Segmenting customers based on touch points along their customer journey on digital platforms

(1)

Segmenting Customers Based on Touch Points

along their Customer Journey on Digital Platforms.

Name: Josje Helmond Student number: 11152583 Date: February 27th

2017

Master: MSc. in Business Administration - Marketing track Supervisor: Umut Konuş

(2)

University of Amsterdam _II

Statement of Originality

This document is written by Josje Helmond, student Business Administration, Marketing track, who declares to take full responsibility for the contents of this document. I, Josje Helmond, declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The faculty of Economics and Business of the University of Amsterdam is solely responsible for the supervision of completion of the work, not for the contents.

(3)

University of Amsterdam _III

Acknowledgement

The completion of this thesis symbolizes a concluding piece of my academic time at the University of Amsterdam. I proudly present this reasearch to you, as it resembles my academic growth of the last 1,5 years. Additionally, this thesis has contributed on a professional level to my position as an e-commerce CRM marketer. Lastly and most importantly, I have really enjoyed the taken steps in this process e.g. ‘the touch points in this personalized digital customer journey’.

I would like to take this opportunity to thank my supervisor Umut Konuş. It was a real pleasure working with him. Next to constructive, clear and effective meetings, fruitful and inspiring conversations occurred as we share the common interest of digitalization and the endless range of opportunities it unlocks. To me, Umut has not only contributed academically, but also professionally and personally. For this, in combination with his involvement and valuable advice throughout the whole process, I want to give him many thanks. Furthermore, I would like to thank the Marketing Manager, Business Intelligence Manager, two Business Intelligence Analysts and the Legal advisor of the involved company for their efforts regarding the quality and feasibility of this research. To conclude, I would like to thank my family, my girlfriend and my friends for their loving support.

Kind regards, Josje Helmond University of Amsterdam

(4)

University of Amsterdam _IV

Abstract

In this digital age, an explosive and increasing amount of touch points –interaction between customer and company- takes place on digital platforms. All these touch points together, including the order and patterns in which they are experienced on different devices, can be seen as a Customer Journey. This research suggests touch points on digital platforms as valuable segmentation basis. Namely, by segmenting on actual behaviour -containing profound information on Online Experiences, Device use and Platform use-, identified segments will contain in-depth segment-specific characteristics. Hence, predictive value lies in the possibility of managing these segments accordingly, tailoring communication strategies and campaigns for different customer journey-segments in the most effective way possible, thus increasing

business results. Although interesting segmentation bases have been introduced in past research, this research addresses a gap in the literature as the suggested

segmentation base has never been used before. Consequently, academic knowledge of the increasingly relevant online market is extended. Next to the efforts of this research to segment online customers based on digital touch points using the method Latent Class Cluster Analysis -revealing profound latent segments-, this research goes a step beyond by executing a profitability segmentation. Accordingly, behavioural and relational drivers of profitable customers within each segment become visible. Five online segments have been discovered; Desktop-Lovers, Young-Mobile-App-Lovers, Older-Tablet-Lovers, Mobile&Tablet-Web-Lovers, and Multichannel-Bigspenders. Furthermore, multichannel use is identified as main behavioural driver of profitability. Additionally, other segment-specific profit drivers are identified for each segment.

(5)

University of Amsterdam _V

Statement of Originality

………...II

Acknowledgement

………....….III

Abstract

………..……..IV

1. Introduction

……….….…….……….…..1

2. Literature Review

……….……..…5

2.1 Customer Journey………...……...5 2.1.1 Customer experiences………..……….……5

2.1.2 Phases of a customer’s buying process……….6

2.1.3 The Journey………...8

2.2 Digital Customer Journey………...12

2.2.1 Digital touch points……….12

2.2.2 Moments of Truth………. ..14

2.2.3 Micro-moments………...15

2.2.4 Conversion, attribution and path of purchase………...16

2.2.5 Digital customer journey in practice………...20

2.3 Digital Customer Journey and Experiences as a Segmentation Base.21 2.3.1 Customer Segmentation in Marketing………21

2.3.2 Behavioural Segmentation………..22

(6)

University of Amsterdam _VI

h Behaviour………24

2.3.4 Customer Journey Touch Points as a Segmentation Base……..28

2.3.5 Segmenting Customer on the Basis of Their Online Experiences………31 2.3.6 Segment Expectations………33

3. Conceptual Framework

………34

3.1 Framework………..34 3.2 Indicators/ Predictors……….38

3.2.1 Online experiences……….39 3.2.2 Device………39 3.2.3 Platform……….40

3.3 Covariates………....40 3.3.1 Active covariates………....40 3.3.2 Passive covariates………..40

4. Methodology

……….…………41

4.1 The Sample……….41 4.1.1 Sample description………...41 4.1.2 Sample selection………...42

4.2 Research Design……….42

4.3 Customer Segments Based on Online Touch Points………...45

4.3.1 Method: Latent Class Cluster Analysis……….45

4.4 Profitability Segments and Their (Varying) Touch Points…………46

4.4.1 Method: Latent Class Regression……….46

(7)

University of Amsterdam _VII

5. Results and Analysis

……….…..50

5.1 Preliminary and Exploratory Analysis……….51

5.2 Results: Customer Segments Based on Online Touch Points…….51

5.2.1 Model Selection……….51

5.2.2 Latent Class Cluster Analysis………....54

5.2.2.1 Indicator Analysis………...54

5.2.2.2 Covariate Analysis………..59

5.2.2.3 Segment labeling……….61

5.3 Results: Profitability Segments and Their (Varying) Touch Points………63

5.3.1 Model Selection……….63

5.3.2 Latent Class Regression……….64

5.3.2.1 Explaining Segment Parameter Estimates………..64

5.3.2.2 Profile Analysis………...67

5.3.2.3 Explanatory power………..71

6. Discussion and conclusions

………..72

6.1 Discussion………. ………...73

6.2 Overall Conclusion………. ………75

6.3 Theoretical Implications……….78

6.4 Managerial Implications……….79

6.5 Limitations and Future Research………..81

7 Appendices

………..83

Appendix 1: The customer journey in practice………...83

Appendix 2: Parameter output during model selection (LCR)…………...84

(8)

