O M N I C H A N N E L S H O P P I N G B E H A V I O U R D U R I N G T H E C U S T O M E R J O U R N E Y

Transcription

1 O M N I C H A N N E L S H O P P I N G B E H A V I O U R D U R I N G T H E C U S T O M E R J O U R N E Y Graduation paper Lieke van Delft October 2013

2 O M N I C H A N N E L S H O P P I N G B E H A V I O U R D U R I N G T H E C U S T O M E R J O U R N E Y An empirical study into the contribution of omni channel shopping characteristics during the customer journey by consumers segments. Eindhoven, October 2013 AUTHOR W.C.J.M. (Lieke) van Delft Student number : liekevandelft@outlook.com Real Estate Management & Development Faculty of Architecture, Building & Planning Eindhoven University of Technology COMPANY Wereldhave N.V. COMMITTEE Dr. ir. A.D.A.M. Kemperman Ir. A.W.J. Borgers Lizzie Saffrie Ir. G.J. Wijnen TU/e TU/e Wereldhave Wereldhave 2

3 CONTENTS Summary English... 5 Samenvatting Nederlands... 7 Preface Introduction Background and relevance Omni channel The Customer Journey The challenge of retail Segmentation on channels Scope Research questions Objectives Research questions Research structure Consumer shopping behaviour Consumer behaviour Online consumer behaviour Omni channel shopping behaviour The Customer Journey Conclusion Evaluation of consumer segmentation Consumer segmentation Segmentation methods Cluster analyses in general Clustering variables Clustering procedure Number of clusters Cluster solution interpretation and validation Segmentation studies Conclusion Research design and implementation Research instrument Which data do we need? Online and offline shopping behaviour Measuring the customer journey Omni channel shopping Personal characteristics Data collection Shopping centers Purmerend, Eggert Shopping Center Arnhem, Kronenburg Shopping Center Maassluis, Koningshoek Shopping Center Capelle aan de Ijssel, Koperwiek Shopping Center Roosendaal, Roselaar Shopping Center Conclusion Data Analysis & Results Data Analysis Differences between consumers shopping behaviour Predicting shopping behaviour from variables Characteristics of respondents Consumers online and offline Shopping behaviour Personal characteristics & online buying behaviour Psychographics & online buying behaviour

4 5.3.3 Description of shopping behaviour Conclusion Shopping behaviour during the customer journey Personal characteristics & the customer journey Psychographics & the customer journey Description of the customer journey Conclusion Omni channel shopping behaviour Personal characteristics & omni channel shopping behaviour Description of omni channel shopping behaviour Conclusion Deriving a typology of omni channel shoppers Cluster analyses Description of Fashion shopper typologies Description of Personal care shopper types Description of Grocery shopper types Comparison of shopper types Conclusions Conclusion and recommendations Conclusions Theoretical research question Practical research question Recommendations for further research Managerial implications References

5 SUMMARY ENGLISH In recent years, there has been an increasing interest in retail channel usage of consumers. Retailers experiences tough times. On the one hand, due to three crises (the financial crisis, the euro crisis and the recession) consumers have been confronted with bad economic conditions. This is resulting in a decrease of visitor numbers in Dutch shopping centers and a decrease in their spending. On the other hand, consumers use lots of online and offline channels during their shopping process. Consumers can shop everywhere and at any time. Therefore, there is a need to understand the online shopping behaviour of consumers. Omni channel shopping behaviour is a trending topic among retailers and managers involved in retail. However, no single study exists which adequately covers omni channel shopping behaviour. Within omni channel shopping all channels work together within one strategy in which the customer is the central point. Furthermore, consumers switch easily and continuously between the channels and they experience all channels together as one complete channel. Omni channel shopping behaviour is related to the customer journey. The customer journey can considered as consumers decision making process. It is the journey a customer follows for purchasing products and services. By reviewing literature, insight was gained in omni channel shopping behaviour and the customer journey. Thereby, consumer segmentation is studied to gain insight in clustering techniques for analysing omni channel shopping behaviour during the customer journey. With information obtained from the literature review, omni channel shopping behaviour and channel usage of consumers have been operationalized into measurable variables. A questionnaire was designed especially for this research. The scope of this research is limited to consumers that live in the catchment areas of five shopping centers located in Purmerend, Arnhem, Roosendaal, Capelle aan den Ijssel and Maassluis. Data is collected among 2025 Dutch consumers through an online questionnaire. The objective of this research is to gain insight in consumers omni channel shopping behaviour during the customer journey. The main questions addressed are as follows: Theoretical research question: What is the relation between omni channel shopping behaviour during the customer journey and personal characteristics? Practical research question: What consumer segments can be found for consumers who live in large and medium-sized cities, taking into account consumers omni channel shopping behaviour during their customer journey? Omni channel shopping behaviour is studied during the customer journey. The customer journey for this research is considered as consumers shopping decision making process. The customer journey consists of the following five phases: Stimulate Search for information Purchase Delivery After sales service In every phase of the customer journey consumers can choose from a range of channels. Nowadays, consumers are well informed because they can gather information through a wide range of channels. Consequently, consumers are very critical in making their purchase decisions. Prior studies have noted that demographics, channel knowledge, perceived channel utility and shopping orientation are related to online shopping behaviour. Whether consumers choose for online channels depends on their channel risk perceptions, pricesearch intentions, search effort and delivery time. However, consumers choose to shop in a brick and mortar shop because of the service and product experience. 5

