory factor analyses (hereafter called factor analysis) as well.

(38)

In this case there are solely high correlations though. To assess if the correlation matrix is an identity matrix, I use the Kaiser-Meyer-Olkin (KMO) measure of sample adequacy and the Bartlett’s test of sphericity. The KMO of 0.889 is meritorious, which indicates that the partial correlation is small and I can assume that the variables share common factors. Further, Bartlett’s test is highly significant (p = 0.000) and hence the variables in the population are correlated. This shows that the correlation matrix is not an identity matrix and the data is suitable for the factor analysis. Next, the investigation of the anti-image covariance matrix indicates that the data is suitable for the factor analysis as less than 25% of the off-diagonal values are higher than 0.09 and thus different from zero. Based on the anti-image correlation matrix, the measure of sampling adequacy (MSA) is assessed. The MSA values range from 0.834 to 0.949 and are therefore high. To recapitulate briefly, the data is suitable for the factor analysis.

Aiming to minimise the number of factors which are able to explain most of the variance leads to the principal component analysis being the appropriate method. Due to high communalities (>0.3) all variables seem to share a lot of variance (communalities: 0.485 – 0.825). To derive the number of factors, I use three criteria. First, the Kaiser criterion stating that the number of factors is equal to the number of Eigenvalues exceeding one. This criterion indicates a one factor solution with one Eigenvalue of 4.039. Next, the percentage of variance criterion derives the number of factors based on a cumulative percentage of variance which should be at least 60%. Again, this leads to a one factor solution (67.31%). Last, I investigate the scree test, where the elbow is at the second component leading to a one factor solution as well. All variables load high on this one factor (factor loadings: 0.697 – 0.908). Thus, instead of investigating the two customer loyalty dimensions attitudinal und behavioural customer loyalty separately, I consider them as one construct in the following analyses. This finding is not astonishing when measuring attitudinal and behavioural customer loyalty by questionnaires.

(39)

(40)

intended customer experience factors. To use the factors in the following regression analyses, I calculate new factor variables using summated scales.

6.3 Segmentation

I intend to cluster the participants based on their touchpoint usage and active covariates as described in chapter 4 Conceptual Framework. Prior to conducting the cluster analysis, I investigate whether the assumptions of independent cluster variables as well as a multinomial distribution for categorical variables variables hold. Based on cross-tabulations, not all touchpoints variables are independent. However, the procedure is still quite robust even though there is a slight violation of the assumption (IBM 2019). The touchpoint variables are categorical and follow a multinomial distribution based on a significant Chi2-test (p = 0.000 for all touchpoint variables). Including all touchpoints from the pre-purchase stage, purchase stage, and post-purchase stage as dummy variables as well as the set of covariates results in a total of 40 clustering variables. Two clusters can be derived, where the most important clustering variables are Pre-purchase social media, Post-purchase Offline Word-of-Mouth, Pre-purchase

physical store of another retailer, Post-purchase peer observation, and Age. The clusters are

not equal in size (62.0% and 38.0%). However, the two cluster solution is the best solution based on the BIC (see table 5). This results in a poor cluster solution though, meaning that the clusters are not very different from each other.

Table 5. BIC Comparison for Cluster Solution 1.

Number of clusters BIC Change in BIC

1 7575.620

2 7487.066 -88.555 3 7587.800 +100.735 Source: Author’s own illustration.

Therefore, I aim to improve the cluster solution to derive more meaningful results. Hence, I test a model solely including the touchpoints the customer uses during the customer journey which are of primary interest. This results in 26 clustering variables and a fair cluster quality solution indicating that the clusters differ. The most important predictor variables are post-purchase

offline Word-of-Mouth, pre-purchase physical store of another retailer, pre-purchase social media, pre-purchase leaflet of another retailer, and pre-purchase offline Word-of-Mouth. SPSS

(41)

Table 6. BIC Comparison for Cluster Solution 2.