University of Amsterdam _VIII

Appendix 4: Descriptive statistics during model selection (LCCA)……86

Appendix 5: Descriptive statistics of indicators (LCCA)……….87

List of Tables

2.1 Online segmentation literature overview………25

2.2 Explanatory power and manageability of segmentation bases. ……….32

4.1 Measures of the indicator variables and covariates. ………..49

5.1 Profile of the segments, indicators (LCCA). ……….55

5.2 Platform and Device use per segment (LCCA). ………58

5.3 Profile of the segments, active covariates (LCCA). ………..59

5.4 Profile of the segments, passive covariates (LCCA). ………60

5.5 Overview of the characteristics of the five identified segments (LCCA)…………...62

5.6 Log-likelihood statistics for model selection (LCR). ……….63

5.7 Predictors parameter results (LCR). ………...65

5.8 Active covariates parameter results (LCR). ………...67

5.9 Profile of the segments, passive covariates (LCR) ………....68

5.10 Profile of the segments, active covariates (LCR) ………70

5.11 Behavioural and relational profitability drivers per segment (LCR). ………..71

6.1 Overview of the five identified segments based on online touch points………75

6.2 Behavioural and relational and profitability drivers per segment. ……….79

List of Figures

2.1 The AIDA-model. ………...7

2.2 Engel Kollat Blackwell Consumer Behaviour Model. ………7

(9)

University of Amsterdam _IX

2.4a Holistic view of a customer journey map. ………10

2.4b Identification of different paths to purchase. ………10

2.5 P&G’s Three moments of Truth. ………11

2.6 Increase of digital touch points. ………..12

2.7 Proposed online touch point taxonomy. ………..13

2.8 An example of the digital customer journey with several touch points across different moments in time. ……….14

2.9 ZMOT, FMOT, SMOT, UMOT………..15

2.10 Google’s Micromoments. ………..16

2.11 Conversions along the customer journey. ………...17

2.12 Conversion credit allocation to different channels. ………...18

2.13 Attribution models. ……….19

2.14 Google’s Path to Purchase tool. ………....20

2.15 Sungevity’s streamlined customer journey. ………...21

2.16 Classification of Segmentation bases. ………. ..22

2.17 Evaluation of Segmentation Bases. ………29

3.1 Conceptual framework research question 1a, using Latent Class Cluster Analysis….36 3.2 Conceptual framework research question 1b, using Latent Class Regression……….37

5.1 A visual representation of the LCCA BIC(LL) values, model 1 through 8………….53

(10)

University of Amsterdam

1

1. Introduction

Digitalization has ‘made a dent in the universe’ –Steve Jobs-, changing the way we eat (JustEat, TakeAway, Deliveroo, Foodora, UberEats), sleep (Airbnb,

Couchsurfing), travel (Uber, Blablacar, Snappcar), communicate (WhatsApp, Facebook, Snapchat, Pinterest, Tinder), search (Google), enjoy (Netflix, YouTube, Instagram), work and live (Apple’s hardware & software) in a lightning pace. On November 3rd 2014, the digital world passed a milestone when the amount of global internet users passed the 3 billion (We Are Social, 2014). Last year, 2016, the amount surpassed the 3.4 billion Internet users (Digital Agency Network, 2016). As a

consequence of this rapid growth of digitalization, people have a lot more online experiences than they had 15 years ago; offline experiences have been fairly complemented and enriched by online and mobile experiences. These customer experiences have often been addressed in literature as the purchase process and several models consisting of multiple phases have been introduced. Today, another perspective has arisen now that proposed models fail to adequately capture the online purchase process. The rapid increase of the use of mobile devices means that

consumers can now take action to address their needs immediately and Google (2015) refers to these instances as micro-moments: moments when a consumer reflexively turns to a (generally mobile) device to fulfill an immediate need. All these micro-moments or so-called touch points together including the way, order and patterns in which they are experienced can be seen as a journey: the customer journey. This new customer journey perspective, meaning all interactions between a business and its consumers in its presence today, would have been unimaginable a few decades ago. Today, virtual places enable many more interactions and many more possibilities across global distances (Norton and Pine, 2013).

(11)

2

This research introduces customer specific touch points of the digital customer journey as a customer segmentation basis. It is argued that customer journey touch points make a good segmentation basis as specific in-depth behaviour around purchases can be identified and latent –present but not visible- differences between specific customer groups will become visible. Research shows that this transactional data –actual behaviour- is a good base for predicting future customer behaviour and tailoring strategy and communication campaigns accordingly, thus showing value of the proposed segmentation based. Hence, research question 1a of this research is the following: Can we talk about distinct customer segments identified on the basis of customers’ behaviour along multiple touch points on online platforms? Building on to this, in order to identify drivers of profitability, research question 1b of this research is proposed: Are there distinct customer segments which can be identified on their profitability that (may) depend on different behavioural and relational drivers? The execution of a segmentation study based on experiences along the digital customer journey (1a), in combination with a deeper understanding of profitability drivers as is being done in this research (1b), is considered as highly important for one main reason; crucial managerial implications.

Taking into account the rapid growth of digitalization and significant influence on our way of living it is increasingly necessary to gain profound

understanding of online customers in order to design effective customer strategies. The possibility to identify distinct groups of customers and distinct profit drivers within each specific segment enables practitioners to effectively tailor offerings to customer types (Rohm and Swaminathan, 2004). Different strategies and campaigns for different customer journeys and different customer journey segments can be developed on the basis of a customer journey touch points-based segmentation. The

(12)

3

obtained segments and insights in these segments regarding their profit drivers and their relational, demographical and behavioural characteristics results in the ability to attract new customers, keep existing customers, and thus growing the customer base. Accordingly, decisions concerning what channels or touch points to invest in can be addressed. In turn, these strategic decisions taken by companies will affect the customer value and the turnover potential (Bhatnagar and Ghose, 2004). The success of these strategic decisions, so the effectiveness in reaching the relevant and

measurable KPI’s (Key Performance Indicators) depends on the quality of the identified segments, and thus on the segmentation base itself; ’The better

identification of your segments the better communication strategies can be developed’ (Thomas and Sullivan, 2005). More specifically, by segmenting customers on the basis of their customer journey touch points using transactional data, this

segmentation will reveal latent segments containing specific customer journey touch points characteristics and will thus have more predictive value compared to other segmentation bases as Demographical or Motivational bases.

As the ultimate desire of practitioners is to be able to manage customers - tailor personal communication strategies which work best and are most desirable for those specific customers- it is argued that segmenting on the basis of the experiences along the customer journey -making latent variance in preferences and behaviour visible- holds significant managerial value as it makes it possible to manage segments in the most efficient way possible. Past research shows many online segmentations based on interesting bases and variables, contributing to the literature. However, being able to manage segments on the basis of customers’ actual behaviour data including many behavioural online touch points is non-existing in current literature. Additionally, segment specific knowledge on profit drivers along customer journey

(13)

4

touch points, something which is also non-existing is current literature, adds even more value to this research. Accordingly, the literature gap upon which this research is based is a segmentation base existing of customer experiences itself within the customer journey, making it possible to really manage the segments from a consumer perspective. In order to fill this gap, this research will entail an online segmentation based on experiences along the digital customer journey using three-dimensional (i) Online Experiences, ii) Device, iii) Platform) transactional touch points-data, where after the identified segments will be profiled according to relational, demographical and behavioural variables. In addition, this segmentation is expanded with the exploration of segment-specific profit drivers by executing a profitability segmentation.