6 This study confirms previous findings and contributes additional evidence that suggests relations between consumers omni channel shopping behaviour during the customer journey and personal characteristics (sociodemographics and psychographics). Interesting is the relation between price consciousness and buying groceries online. The results show that non-price conscious consumers are likely to buy groceries online. A possible explanation for this is that budget supermarkets do not offer the opportunity to buy groceries online. As stated before consumers online channel usage is related to age and innovativity. As found in the literature channel usage is related to channel knowledge, young and innovative consumers are likely to have knowledge about online channels. Surprisingly, especially consumers between 30 and 49 years old use applications on mobile devices more often than consumers below 30 years old. Consumers below 30 years old are more likely to use social media. Further, consumers with high incomes are more likely to use applications on mobile devices than consumers with low incomes. Retailers should consider the type of consumers they reach through their applications and social media channels. Because it is important for retailers to offer the right information, on the right channel, during the customer journey. Interesting is to note that using advertisement leaflets and catalogues is related to education and age. Low educated consumers are likely to search for information about personal care products and buy fashion through advertisement leaflets and catalogues. Surprisingly, these consumers are also likely to use advertisement leaflets and catalogues while shopping in a traditional shop. Further, young consumers are unlikely to search through advertisement leaflets to buy groceries. Through online sales the number of pick-up points increase. The results of this study show important findings for collecting products at a pick-up point. Pick-up points are most often used by young consumers (below 50 years old) and low educated consumers are unlikely to use pick-up points. Further, interesting is the fact that price un-conscious consumers are very likely to use pick-up points to collect their online bought groceries. Prior studies have noted that segmenting consumers based on their shopping behaviour can lead to interesting shopper typologies. Within this research, interesting shopper typologies based on consumers shopping behaviour were found by using the TwoStep clustering technique. Six fashion shopper typologies were found (omni channel shoppers, omni channel adopters, online searchers, targeted offline purchasers, offline single channel shoppers and laggards), three grocery shopper typologies were found (omni channel grocery shoppers, offline targeted grocery purchasers and offline grazers), and two personal care shopper typologies were found (omni channel personal care shoppers and totally offline personal care shoppers). This suggests structural differences in the buying behaviour across product categories. However, interesting is the fact that for every product category a type of omni channel shopper was found. These omni channel shopper types show similarities in psychographics and socio-demographic characteristics. These omni channel shopper types are often below 50 years old, have high incomes and are high educated. Further omni channel shoppers are innovative, loyal, they enjoy shopping, feel pressured in time and they are price conscious. This research offers a good basis for studying omni channel shopping behaviour during the customer journey. It was a big challenge to research omni channel shopping behaviour and the customer journey, because there was no research conducted yet on these topics. From this research we have learned which channels consumers use and how often they use those channels. For further research it would be interesting to gain insight in consumers spending via those channels. Nowadays, retailers are implementing online channels in brick and mortar shops. It would be interesting to investigate the role and impact of online channels in brick and mortar shops. In addition, there is a generation gap for using online channels while shopping. In an omni channel strategy it is therefore important to take elderly consumer into account. Elderly consumers are often forgotten in online strategies, but they represent a significant amount of consumers and thus a significant part of the sales. The number of consumers that use online channels for shopping will probably increase in the upcoming years, because the number of internet users still increases. It is therefore important to conduct this research again over five years. 6