Number of clusters BIC Change in BIC

1 3252.473

2 2990.883 -261.590 3 2853.149 -137.734 4 2833.990 -19.159 5 2852.980 +18.990 Source: Author’s own illustration.

Thus, forcing four clusters still leads to a fair cluster quality, but changes the most important predictor variables to Purchase, pre-purchase online shop, pre-purchase social media,

post-purchase offline Word-of-Mouth, and pre-post-purchase physical store of another retailer. The

cluster sizes are 88 (43.6%), 62 (30.7%), 18 (8.9%), and 34 (16.8%). The third cluster could be described as an online-prone cluster. However, the cluster size is extremely small and cluster one is almost five times as large. This also becomes an issue when conducting the regression analyses per cluster. Taking this into consideration, I will continue with the three cluster solution even though it does not have the lowest BIC. The change in BIC from a three cluster solution to a four cluster solution is minor though while a change from three to two clusters results in a larger change in BIC. The ratio of the largest to the smallest cluster further decreases to 2.38. Therefore, three clusters seem to provide a meaningful solution.

For validation purpose, I randomly exclude 10% of the sample and conduct a new cluster analysis. I repeat this process ten times (see table 7). As in the complete data set, the BIC mostly indicates a four cluster solution. A five cluster solution would be superior twice. The change in BIC is always only minor though. A three cluster solution would also be superior twice. Even though the result is slightly ambiguous, considering the sample size, a three cluster solution seems to be reasonable even in the validation tests.

Table 7. Cluster Validation.

Number of Clusters BIC 1a BIC 2a BIC 3a BIC 4a BIC 5a 1 2989.42 2834.22 2959.39 3046.34 3056.89 2 2763.69 2610.37 2745.42 2805.12 2826.30 3 2722.66 2532.87 2690.15 2714.79 2771.99 4 2701.34 2500.44 2668.81 2687.30 2747.25 5 2492.62 2650.72 Number of Clusters BIC 6a BIC 7a BIC 8a BIC 9a BIC 10a

1 2945.19 3002.22 2733.92 2881.75 2733.92 2 2745.24 2789.77 2554.72 2657.83 2554.72 3 2671.19 2663.75 2479.11 2567.43 2479.11 4 2617.86 2478.27 2478.27

a_{Number of validation check. Each time 10% of the sample are randomly excluded from the Cluster analysis.}

(42)

do not use any online touchpoints in the post-purchase stage at all, and only rarely use online touchpoints in the pre-purchase stage (mean: 0.1364). Further, 14 out of 25 touchpoints are not used at all by any customer in segment one. In the pre-purchase stage, the only touchpoint that was used by the majority is the physical store and in the post-purchase stage the consumption or usage touchpoint. The customers in this segment all purchase their product in the physical store. Thus, the customers in this segment aim to have an efficient shopping trip without using a lot of touchpoints and buy their product frequently, which is why I classify them as

experienced pragmatic shoppers.

The second cluster does not differ much in the touchpoints which are used by the majority of customers in this cluster from cluster one. Again, solely the physical store in the pre-purchase stage and the consumption or usage in the post-purchase stage are used by the majority. However, the physical store of another retailer is also used by a relatively large number of customers in this segment (approx. 60.0%). The same applies to the offline Word-of-Mouth in the pre-purchase stage (46.4%) and in the post-purchase stage (48.0%) indicating that the customers tend to share their opinion with others. 87% shop in a physical store. Nevertheless, this cluster is the only cluster that also contains customers that use the online shop. Customers in this segment use more touchpoints than the experienced pragmatic shoppers even if only just (mean pre-purchase: 2.2727; mean post-purchase: 1.4675). Again, the customers use more touchpoints in the pre-purchase stage. Most touchpoints used are offline (mean pre-purchase: 1.8052; mean post-purchase: 1.3117). Even though this segment includes the online shopper, the number of online touchpoints is extremely low (mean pre-purchase: 0.4416; mean post-purchase: 0.1558). The low number of touchpoints used also indicates that the customer does not want to occupy himself/herself too much with the shopping process in these product categories. However, they do talk about the product indicating more involvement, which leads to the classification as receptive pragmatic shoppers.