The structure of this research is as follows. Firstly, the literature on (Digital) Customer Journey and Segmentation will be reviewed. Secondly, a visualization of the two conceptual frameworks of research question 1a and 1b will be given, followed by an explanation of segmentation indicators/ predictors, active covariates and

descriptive covariates. Thirdly, the methodology used will be explained followed by a detailed chapter on the segmentation results. Finally, a detailed discussion and final conclusion on findings followed by theoretical implications, managerial implications, limitations and future research suggestions will be given.

(14)

5

2. Literature review

2.1 Customer Journey

2.1.1 Customer experiences

One can look at life as a rollercoaster, passing through a lot of experiences –a first kiss, a new job, the discovery of new things or raising children- accompanied by a pallet of different feelings as –happiness, frustration or disappointment-. People also pass through experiences as a customer; customer experiences. These experiences can occur either offline –while shopping in-store or during the use of a product for

instance-, or online –when making an online transaction or reading an email for instance-. According to Schmitt (1999), the customer experience itself includes all the moments of contact and emotions during the experience and consists of perceptions that shape emotions, thoughts, and attitudes (Beauregard, Younkin, Corriveau, Doherty and Salskov, 2007). According to Shaw and Ivans (2005), more than 50% of a customer experience exists of emotions as customers are people and people are driven by emotions. In general, practitioners and scholars have come to agree that the total customer experience is a multidimensional construct that involves cognitive, emotional, behavioural, social and sensorial components (Schmitt, 1999; Schmitt, 2003; Verhoef et al., 2009). However, Verhoef and Lemon (2016) argue that specific aspects of the offering such as technology (McCarthy and Wright, 2004) or brand (Brakus, Schmitt, and Zarantonello, 2009) could be related to the experience. Also, they argue that the experience consists of individual contacts between the firm and the customer at distinct points in the experience; touch points (Homburg, Jozic, Kuehnl, 2015; Schmitt, 2010). Therefore, it is argued that an experience is built up through a collection of these touch points in multiple phases of a customer’s buying process (Pucinelli et al., 2009; Verhoef et al., 2009). Based on this, Verhoef and Lemon

(15)

6

(2016) conclude that customer experience is a multidimensional construct focusing on a customer’s cognitive, emotional, behavioural, sensorial, and social responses to a firm’s offering.

2.1.2 Phases of a customer’s buying process

Looking at the foundations of Verhoef and Lemon’s (2016) definition of customer experience, taking a holistic scientific perspective on customer experiences overtime, different authors have introduced models in which customer experiences have been divided in different phases. The first person to introduce a model addressing these clustered experiences within stages of a customer’s buying process was E. St. Elmo Lewis in 1989. He developed a model that includes customer experiences from the moment a brand or product attracts consumer attention to the point of action or purchase (Strong, 1925). This first model is often referred to as the AIDA-model, which represents the phases Awareness - the customer is aware of the existence of a product or service-, Interest – the customer is actively expressing an interest in a product group-, Desire – the customer is aspiring to a particular brand or product-, and Action – the customer is taking the next step towards purchasing the chosen product-, see figure 2.1. The AIDA-modelhas been frequently illustrated in the diagrammatic format of a funnel, indicating that a larger quantity of potential customers become aware, where after a smaller subset becomes interested, then an even smaller part will experience desire and only few potential customers will take action towards a purchase. This phenomenon is often referred to as a “purchase funnel’’ “customer funnel,” “marketing funnel,” or “sales funnel”. Later, new phases were introduced as an extension of the AIDA-model, such as satisfaction (AIDAS) (Sheldon, 1911) and confidence (AIDCAS) (Kitson, 1920).

(16)

7

Figure 2.1. The AIDA-model. Source: KnowledgeBrief.

Since the introduction of the AIDA-model other authors introduced models where experiences within the buying process where represented in phases as well. In 1968, Engel, Blackwell and Kollat developed a model, the EBK-model, of the consumer buying decision process where they included a feedback phase. Their model consists of five steps: i) problem/need recognition, ii) information search, iii) evaluation of alternatives to meet this need, iv) purchase decision and v) post-purchase behaviour (see figure 2.2).

Figure 2.2. Engel Kollat Blackwell Consumer Behaviour Model.

Source: Alvarez and Casielles (2008).

(17)

8

Furthermore, Shaw and Ivens (2005) introduced a model where they have broken down customer experience in five stages: the expectations setting; the pre-purchase; the purchase; the product or services consumption; and the post-experience review.

2.1.3 The Journey

With the increasing amount of customer-company touch points as time goes by, nowadays, the concept of a funnel fails to capture all touch points. A more sophisticated approach is required to help marketers navigate and understand this customer experience-environment, less linear and more complicated than the funnel suggests; the consumer journey (McKinsey, 2009), see figure 2.3.

(18)

9

Customer journey means the sequence of events that customers go through to learn about, purchase and interact with company offerings, including commodities, goods, services or experiences (Norton and Pine, 2013). It entails a customer-specific

personalized journey, opening up countless different combinations of customer paths, varying in experienced touch points, sequence, timing and frequency. Nenonen, Rasila, Junnonen, and Kärnä (2008) describe the customer journey as the cycle of the relationship interaction between the customer and the company, where customer journey mapping is a visual and process-oriented method for conceptualising and structuring people’s experiences. In 1999 the customer journey was mapped for the first time through the Acela high-speed rail project of IDEO (Yoo and Pan, 2014). See figure 2.4a for an example from a holistic view of a customer journey map. Customer mapping makes it possible to identify different customer journeys for different segments, making different paths to purchase visible (figure 2.4b). For instance, in practice, customers which take different paths to purchase, thus which have different journeys, could have had exactly the same customer-company touch points but in a different order. Moreover, people with completely different journeys could end up buying exactly the same product, and people with very similar journeys could end up buying very different products, or they can differ in purchase frequency and value. Consequently, myriad numbers of journeys could exist for one company.

The customer journey helps organisations understand how prospective and current customers use the various channels and touch points, how these customers perceive the organisation at each touch point and how they would like the customer experience to be.

(19)

10

Figure 2.4a. Holistic view of a customer journey map. Source: Maggi (2012).