7 SAMENVATTING NEDERLANDS De laatste jaren, is er een groeiende interesse in onderzoek naar het kanaalgebruik van consumenten. Retailers hebben het zwaar. Aan de ene kant, worden consumenten geconfronteerd met slechte economische omstandigheden vanwege drie crisissen (de financiële crisis, de euro crisis en de recessie). Dit resulteert in een afname van bezoekers in Nederlandse winkelcentra en een afname in consumentenbestedingen. Aan de andere kant zijn keuze mogelijkheden voor consumenten toegenomen door de groei van online en offline winkel kanalen. Consumenten kunnen overal winkelen en op ieder moment van de dag. Hierdoor is het noodzakelijk om het online winkelgedrag van consumenten inzichtelijk te maken. Omni channel winkelen is een trending topic onder retailers en managers in retail. Maar het omni channel winkelgedrag van consumenten is nog niet eerder wetenschappelijk onderzocht. Bij omni channel winkelen werken alle kanalen samen volgens één strategie waarin de consument het middelpunt is. Daarnaast wisselen consumenten eenvoudig tussen kanalen en zij ervaren alle kanalen samen als een compleet kanaal. Omni channel winkelgedrag is gerelateerd aan de customer journey. De customer journey kan worden gezien als het consumenten-beslissingsproces. Het is de reis die een consument aflegt om een product of service te kopen. Door literatuuronderzoek is inzicht verkregen in het omni channel winkelgedrag en de customer journey. Daarnaast is consumentensegmentatie bestudeerd om inzicht te krijgen in clustertechnieken voor het analyseren van omni channel winkeltypen. Met informatie die is verkregen vanuit de literatuurstudie, is het omni channel winkelgedrag en het kanaalgebruik van consumenten geoperationaliseerd in meetbare variabelen. Speciaal voor dit onderzoek is een enquête ontwikkeld. Dit onderzoek is ingekaderd tot consumenten woonachtig in de verzorgingsgebieden van vijf winkelcentra in Purmerend, Arnhem, Roosendaal, Capelle aan den IJssel en Maassluis. Data van 2025 Nederlandse consumenten is verzameld door middel van een online enquête. De doelstelling van dit onderzoek is inzicht te verkrijgen in het omni channel winkelgedrag van consumenten tijdens de customer journey. De hoofdvragen zijn als volgt geformuleerd. Theoretische onderzoeksvraag: Wat is de relatie tussen omni channel winkelgedrag gedurende de customer journey en persoonlijke kenmerken? Praktische onderzoeksvraag: Welke consumentensegmenten kunnen worden gevonden voor consumenten die wonen in grote en middelgrote steden, rekening houdend met het omni channel winkelgedrag van consumenten gedurende hun customer journey? Omni channel winkelgedrag is bestudeerd gedurende de customer journey. De customer journey voor dit onderzoek is beschouwd als het consumenten beslissingsproces. De customer journey bestaat uit de volgende vijf fases: stimulatie zoeken naar informatie kopen bezorging service In iedere fase van de customer journey kiezen consumenten uit een aantal kanalen. In deze tijd zijn consumenten goed geïnformeerd omdat ze overal informatie over producten kunnen vinden. Hierdoor zijn consumenten kritisch in hun aankoop beslissingen. Voorgaande studies hebben aangetoond dat demografische kenmerken, kanaalkennis, verkregen voordelen uit een kanaal en winkel oriëntatie zijn gerelateerd aan online winkelgedrag. De keuze van consumenten voor online kanalen hangt af van kanaal-risico-percepties, prijszoekintenties, zoek doelstellingen en de levertijd van een product of service. Consumenten winkelen graag in traditionele winkels vanwege de service en product ervaring. 7