(43)

leaflet of another retailer and perform peer observation, respectively. In the post-purchase stage, the majority of the cluster members talk about the product in an offline environment (offline Word-of-Mouth) and consume or use the product (consumption or usage). Considering these characteristics leads to the classification of the cluster members as engaged multi-channel

shoppers.

(44)

Table 8. Segment Description.

Segment 1 Segment 2 Segment 3 Age 55.7% younger than 35 years

(index: 0.8582) 70.1% younger than 35 years (index: 1.0801) 75.7% younger than 35 years (index: 1.1664) and 48.6% are between 25 to 34 years (index: 1.1970)

Gender Females: 65.9% (index: 0.9940)

Females: 62.3% (index: 0.9397)

Females: 75.7% (index: 1.1418) Education Mainly university degree:

51.1% (index: 0.9060) or A- levels: 34.1% (index: 1.1482)

Mainly university degree: 61.0% (index: 1.0816) or A-levels: 24.7% (index: 0.8317)

Mainly university degree: 59.5% (index: 1.0550) or A- levels: 29.7% (index: 1.0000)

Income Rather equally distributed (below 1000€: 20.5%, index: 0.8436;; 1000€ - 2000€: 18.2%, index: 0.8161;; 2001€ - 3000€: 18.2%, index: 0.9681;; 3001€ - 4000€: 14.8%, index: 1.1473;; more than 4000€: 14.8%, index: 1.4231)

54.5% earn up to 2000€ (index: 1.1695)

Most earn under 1000€ (29.7%;; index: 1.2222), 2001€ - 3000€ (24.3%;; index: 1.2926) or 1000€ - 2000€ (18.9%;; index: 0.8475) Area Urban area: 58.0% (index:

0.9080)

Urban area: 63.6% (index: 0.9953)

Urban area: 78.4% (index: 1.2269) Journey

Duration

43.2% only a few moments (index: 1.1967) and 73.9% up to one day (index: 1.0438)

Mainly a few moments (29.9%;; index: 0.8283) or one day (16.9%;; index: 1.4202) or 2 – 3 days (16.9%;; index: 1.0060)

Mainly a few moments (32.4%;; index: 0.8975) or 2 – 3 days (21.6%;; index: 1.2857) Time Pressure 89.8% at least partly

concerned (index: 0.9804)

94.8% at least partly concerned (index: 1.0349)

89.2% at least partly concerned (index: 0.9738)

Price

consciousness

75% at least rather price conscious (index: 0.9908d)

76.6% at least rather price conscious (index: 1.0119)

75.7% at least rather price conscious (Index: 1.000) Frequency of

Buying

84.1% buy the product at least rather frequently (index: 1.1350)

64.5% buy the product at least rather frequently (index: 0.8704)

70.3% buy the product at least rather frequently (index: 0.9487) Product

Importance

For 66.7% at least rather important (index: 0.9926)

For 66.2% at least rather important

(index: 0.9851)

For 70.3% at least rather important (index: 1.0461) Shopping

experience

Experienced in a physical store, but not in an online shop

Experienced in a physical store and at least a little experienced in an online shop (28.6% have

experience;; index: 1.5628)

Experienced in a physical store, but not in an online shop Product Category Sweets (58.0%;; index: 1.1373);; coffee (28.4%;; index: 1.1687);; alcohol (13.6%;; index: 0.5484) Sweets (44.2%;; index: 0.8667);; alcohol (32.5%;; index: 1.3105);; coffee (23.4%;; index: 0.9630) Sweets (48.6%;; index: 0.9529);; alcohol (35.1%;; index: 1.4153);; coffee (16.2%;; 0.6667) The segment description is based on frequencies.