(20)

11

This knowledge can be used to design an optimal experience that meets the

expectations of major customer groups, achieves competitive advantage and supports attainment of desired customer experience objectives (Nenonen et al., 2008). In 2005, A.G. Lafley, Chairman, President and CEO of Procter & Gamble introduced two and somewhat later three Moments of Truth (MOT); moments within the customer

journey where a customer interacts with a brand, product or service to form or change an impression about that particular brand, product or service. The First moment of truth (FMOT) takes place when a customer is confronted with the product in-store or in real life. The Second moment of truth (SMOT) takes place when a customer purchases a product and experiences its quality as per the promise of the brand. The Third moment of truth (TMOT) takes place when consumers feedback or reaction towards a brand, product or service i.e. consumer becomes brand advocate and gives back via word of mouth or social media publishing (Cohen, 2013), see figure 2.5.

(21)

12

2.2 Digital Customer Journey

2.2.1 Digital touch points

Due to the explosion of digital channels, technology innovation and product choices, the moments where customers interact with a company –touch points- have increased vastly. Moments where advertisements were seen or where the product itself was inspected are nowadays complemented with digital touch points such as receiving push notifications via an application, personalized e-mails such as recommendations for you based on your online behaviour or online cross-sell opportunity’s, see figure 2.6.

(22)

13

Other digital touch points are customer care interaction via Twitter, Facebook or Whatsapp, interaction on social media, reading a companies’ Blog or watching a companies’ Vlog, Pinterest, Instagram or My story on Snapchat. Anderl, Schumann and Kunz (2016) proposed an online touch point taxonomy, in order to create distinctness in this constantly changing and growing field (figure 2.7). In this

taxonomy, the significant and increasing power of the customers’ voice and actions in the online environment is taken into account by introducing two main categories ‘Customer-initiated channels’ and ‘Firm-initiated channels’.

Figure 2.7.Proposed online touch point taxonomy. Source: Anderl et al. (2016).

With the exponential growth of digitalization, touch points in the customer journey has grown exponentially. So basically, the digital customer journey is a sophisticated and extended version of the initial customer journey, with many online touch points overtime. See figure 2.8 for an example of a digital customer journey and its

(23)

14

Figure 2.8. An example of the digital customer journey with several touch points across different moments in time.Source: Crownpeak (2014).

2.2.2 Moments of Truth

Because of the rise of full internet adoption and the corresponding increased search engine use, social media use and review use, many consumer-company interactions take place before that consumer ever sees a product on a shelf or in a webshop (Google, 2011). According to research conducted by Google, 88% of US customers are researching online before actually buying the product (Cohen, 2013).After detecting this development, Google (2011) introduced another moment of Truth in addition to P&G’s three moments of truth, the Zero Moment of Truth (ZMOT). This Zero Moment of Truth refers to the moment when the user is searching online for a product with an intention to understand the product that he or she intends to consume or buy (Brandalyser, 2011). So hereafter, the Three moments of truth had become Four moments of truth. In the following years, due to the extensive use of the internet, the world became ultra-connected; consumers’ first impressions of a brand often

(24)

15

comes from fellow consumers sharing experiences online (LostinGoogle, 2015). In this light, it seemed that something important was missing to get the whole picture of the digital customer journey. So in 2014, Brian Solis, successful author and digital analyst at Altimer Group, introduced the notion of “shared experiences” which is at the core of the “Ultimate Moment Of Truth” model (see figure 2.9).

Figure 2.9. ZMOT (Google, 2011), FMOT, SMOT, UMOT (P&G, 2005).

2.2.3 Micro-moments

In 2015,the ongoing exponential growth of touch points in the digital journey is expanding in such a rapid pace, that the concept of just a few ‘moments of truth’ doesn’t seem to cover as a whole what is happening anymore. In the spring of 2015 Google introduced the micro-moments framework. This framework addresses the insight that through the proliferation of mobile devices, consumers can now take action to address their needs immediately. Important instant-experiences among all touch points arise, moments a company wants to be ready for, with the right personal

(25)

16

message at any moment. Google refers to these instances as micro-moments; moments when a consumer reflexively turns to a (generally mobile) device to fulfill an immediate need. Nowadays, consumers check their phone 150+ times a day (Google, 2016). These micro-moments are classified as one of the following; I-want-to-know moments, I-want-to-go moments, I-want-to-do moments and I-want-to-buy moments (see figure 2.10).

Figure 2.10. Google’s Micromoments. Source: Google (2015).

These micro-moments are all critical touch points within today’s customer journey. Using the micro-moments framework is an excellent way to understand the online customer journey from search discovery intent through to purchase intent (Yi, 2016).

2.2.4 Conversion, attribution and path to purchase Conversion

Along all customer-company touch points within the customer journey, conversions take place (figure 2.11). A conversion is a desired action of the company, taken by the

(26)

17

customer, such as providing an email address to receive more information or

purchasing a product in response to an advertising (Burt, Dunham, Ives and Jarvinen, 2010). Different metrics are used to measure the effectiveness of the use of

communication channels. For advertisements on affiliate sites –other websites providing commercial space- for instance, it is likely that a Click Through Rate (CTR) is measured. For instance, a total amount of views counts 1000, and the total amount of clicks is 20. The CTR is then 20/1000= 2%. However, other metrics instead of clicks can be used to measure specific conversion rates within a customer journey, for instance views, opens, purchases, likes or comments on Facebook, or other company, industry or channel specific conversions.

(27)

18

Attribution

While marketers employ various online channels to reach consumers, such as Facebook (paid), Facebook (unpaid), SEA (Search Engine Advertising, Google’s advertising tool), Instagram, Pinterest, Snapchat, Affiliates, E-mail and optional Push Notifications (via an application), it is often not known to what degree each channel actually contributes to the overall marketing success (Anderl, Becker, Wangenheim and Schumann, 2014). Allocating credit to each channel is called attribution.

Attribution is a challenge (Neslin and Shankar, 2009), which involves finding ways to measure “the partial value of each interactive marketing contact that contributed to a desired outcome” (Osur, 2012) (Figure 2.12). The purpose of attribution is to quantify the influence each marketing impression has on a consumer’s decision to make a purchase decision, or convert. Based on this knowledge, strategic decisions for marketing campaigns and budgeting can be made, performance can be evaluated and KPI’s can be set and optionally be adjusted accordingly.

Figure 2.12. Conversion credit allocation to different channels. Source: Yamaguchi (2014).

Different attribution models are used to allocate credit among different channels (figure 2.13). For example, in “last interacion’’ or in other words ‘’last clicks wins’’ the value gets attributed solely to the marketing channel that directly preceded the

(28)

19

conversion (Econsultancy, 2012; The CMO Club & Visual IQ, Inc., 2014). Any prior customer interactions are disregarded.

Figure 2.13. Attribution models. Source: Nebo (2013).