8 Dit onderzoek bevestigd eerdere bevindingen en geeft aanvullend bewijs dat er relaties zijn tussen omni channel winkelgedrag tijdens de customer journey en persoonlijke kenmerken (sociaaldemografische en psychografische). De resultaten laten zien dat niet-prijsbewuste consumenten vaker dagelijkse boodschappen online kopen dan prijsbewuste consumenten. Waarschijnlijk heeft dat te maken met het feit dat budget supermarkten (nog) geen dagelijkse boodschappen online verkopen. Daarnaast is in de literatuur gevonden dat kanaal-gebruik gerelateerd is aan kanaal-kennis. In dit onderzoek komt naar voren dat met name jonge en innovatieve consumenten gebruik maken van online kanalen. Waarschijnlijk hebben zij de meeste kennis over online kanalen. Verassend genoeg gebruiken consumenten tussen 30 en 49 jaar vaker applicaties op mobiele apparaten dan consumenten met een leeftijd onder 30 jaar. Consumenten jonger dan 30 jaar gebruiken juist vaker sociale media. Verder maken consumenten met hoge inkomens vaker gebruik van applicaties om te winkelen dan consumenten met lage inkomens. Het is voor retailers daarom ontzettend belangrijk dat zij weten welke typen consumenten zij bereiken met applicaties, websites en sociale media. Dit is belangrijk omdat retailers dan de juiste informatie kunnen aanbieden aan consumenten, via het juiste kanaal, gedurende de juiste fase van de customer journey. Het is interessant dat het gebruik van reclame folders en catalogi gerelateerd is aan opleidingsniveau en leeftijd. Met name laag opgeleide consumenten gebruiken reclamefolders en catalogi om te zoeken naar informatie over persoonlijke verzorgingsproducten en voor de aankoop van mode. Daarnaast gebruiken zij vaker reclame folders en catalogi tijdens het winkelen in een traditionele winkel. Daarbij is het onwaarschijnlijk dat jonge consumenten gebruik maken van reclamefolders en catalogi om dagelijkse boodschappen te kopen. Door de toename van het aantal online aankopen neemt het aantal pick-up points in Nederland toe. De resultaten van dit onderzoek tonen aan dat pick-up points voornamelijk worden gebruikt door jonge consumenten (onder de 50 jaar oud). Het is onwaarschijnlijk dat laagopgeleide consumenten pick-up points gebruiken. Daarnaast is het interessant dat niet prijsbewuste consumenten vaker gebruik maken van pick-up points om dagelijkse boodschappen op te halen dan prijsbewuste consumenten. Voorgaande studies hebben aangetoond dat het segmenteren van consumenten op basis van hun winkelgedrag kan leiden tot interessante (doel)groepen. In dit onderzoek zijn consumentensegmenten gevonden gebaseerd op het omni channel winkelgedrag door gebruik te maken van de TwoStep cluster methode. Er zijn zes segmenten gevonden voor de aankoop van mode (omni channel shoppers, omni channel adopters, online searchers, targeted offline purchasers, offline single channel shoppers and laggards), drie voor dagelijkse boodschappen (omni channel grocery shoppers, offline targeted grocery purchasers and offline grazers) en twee voor persoonlijke verzorging (omni channel personal care shoppers and totally offline personal care shoppers). Dit geeft aan dat er structurele verschillen zijn in koopgedrag bij de verschillende product categorieën. Opvallend is het feit dat er voor iedere product categorie een omni channel consumentensegment is gevonden. Deze omni channel consumenten hebben gelijkenissen in psychografische en sociaaldemografische kenmerken. Omni channel consumenten zijn vaak jonger dan 50 jaar, hebben hoge inkomens en zijn hoog opgeleid. Daarnaast zijn omni channel consumenten innovatief, loyaal, vinden ze het leuk om te winkelen, ervaren een tijdsdruk en zijn ze prijsbewust. Dit onderzoek biedt een goede basis voor het bestuderen van omni channel winkelgedrag tijdens de customer journey. Het was een grote uitdaging om het omni channel winkelgedrag en de customer journey van consumenten te onderzoeken, omdat er nog geen wetenschappelijk onderzoek naar deze onderwerpen was gedaan. In dit onderzoek is inzicht verkregen in welke kanalen consumenten gebruiken en hoe vaak ze deze kanalen gebruiken. Voor verder onderzoek is het interessant om inzicht te krijgen in het bestedingspatroon van consumenten via de verschillende kanalen. Daarnaast is het interessant om de rol en impact van online kanalen ter versterking van de fysieke winkel te onderzoeken. Er is een generatiekloof zichtbaar in het gebruik van online kanalen tijdens het winkelen. Bij het ontwerpen van een online strategie is het daarom van belang om ook rekening te houden met oudere consumenten. Ouderen worden namelijk vaak vergeten bij de ontwikkeling van online strategieën, maar zij vertegenwoordigen een significant deel van de consumenten en dus een significant deel van de bestedingen. Het aantal consumenten dat online kanalen gebruikt om te winkelen, zal waarschijnlijk in de komende jaren toenemen. Dit is waarschijnlijk omdat het aantal internet gebruikers ook toeneemt. Het is daarom interessant om dit onderzoek over vijf of tien jaar opnieuw uit te voeren. 8

9 PREFACE Proudly I present the result of my graduation research project. This project is part of the Master track Real Estate Management & Development at the Eindhoven University of Technology and was completed during an internship period at Wereldhave. The subject of this report, omni channel shopping behaviour of Dutch consumers during their customer journey, turned out to be a challenging topic. Shopping behaviour in general is well researched. In addition, omni channel shopping and the customer journey are popular concepts frequently used by retailers, but there was little scientific research conducted about these concepts. I managed to translate these popular concepts into measurable variables. And I was able to find very interesting results. Thanks to this research I found a new love within Real Estate called Retail. This love will probably chase me the rest of my life. However, this research was made possible by a number of individuals whom I am extremely grateful. My word of thanks goes out to Aloys Borgers, Astrid Kemperman, Geri Wijnen and Lizzie Saffrie for their good directions and guidance during this research. I would also like to thank Wereldhave for the possibilities they offered me, and for the nice time I have had there. My special thanks goes out to my family, friends and my boyfriend Mike, for their understanding, support and the necessary distraction. Without them I had not nearly so much fun during this research. Please enjoy reading my master thesis, Lieke van Delft Haarsteeg, September

10 In the beginning, there were shops Places we visited to have a look round, chat with an assistant, and buy stuff from the rows and racks of stock vying for our attention. We knew what we needed and a shop was the place to get it. Life was simple Fitch,

11 1 INTRODUCTION This chapter introduces the research which is conducted on omni channel shopping behaviour during the customer journey of Dutch consumers. This chapter is organised in the following way. In the first section, the background and relevance of this research are discussed. Omni channel shopping behaviour and the customer journey are relatively new concepts; these concepts are explained in section and section The next section (1.2) gives insight in the scope of the research. With these insights the research questions are defined (section 1.3) and the research design is developed (section 1.4). 1.1 BACKGROUND AND RELEVANCE In recent years, the number of purchase and orientation channels has increased. This creates widespread challenges for researchers and for practitioners involved in retailing. Because of these challenges, there is a need for studies about multiple channel customer management with the objective to increase customer value. This customer value can be increased through proper customer acquisition, retention, and development (Neslin, Grewal, Leghorn, Shankar, Teerling, Thomas & Verhoef (2006). However, it is a challenge to understand the behaviour of consumers in an environment where consumers can shop anytime and anywhere. Thereby, it is important to understand the characteristics of consumer segments, in order to target and position online and offline channels well, to be able to reach new customers and serve current customers (Konuş, Verhoef & Neslin, 2008) OMNI CHANNEL A few years ago, omni channel shopping behaviour was something no one knew about. Recently, the popularity of the concept omni channel shopping has increased, but little research is conducted (yet) on omni channel shopping behaviour. However, research is conducted about multi channel and cross channel shopping behaviour (Konuş et al., 2008; Neslin et al., 2006; Verhoef, Neslin, Vroomen, 2007). Multi channel shopping is shopping on different channels, for instance in the brick and mortar shop and online on a website. Within multi channel shopping every channel has its own strategy. In cross channel shopping there is one strategy for all channels and consumers use different channels. Omni channel shopping is seen as an advanced form of cross channel shopping. Consumers use several channels during the buying process, both online and offline: channels such as websites, webshops, social media, brick and mortar shops, applications on mobile devices, catalogues, and many more. In addition, consumers switch easily and continuously between these channels and they experience all channels together as one complete channel. For instance, when consumers are going to shop for groceries at Albert Heijn they can use the Appie application on their smartphone; this application can show consumers shopping list and a personalized shopping route. Consumers experience this application and the shop as one complete channel. This seamless shopping experience is called omni channel shopping. Consumers consider the pros and cons of each channel before using them. The choice behaviour of consumers regarding channels depends on characteristics of consumers, products, shopping channels and the retailer (Zijlmans, 2010). Based on their wants and needs at a moment, the consumer decides which channel to use to search for information or buy products. For consumers, online shopping has, among others, the advantage that it saves both time and effort. In addition, information can be easy found and compared online (Kollmann, Kuckertz & Kayser, 2012). Recent, new media have emerged, resulting in a closer connection between retailer and consumer. In the Netherlands, social media have integrated in daily life; almost eighty percent of the Dutch uses social media (Newcom research, 2013). Retailers use of social media has also increased. Research of EtailTrends has shown that social media use by retail formulas increased extremely last year, the number of fans and followers increased sevenfold in one year. Social media can be defined as media for social interaction, using highly accessible and scalable communication techniques (Markova and Petkovska-Mirčevska (2013)). Social media is, among others, accessible by personal computer, tablet and smartphone. In the Netherlands, the most used social media channel is Facebook. 11