Path to purchase

Looking at the customer journey from the perspective of conversion, a different view of the customer journey is the perspective of a path to purchase. A path of purchase is defined as a path that includes all contacts of any individual customer over all online marketing channels, prior to a potential purchase decision (Haan, Wiesel, and

Pauwels, 2013; Xu, Duan, and Whinston, 2014). In fact, there can be several paths to purchases in one customer journey. In line with the customer journey, the idea of a path to purchase sequence holds that consumers proceed through a series of stages on the path to purchase. This series begins with awareness and knowledge-building (cognition or thinking), followed by liking and preference (affect or feeling) and ends with conviction and purchase (Srinivasan, Rutz and Pauwels, 2016). Multiple

pathways exist for the consumer’s path to purchase (Vakratsas and Ambler, 1999). In order to provide a tool to make the path to purchase visible, Google developed a Path to Purchase-tool called “Customer Journey to Online Purchase” (Figure 2.14).

(29)

20

Figure 2.14. Google’s Path to Purchase tool. Source: Google (2013).

2.2.5 Digital customer journey in practice

An example of a company that is shaping the customer journey in a seamless way, personalizing every touch point, is California-based solar panel provider Sungevity which is operating worldwide. See appendix 1 for detailed information on their step by step personalized customer journey. In short, Sungevity customizes and automates each touch point/ step of the journey, making it so simple and compelling to move from one step to the next that it is hard to even consider alternative providers. In essence, the company jumps some steps in the decision journey, immediately

minimizing the choice of the consideration set to only its own company, streamlining the evaluation phase and delivering the customer directly into a “loyalty loop,” where a monogamous and open-ended relation remain with the firm, see figure 2.15. For Sungevity, its personalized journey strategy is working. Sales have doubled in 2014 to more than $65 million, exceeding growth targets and making Sungevity the fastest-growing player in the residential solar business (Edelman and Singer, 2015).

(30)

21

Figure 2.15. Sungevity’s streamlined customer journey. Source: Edelman and Singer (2015)

2.3 Digital Customer Journey and Experiences as a Segmentation

Base

2.3.1 Customer Segmentation in Marketing

Since the concept of segmentation emerged in 1956, introduced by Smith, it has been one of the most researched topics in marketing literature (Wedel and Kamakura, 2012). As stated by Smith (1956), market segmentation involves viewing a

heterogeneous market as a number of homogeneous markets, in response to differing preferences, attributable to the desires of consumers for more precise satisfaction of their varying wants. This concept has led to research that divides markets into

homogeneous sub-markets in terms of customer demand (Dickson, 1993), resulting in the identification of groups of consumers that respond similarly to marketing

activities (Wedel and Kamakura. 2012). This makes market segmentation one of the most one of the most efficient and promising tools in a marketer's toolbox

(Fachantidis, Tsiaras, Tsoumakas and Vlahavas, 2016).

Segmentation is an essential element of marketing since business cannot exist without considering customers’ needs and recognizing the heterogeneity of those

(31)

22

needs. Firms that are able to identify specific needs for specific subgroups and to develop the right tailored offer thus create a competitive advantage. Questions as why people behave differently, why they buy differently and why their preferences are different can be explored and understood on the basis of segmentation.

The way the market is divided can be done in various ways, using one or combining several segmentation bases. A segmentation basis is defined by Wedel and Kamakura (2012) as a set of variables or characteristics used to assign potential customers to homogeneous groups.

2.3.2 Behavioural Segmentation

Looking at the segmentation literature, there are literally millions of ways to divide up the market (Aaker, 1995). Wind (1978) has introduced a currently still frequently used classification of segmentation bases, see figure 2.16. Even though the

classification is dated, Wagner and Kamakura (2012) state that despite substantial developments in IT and data analysis technology, no significant scientific

developments in market segmentation has appeared since its publication.

(32)

23

Demographic bases are argued to be easy to use but also to be limited in utility. Also, it has been suggested in previous research that a segmentation based on either

demographic or socio-demographic bases are of limited value when consumer behaviour is being investigated (Sell, Mezei and Walden, 2014). Furthermore, many segmentations are done on the bases of motivations or attitudes. In 1934, LaPiere concluded that there is a large gap between attitudes and behaviour, and that therefore, data obtained from a questionnaire could not always be trusted to be reliable (Armitage and Christian, 2003). This is in line with the research by

MacDonald, Wilson and Konuş (2012), who argue that intentional or motivational data is flawed, since the respondent relies on its memory, which are often biased by context and which decay rapidly. In light of the gap between intentions and

behaviour, it is preferable to segment on actual behaviour e.g. transactional data, as is being done in this research. Namely, if one aims to use the segmentation results for marketing practices research has indicated that transactional data is a better predictor of future behaviour (Armitage and Christian, 2003). Also Sell et al. (2014)

emphasizes the predictive value of segmenting on the bases of experiences, which is in line with Wedel and Kamakura’s research (2000), who argue that there has been increased recognition in recent years of the value of segmentation based on actual and experimental choice data (Bock and Uncles, 2002).

Segmentation has proven to be a very useful concept and has recently become very effective in industries where customer retention is an important goal; identifying, profiling, targeting and reaching segments becomes possible using customer

transaction data (Wedel and Kamakura, 2002).Ultimately, obtained segment

knowledge based on transactional data enhances customer value thus increases firm’s profit potential (Bhatnagar and Ghose, 2004).

(33)

24

2.3.3 Customer Segmentation Literature on Online Customer Behaviour

The second fastest evolving discipline after information technology i.e. ITis

marketing. New approaches appear every day and strategies and tools that were first pre-cursors may have become obsolete today (Aarons, van den Driest and Weed, 2014). The rapid changes of both IT and marketing makes it extremely challenging for e-commerce companies to keep up and beat the competition. Therefore, it is argued that the understanding of online customer segments is of great value to marketers, who currently face the challenge of resource allocation across a broad range of touch points (Baxendale, MacDonald and Wilson, 2015).Hence, thorough understanding of the online market is needed in order to spend the budget effectively and to target specific segments with the right message via the right channel in order to retain and grow the customer base.

In past research, many online segmentations have been executed, making use of many different segmentation bases. However, segmenting the online market based on transactional data, with incorporation of platform- and device-use over a longer period of time, as is being done in this research, has not been done before (table 2.1). Very often, segmentations in past research are executed on the basis of survey or experiment-methods. Consequently, the discussed gap between intentional/ motivational data and actual behaviour is not being addressed.

(34)

25

Table 2.1. Online segmentation literature overview.

Relevance of online segmenting is proven in the literature where several researches have shown that online consumers have divergent purposes when using the internet (Kau, et al. 2003; Mathwick, 2001), revealing that online customers do no form one homogenous group (Aljukhadar, 2011).