12 Tablets and smartphones are very popular. In June 2012, twenty-three percent of the Dutch population had a tablet and forty-eight percent had a smartphone. Smartphones are most popular among years old consumers (71% has a smartphone) and years old consumers (61% has a smartphone). Tablet owners are generally a bit older, which is presumably due to the price of tablets. Tablets are most used in the age groups between 35 and 49 years old (27% has a tablet), between 50 and 64 years old (24%) and between 18 and 34 years old (22%) (GfK, 2013). New media and mobile devices have created a new type of shopper. The new shopper has a strong functional and emotional bond with his smartphone. The new shopper is always online; has fast access to information; compares and evaluates prices actively; uses several channels, online as well as offline to search, shop and buy; shares his product experience through social media, and at last the new shopper has high expectations for meeting his needs and wants, anytime and anywhere. Nowadays, consumers have many opportunities to orientate, gain information and buy products. Many retailers have expanded their shopping channels, offline and online, to better serve their customers and to increase sales (Benedicktus, Brady & Dark, 2008; Verhoef et al., 2007). This is necessary because, cross channel shoppers are more profitable and more loyal than single channel shoppers according to the Cross Channel Integration study of Accenture (2010). More and more existing retailers are developing a strategy for serving shoppers across all channels. They want to improve the relation with their customers by expanding the touch points with customers THE CUSTOMER JOURNEY In recent years, there has been an increasing amount of literature on online shopping behaviour (Javadi, Dolatabadi, Nourbakhsh, Poursaeedi & Asadollahi, 2012; Verhoef et al., 2007; Kim & Eastin, 2011; Jepsen, 2007). The popularity of online shopping is still increasing because of the convenience of the internet, the possibilities to compare price and quality, and the increased amount of information. Online shopping behaviour has changed consumers decision making process. Consumers decision making process in this research is called the customer journey. This is the journey a customer follows to purchase products and services, starting from need recognition to the end of the purchase and even after sales service (Engel, Blackwell & Miniard, 1995). Consumers can experience the customer journey both psychologically and physically. Consumers use of online channels during their customer journey is increased, but online channels are not the only Holy Grail. Not all goods or services are suitable for online sales. Goods which need to be experienced personally are more suited for brick and mortar shops. Consumers online purchasing is different from brick and mortar shopping (Hogg & Penz, 2008). Consumers like brick and mortar shops because of the importance of touch, feel, and service in their shopping experience. On the other hand, online shopping gives consumers an anywhere anytime convenience. For retailers it is important to offer customers the right mix of information sources during the different phases in the customer journey. With insight in the customer journey and omni-channel shopping behaviour of consumers, retailers can better serve their customers. Consequently, they can beat their competition and gain extra market share THE CHALLENGE OF RETAIL At the moment, most Dutch retailers experience tough times. They suffer from the economic uncertainty in the Netherlands. Consumers have less confidence in the economy; as a result they are more conscious about their purchases. In February 2013, consumer confidence dropped to a historic low, at -44 points (Figure 1) (CBS, 2013). In addition to the dropped consumer confidence, the retail market is saturated because of the continuous expansion in the last decades. This increased the competition between retailers (NRW, 2011). Besides, turnovers of offline retailers decreased because consumers buy products more often online. 12

13 FIGURE 1, CONSUMER CONFIDENCE NETHERLANDS, SEASONS CORRECTION (CBS, 2013) For retailers, a way to survive is by combining the best of both worlds. The Dutch Shopping Center Council (NRW) calls this The new way of shopping ; retailers want to assemble the best of internet shops and brick and mortar shops (Hoofdbedrijfschap Detailhandel, 2011). According to the new way of shopping, internet shopping is not a threat but a chance for retailers. Therefore, retailers should offer consumers quality, service and experience offline as well as online. This new way of shopping can be conducted, among others, in shopping centers. The inner-city of Veenendaal has already implemented this concept; consumers can shop with their smartphone and pay online for parking (Jansen, 2012). Several shopping centers in the Netherlands have a Facebook page, and some have created an interactive shopping center app to guide customers to and through the shopping centers, which is another example of the new way of shopping (Houwelingen, 2013). In July 2013, IKEA launched a new app, which allows users to try 90 products at home. This app makes it possible to create an augmented reality image of the furnishings in your own home (IKEA, 2013) SEGMENTATION ON CHANNELS Previously consumers came automatically to the shops, but those days are over. Nowadays, consumers need to be attracted to shops. Hence, information about consumers is needed to attract new consumers, and to increase involvement and loyalty. Eventually, this should result in more profit for the retailer. A method to study consumers and their buying behaviour is consumer segmentation (Gilboa, 2009). Consumer segmentation is the process of dividing a heterogeneous market into groups of customers which have nearly the same characteristics, wishes, needs, buying habits or reactions on marketing activities. Consumer segmentation assumes that consumers exhibit heterogeneity in product preferences and buying habits (Dibb, 1998). Through segmentation, retailers can determine the most interesting customer groups to focus on. Current studies on consumer segmentation, do not consider omni channel shopping behaviour and the customer journey. While insight in omni channel shopping behaviour during the customer journey can help retailers and investors in retail real estate by developing a strategy to beat the competition. 13

14 1.2 SCOPE The relevance of this research on omni channel shopping behaviour during the customer journey of Dutch consumers became clear in the previous section. Another practical relevance for this study follows from a question posed by Wereldhave, an investor in retail real estate. Wereldhave wants to gain insight in the shopping behaviour of consumers that live in the catchment area of their shopping centers in Arnhem, Capelle aan den Ijssel, Etten-Leur, Maassluis, Leiderdorp, Purmerend and Roosendaal (information about these shopping centers can be found in section 4.3). These shopping centers are situated in large and medium sized cities. Therefore, the scope of this study is omni channel shopping behaviour during the customer journey of consumers who live in large and medium-sized cities. 1.3 RESEARCH QUESTIONS OBJECTIVES The objectives for this research were derived from the description of the background, relevance and the scope of this research. In summary, there is a need for research that gives insight in omni channel shopping behaviour of Dutch consumers during the customer journey. This shopping behaviour can be analysed through segmentation; therefore, consumer segments need to be developed that incorporate aspects such as, the customer journey and omni channel shopping behaviour. Existing literature does not provide this information. Because, for this study it is important to gain insight in shopping behaviour, the main objectives of this research are as follows: 1. Gain insight in consumers online and offline shopping behaviour. 2. Gain insight in consumers omni channel shopping behaviour during the customer journey. 3. Gain insight in consumer segments which give information about omni channel shopping during the customer journey RESEARCH QUESTIONS For this research, the following research questions (one theoretical and one practical) are derived from the objectives. Theoretical research question: What is the relation between omni channel shopping behaviour during the customer journey and personal characteristics? Practical research question: What consumer segments can be found for consumers who live in large and medium-sized cities, taking into account consumers omni channel shopping behaviour during their customer journey? To be able to answer these research questions, the following sub-questions must be answered: 1. What is omni channel shopping behaviour during the customer journey? 2. How can consumer segmentation be used for analysing omni channel shopping behaviour during the customer journey? 3. How can omni channel consumer behaviour be investigated during the customer journey? Shopping behaviour of consumers will be studied to form shopper typologies based on consumers omni channel shopping behaviour during the customer journey. 14

15 1.4 RESEARCH STRUCTURE This research is a quantitative study. This research focuses on omni channel shopping behaviour during the customer journey of consumers. The research took place between November 2012 and September The structure of this research is displayed in Figure 2. The research is divided in six phases, which are documented in six chapters in this report. During the initial phase of this study, a literature research is conducted on shopping behaviour, channel usage, the customer journey and consumer segmentation. Chapter 2 begins by laying out the theoretical dimensions of the research, and look at omni channel shopping behaviour and the customer journey, answering sub question 1. Chapter three explores and discusses consumer segmentation methods in order to answer sub question 2. Literature research on segmentation methods is conducted to gain more insight in segmentation techniques for developing shopper typologies. Within the second phase of this research data is collected through an online questionnaire among Dutch consumers who live in large and medium-sized cities. The questionnaire is designed with the input from the literature research. The research design and implementation are discussed in chapter 4. The third phase of the study shows the interpretation of the results and the formulation of conclusions and managerial implications. Analyses and results are shown in chapter 5. In addition, segments are formed and described regarding omni channel shopping behaviour during the customer journey. Conclusions and managerial implications can be found in chapter 6. Phase 1: Initial phase Literature research on shopping behaviour (chapter 2) Literature research on segmentation (chapter 3) Phase 2: Research design and implementation (chapter 4) Phase 3: Results & Conclusions Data analysis and results (chapter 5) Conclusion and managerial implications (chapter 6) FIGURE 2, RESEARCH STRUCTURE 15