Already in 1998 the identification of consumer segments was highlighted as

Authors & Publication year Segmentation base Multiple

devices / channels Longer period of time Transactional data used

Aljukhadar (2011) Internet use pattern ✓ ✓ ✗

Barnes, Bauer, Neumann and Huber (2007)

Constructs of purchasing behaviour on the internet

✗ ✗ ✗

Bhatnagar and Ghose (2004) Benefit perceptions and purchase behaviour

✗ ✗ ✗

Brengman, Malaika, Geuens, Weijters, Smith and Swinyard (2005)

Web-usage related lifestyle

✗ ✗ ✗

Bressolles, Durrieu, and Senecal (2014)

E-satisfaction ✗ ✗ ✗

Christodoulides, Michaelidou and Siamagka (2013)

Online affective states ✗ ✗ ✗

Ganesh, Reynolds, Luckett and Pomirleanu (2010)

Online Shopping Motivations, & e-Store Attributes

✗ ✗ ✗

Gehrt, Rajan, Shainesh, Czerwinski and O'Brien, (2012)

Shopping orientations ✗ ✗ ✗

Hill, Beatty, Walsh (2013) Online usage motives ✗ ✗ ✗ Kau, Tang and Ghose (2003) Buying behaviour ✗ ✓ ✗ Muthitacharoen, Gillenson and Suwan

(2006)

Sales channel ✓ ✗ ✗

Rohm and Swaminathan (2004) Shopping motivations ✗ ✗ ✗

Sell, Mezei and Walden (2014) Attitude ✓ ✓ ✗

This research (2017) Digital customer journey touch points

(35)

26

one of the necessary and important fields of research needed in e-commerce (Chang, 1998).This is also emphasized by Hoffman (2000) and Habul and Trifts (2000), stating that research focused on the Internet and its opportunities lags behind business practice pretty severe. Accordingly, Shim, Eastlick, Lotz and Warrington (2001) conclude that this lag already gave rise to several high quality researches but that there needs more to be done in order to understand the internet consumer

(Jayawardhena, Wright and Dennis, 2007). Nowadays, it is still argued that the knowledge in the area of consumer experiences in online shopping environments is emergent, which provides fertile ground for ongoing research (Klaus, 2013; Martin, Mortimer and Andrews, 2015; Rose, Clark, Samouel and Hair, 2012; Trevinal and Stenger, 2014).

Looking at the online segmenting literature, many researchers have done online segmentations, based on different segmentation and profiling bases. A limited amount of online segmentation researches show segmentation bases based on

consumer experience. Aljukhadar (2011) has segmented online customers based on the various uses of the internet. Despite the fact that the segmentation bases are experience based in contrast to motivations or attitudes, they did not use transactional data of the customers’ data use. Instead they sent a sample of 407 participants an online survey, which provided a pattern of Internet use, Internet experience, and psychological characteristics. Furthermore, Sell et al. (2014) segment consumers with regards to their usage of mobile technology. Again, for this ‘usage’-based

segmentation transactional data wasn’t used as segmentation base, but perceptions and socio-demographic variables were used instead. Brengman, Geunens, Weijters, Smith, and Swinyard (2005) established a life-style based segmentation of online shoppers. In this research a lifestyle scale was validated, and thus no transactional

(36)

27

data but an online survey was used as research strategy. Furthermore, Kau et al. (2003) segmented the online market based on buying behaviour. However, the research strategy used was a questionnaire, asking customers about their buying behaviour and information-seeking patterns. Thus, in line with other mentioned past online segmentation studies on customer behaviour, the data used thus wasn’t transactional data representing actual customer behaviour.

The grounds for the argument presented that segmenting on the basis of various touch points in the customer journey using transactional data is valuable are the following. Firstly, the managerial implications are of great value. Namely, a segmentation based on touch points along the consumers’ personal journey contains high descriptive/predictive power. This mainly stems from the three segmentation characteristics outlined in figure 2.1, as this research meets all three, where obtained segments will be accompanied by profound information on behavioural

characteristics. In line with this, Kau, Thang and Ghose (2003) emphasize the existence of important implications for predicting future customer behaviour when a segmentation is based on actual experiences. Namely, the overall predictive power of a segmentation based on customer journey touch points on digital platforms is the in-depth information on who these customers are, what they do, and thus, what they are likely to do in the future. In this way, varying preferences and behaviour can be interpreted, resulting in the opportunity to tailor communication strategies efficiently. Consequently, more positive business results will occur, as is substantiated in

subchapter 2.3.4. Secondly, a customer journey touch points based segmentation is a non-existence research which adds knowledge to science in its current state. This research fills an existing gap in literature, therefore contributing knowledge to science and thus to humanity itself.

(37)

28

To summarize, looking at the current online segmenting literature, shortcomings are found in the predictive value of used segmentation bases. This means a gap becomes visible for segmenting online customers on the basis of multiple touch points of the customer journey using transactional data, containing device-specific and multichannel characteristics. A segmentation in this form carries latent and predictive value for future customer behaviour and therefore holds implications for practitioners in managing their online channels.

2.3.4 Customer Journey Touch Points as a Segmentation Base

An important topic in outperforming your competitors in this digital era is total experience. A research of Aarons et al. (2014) shows that total experience e.g. the combination of personalized offerings and the focus on the breadth of the relationship by adding touch points, is one of the characteristics of high performing brands. It is even stated that the researchers believe the future’s most important marketing metric will be ‘’share of experience’’. According to Keyser, Schepers and Konuş (2015) a thorough understanding of different customer segments and their unique

characteristics is necessary if one aims to deliver a unified customer experience. A segmentation of the online market based on touch points in the digital customer journey is therefore considered as highly important. As indicated, the reason for the value of this segmentation base is the extent to which variance between customers in preferences and buying behaviour can be explained. After all, this is the reason why we segment customers; it is a strategic marketing tool to define markets and allocate resources accordingly (Assael and Roscoe, 1976).

(38)

29

segmentation bases (see figure 2.17), managerial implications for segmentation based on past behaviour is shown.

Figure 2.17. Evaluation of Segmentation Bases. Source: Wagner and Kamakura (2012).

Touch points in the customer journey are categorized in the category Specific and Observable, as introduced in figure 2.17. As seen in Wagner and Kamakura’s (2012) research the criterion Responsiveness is satisfied. This means that segments will respond uniquely to marketing efforts targeted at them. As stated by Wagner and Kamakura (2012), responsiveness is critical for the effectiveness of any market segmentation strategy because differentiated marketing mixes will be effective only if each segment is unique and homogeneous in its response to them. In this research, 28 Specific and Observable variableswill be used as a basis of this segmentation.

(39)

30

Therefore it is argued that the responsiveness-potential is relatively to other segmentation bases really high. Furthermore, Wagner and Kamakura’s research (2012) also shows a plus, meaning good, on stability for Specific and Observable variables, see figure 2.17. As argued by Wagner and Kamakura (2012) only segments that are stable in time can be the underlying basis for the development of a successful marketing strategy, as the targeted segment change composition, the marketing effort is likely to be unsuccessful. Therefore, the satisfaction of Stability of touch points in the customer journey as segmentation basis is another argument for the use of this basis. Lastly, it is seen from figure 2.17 that the criterion Substantiality has two plusses, meaning very good for the basis used in this research. This contains huge value, since this entails that the targeted segments are expected to ensure the

profitability of targeted marketing programs; higher marginal revenues compared to the marginal cost for the firm (Wagner and Kamakura, 2012).

In line with the arguments made, in order to allocate resources and to make a personalized communication strategy based on the obtained segments, a requisite is that the segments are manageable. Only then practitioners can utilize segmenting the customer base by making knowledge-based decisions. In order to being able to manage segments, one should know which specific experiences influence specific customers in their future behaviour. In this light, answers on the following questions, which can be answered on the basis of this research, are very valuable. How many times did a customer purchased an item? How many times did a customer click in the email to visit the website, or to visit the application? What was the total visit time? What was the device used during that visit? Was it desktop or phone, and if it was a phone, was the platform entered the website or the application? If many of these experiences are known and combined with the problem solving method Latent Class

(40)

31

Cluster Analysis (LCCA), -as is being done in this research- it is possible to see what it is that drives specific customers and therefore predictive value is derived; latent segments become visible. After segmenting the customer base on the basis of customer journey touch points, it has become possible for marketers to develop different strategies and campaigns for different customer journeys and different customer journey segments. Based on this, it is argued that various touch points in the customer journey used as a segmentation basis and thereafter analysed using LCCA is a perfect segmentation base, holding significant managerial value.

2.3.5 Segmenting Customers on the Basis of Their Online Experiences

All in all, past research shows online segmentations based on interesting bases and variables, contributing to the literature, but none of them are based on actual behaviour or experiences along the digital customer journey. As a result, the explanatory power of segmentation done in past research is limited, as they cannot explain exactly what is influencing the segment and what is shaping their preferences. Accordingly, the segmentations done in the past are not fully manageable, meaning that information is missing in order to tailor personalized communication strategies for each discovered segment. This leaves a gap for an online segmentation based on transactional data/ online experiences along the customer journey, revealing latent segments and making it possible to manage them. See table 2.2for a visual

(41)

32

Segmentation bases Explanatory power, on a scale from 1 (no

power) to 5 (full power)

Manageable Demographic variables 1 ✗ Psychographic variables 2 ✗ Motivation variables 3 ✗ 1-dimensional Behaviour variables 4 ✗/✓

3-dimensional Customer journey Touch Points

5 ✓

Table 2.2. Explanatory power and manageability of segmentation bases.

This research includes segmentation/ indicator variables containing one and three dimensions; Online experiences, Device and Platform. This reveals the in-depth characteristics of customer behaviour, for instance: a customer clicked in an email (Online experience) on a phone (Device) to go to the website (Platform), because the application wasn’t installed by that specific customer. The inclusion of these three dimensions-touch points enhances the explanatory power of this research, thus increasing predictive value to tailor segmented based communication in the future. Next to the dependence on indicator variables, classification decisions on customer membership e.g. the decision to which segment a customer gets assigned, also depends on active covariates. After segments are identified on the basis of indicators and active covariates, they will be profiled by passive covariates in order to gain a better understanding of who these people are. Accordingly, the purpose of this research is to fill the identified gap by investigating research question 1a, which is presented as follows.

1a. Can we talk about distinct customer segments identified on the basis of customers’ behaviour along multiple touch points on online platforms?

(42)

33

On the basis of the first research question, an analysis of a digital touch point based segmentation using the method Latent Class Cluster Analysis is executed. Next to this, this research introduces an additional study to go a step beyond. On the basis of the same dataset, an additional segmentation will be executed with variable

TotalMoneySpent as dependent variable, using the method Latent Class Regression. This is done in order to identify drivers of profitability overall and within each specific segment. Accordingly, research question 1b is formulated as follows.

Are there distinct customer segments which can be identified on their profitability that (may) depend on different behavioural and relational drivers?

Similar to the research of Sell et al. (2014), this research will abide to the idea of Minhas and Jacobs (1996); first, interesting behaviours will be identified and thereafter consumers in those segments will be investigated in order to understand who these people are. This is in contrast with a more traditional approach where consumers are grouped based on observable general similarities where after their behaviour is investigated (Sell et al. 2014).

2.3.6 Segment Expectations

Based on business practice and existing literature, it is broadly expected that the segments found on the bases of experiences along the customer journey are in the direction of the following.

An ‘innovative’-segment is expected, where new options are explored but also where options are abandoned soon, when expectations are not fulfilled as wished for.

(43)

34

Thus, it is expected that such a segment is mainly using relatively new and modern technologies as an application. This might be a relatively young segment.

Also, a ‘stick with what is known’-segment is expected, where stable patterns occur and not a lot of discovery of new options is seen.

Furthermore, a ‘go with the flow’ is expected where patterns on first sight are hard to discover; behaviour is very dependent on events in the consumer’s

environment (spontaneous behaviour, influenced by trends, influenced by other people etc.). Thus, the use of different devices and platforms is expected.

Lastly, a ‘convenience’-segment is expected, where all experiences occur from a convenience desire. Accordingly, device use might show high tablet –convenience device- use. In line with this, it is expected that this segment is a relatively old segment.

3. Conceptual Framework

This chapter visualizes and explains the two conceptual frameworks of research question 1a and 1b. Additionally, this chapter describes the indicators and the active covariates used to identify customer segments along the experiences in the digital customer journey. Lastly, the passive covariates are described, used to profile the segments in order to get a deeper knowledge of the segment-specific characteristics.

3.1 Framework

Given the nature of the research, no hypothesis will be included in neither of the two frameworks. Namely, Latent Class Modelling is a post-hoc analysis and

(44)

35

Therefore, as a general rule based on other leading segmentation studies in this field, no hypotheses are proposed (Konus, Verhoef and Neslin, 2008).

Figure 3.1 provides an overview of the conceptual framework of the first analysis of this research, where the overall purpose is to explore ifwe can talk about distinct customer segments identified on the basis of customers’ behaviour along multiple touch points on online platforms (research question 1a). The method used to address this question is Latent Class Cluster Analysis. Subchapter 4.3.1 provides a detailed description of the nature of this method.

Figure 3.2 provides an overview of the conceptual framework of the second and last analysis of this research, where the overall purpose is to explore if there are distinct customer segments which can be identified on their profitability that (may) depend on different behavioural and relational drivers. Here, drivers of profitability are identified using the method Latent Class Regression (LCR), for which subchapter 4.4.1 provides a description. The Dependent Variable used is TotalMoneySpent, as this variable is a proper indication for profitability. On the basis of this LCR, profitability drivers, if existent, can be identified for each specific segment.

(45)

36

(46)

37

(47)

38

The dataset used for this research exists of three different types of variables: i) Indicators for LCCA and Predictors for LCR, ii) Active covariates and iii) Passive covariates. Indicator variables and active covariates are used to segment the online customers. In addition, passive covariates are included in the framework to adequately profile the discovered segments. Profiling is done in order to increase the explanation power of the segments and thereby increase the value with regards to the managerial implications.

3.2 Indicators/ Predictors

Firstly, although having the same function, it is noted that all indicators in LCCA, will be used as predictors in LCR. One exception is the variable Total Money Spent, which is an indicator in LCCA but a dependent variable in LCR.

All indicators used in LCCA and all predictors used in LCR, with the exception of the one-dimensional variables Customer Care Encounters, Orders and Total Money Spent are three-dimensional. They consists of i) Online Experiences ii) Device and iii) Platform. The three mentioned variables which are one-dimensional consist of only Online Experiences, as showed in figure 3.1.

This research introduces this broad and in-depth touch point dataset in order to gain insight into the specific experience and characteristics of touch points. Namely, a customer can have an Online experience ‘visit’. This visit can take place via different devices, for instance, via a tablet. However, using a tablet, a customer can enter the company website through an internet browser, but a customer can also visit the company via the company application when a customer has downloaded this

application. The same applies for a phone device. In the next subchapter an overview is given of the three discussed dimensions concerning the indicator variables.

(48)

39

3.2.1 Online Experiences

The indicator/predictor dimension Online Experiences used in this research consists of the following three-dimensional indicators/predictors.

1) Visits 2) Visit time 3) Email clicks 4) Pageviews

Additionally, the one-dimensional indicators/predictors within the dimension Online Experiences are the following.

5) Customer Care Encounters 6) Orders

7) Total Money Spent*

* In LCR, the indicator Total Money Spent used in LCCA is used as a dependent variable.

3.2.2 Device

The indicator/predictor dimension Device used in this research consists of the

following. There are three different devices included in this research on which a touch point can take place.

1) Desktop (existing of a laptop or a computer) 2) Tablet

(49)

40

3.2.3 Platform

The indicator/predictor dimension Platform used in this research consists of the following. There are two different platforms included in this research on which a touch point can take place.

1) Website 2) Application

3.3 Covariates

3.3.1 Active covariates

Active covariates are, next to indicator variables, used in this research to segment consumers; indicate which customer is assigned to which segment. Moreover, all active covariates are three-dimensional, consisting next to Online Experience of a Device dimension (3.2.2) and a Platform dimension (3.2.3). The online experience dimension of the covariate variables is:

1) Transactions

Transactions in this dataset are different than orders. Instead, transactions indicate a conversion before the step of actually purchasing the product, in this case, they indicate a bid as the model used by this company is an auction model.

3.3.2 Passive covariates

Passive covariates are used in this research to profile the obtained segments in order to increase the explanation power of the segments. The passive covariates used in this research are the following.

(50)

University of Amsterdam 41 2) Gender 3) Registration date 4) Email opt-in 5) Email opt-out 6) Net Promoter Score

4. Methodology

4.1 The sample

4.1.1 Description of sample

In order to identify segments based on digital customer journey touch points, a sample has been derived from the18-plus Dutch online population. Because no sampling frame could be selected from the enormous population of online customers, the sample selection method that has been used is a non-probability (quota) sampling technique distributed via the consumer base of an anonymous Dutch e-commerce company. This way it is still possible to generalize from the findings, only not on statistical grounds (Saunders, Lewis and Thornhill, 2009). A vast customer base of a Dutch e-commerce company is used, which exists of approximately 500.000 active consumers. From this base, a sample of 5000 customers is randomly derived.The company providing the dataset is a company active in the online leisure industry, making use of an auction model. The company is chosen on the basis of satisfying the following criteria; sufficient data tracking, sufficient data availability, minimum active consumer base and accessibility.

(51)

42

4.1.2 Sample selection

The sample selection is a random selection of 10% of the active customer database existing of +/-500.000 customers. Within this selection a selection of 1%is made, consisting of 5000 active customers, still keep the reliability of the sample to a sufficient extend. This vast sample amount is chosen to increase the

representativeness regarding the population and the reliability of the sample. In addition to this, a relatively large initial sample is needed to secure reliability on profiling the segments.Moreover, an active customer is a customer that did at least one order in the chosen timeframe of which the dataset consists; 01-09-2015 till 01-9-2016. In this timeframe, all occurred touch points –which are included in this

research- within the customer journeys of the 5000 customers in the sample are included in the dataset. This timeframe is chosen in order to control for seasonal influences in the research.

4.2 Research design

In this research, data is retrieved from an anonymous Dutch e-commerce company in the leisure industry. Firstly, the research proposal of this research has been discussed with the Marketing Manager of the company. After accordance of conducting the research using data of the company, optional company-specific touch points are analysed and included into a proposed variables list. This list of optional touch points is discussed with the companies’ Business Intelligence Analysts, who consulted on the achievability of the proposed variable list. Accordingly, from this final list of touch points, final indicator/predictor variables (23) and final active covariates (5) are indicated in order to base the segmentation on. Additionally, final passive covariates

(52)

43

(6) are indicated in order to profile the identified segments. The source of this data are the internal databases of the company; Datawarehouse and Qlikview. The dataset is established among the common variable customer ID, which is a specific number in order to identify specific customers and their specific touch points/ behaviours in the past. Among the first column of 5000 customer id’s, the dataset exists of additional columns; one column existing of 5000 data observations for every of the 34 touch points. The final dataset is obtained with guidance of two of the companies’ Business Intelligence analysts.

Moreover, due to time limits, no variables are included in the dataset

concerning the order in which the touch points are experienced. As sequence in touch points defines the customer journey, this research presents touch points of the digital customer journey as a segmentation base instead of the digital journey itself. This is addressed in more detail in subchapter 6.5, Limitations and Future Research.

Nonetheless, touch points along the customer journey as presented in this research contains valuable behavioral information, providing profound segments when used as a segmentation base.

4.3 Customer Segments Based on Online Touch Points

4.3.1 Method: Latent Class Cluster Analysis

The method used to answer research question 1a is Latent Class Cluster Analysis. In segmenting, Latent Class analysis holds many advantages and has been shown to perform well for clustering applications (Magidson and Vermunt, 2002). Nevertheless it is a less used technique due to relative complexity in operationalizing the

algorithms and low availability (Green, Carmone and Wachspress, 1976). However, there is a current development of extended computer algorithms, which allow today's

Segmenting customers based on touch points along their customer journey on digital platforms