16 16 Source: Google

17 2 CONSUMER SHOPPING BEHAVIOUR This chapter will focus on consumer shopping behaviour in an omni channel shopping environment during the customer journey. Sub question 1 is the central question in this chapter: What is omni channel shopping behaviour during the customer journey? Before proceeding to answer the central question, it will be necessary to explain consumer behaviour in general (section 2.1). What follows is an outline of omni channel shopping behaviour. The third section gives insight in the buying process of the consumer, which is also defined as the customer journey. Finally, conclusions are drawn about the most important characteristics of consumers shopping behaviour related to the research question. 2.1 CONSUMER BEHAVIOUR Consumer behaviour are all consumers activities directly related to the acquisition, use and disposal of products. Including the information and decision-making processes that precede and follow from these activities. Consumer behaviour has four main aspects; communication behaviour, purchase behaviour, usage behaviour and disposal behaviour. This research has its main focus on the communication and purchase behaviour. It is complicated to make the whole decision making process of consumers transparent. Often, only the buy decision is visible. Many factors influence the consumer during their decision making process; such as social, socio-cultural and external non-controllable stimuli. The response of the consumer is the result of those stimuli (Broekhuizen, 2011). There are different types of buying behaviour with similarities in function and motivation. Research distinguishes two forms of shopping; utilitarian and hedonic shopping motivations. Utilitarian shopping is task-oriented shopping, while hedonic shopping is shopping for personal gratification and self-expression (Babin, Darden & Griffin, 1994; Wijnen, 2010). However, the basics of shopping are simple, consumers like to have a choice in merchandise; they want to be able to choose from a wide range of shops. Consumers are convenience shoppers and feel free and flexible while switching between channels. Consumers can shop for chore, social pleasure, relaxation, stimulus and for many other reasons (Dennis, Newman & Marsland, 2005) ONLINE CONSUMER BEHAVIOUR Online consumer behaviour has been studied from different scientific perspectives, such as marketing, management, information systems and (social) psychology (Hoffman and Novak, 1996; Koufaris, 2002; Gefen, Karahanna & Straub, 2003; Pavlou, 2003; Pavlou & Fygenson, 2006; Cheung, Chan & Limayem, 2005; Zhou, Dai & Zhang, 2007). The differences between online and offline consumer behaviour is the use of the internet. There are many factors that influence online consumer behaviour (i.e., risk, usefulness, ease of use, consumers attitude towards online channels, social influence, personal online skills and web site features). But research also has turned out that online consumer behaviour is associated with demographics, channel knowledge, perceived channel utility and shopping orientation (Javadi et al., 2012). The internet and new media offer unprecedented information collection and processing capabilities, which changed the buying behaviour of consumers. The internet gives consumers the possibility to obtain more information about price and non-price attributes. Five factors influence consumers online and offline shopping behaviour, these factors are channel-risk perceptions, price-search intentions, search effort, evaluation effort and delivery time (Gupta, Su and Walter, 2004; Gong & Maddox, 2011). Consumers search online for product features, compare prices, read reviews, select products, obtain services, place orders, payments, and much more. The internet has become a worldwide place to sell and buy goods and services 24/7. For retailers, online shops became significant sales channels (Javadi et al., 2012). Online shopping is popular because it has many advantages compared to physical shops. First, online shopping gives consumers an anytime anywhere convenience. There is no travelling and waiting time. Online shops also offer much information about products and services, which makes shopping for bargains more attractive. On the other hand, consumers like brick and mortar shops because they like to experience products; touch, feel, sensory and service are important factors in that experience. The lack of face-to-face communication in online 17

18 shopping reduces consumers faith in shopping (Javadi et al., 2012). A combination between online and offline shopping can offer both the retailer and the consumer the best of both worlds. Social media has become a significant component of the current media landscape. From a business point of view, social media can offer many advantages. Through social media companies can gain insight in their consumers, improve product consciousness, reduce costs and optimize marketing and communication plans. In the Netherlands, social media is integrated in daily life; 8 out of 10 persons in the Netherlands use social media. Facebook, Hyves, Twitter and YouTube are the most popular in the Netherlands. But also Pinterest, MySpace, Foursquare, Instagram, Snapchat and LinkedIn are frequently used by the Dutch population. Social media platforms offer a wide range of opportunities such as sharing photos, videos, music and widgets (Singh, Lehnert & Bostick, 2012). In retail, there is a huge opportunity for social marketing, especially social location marketing can offers retailers great advantages. Social location marketing is a marketing channel for companies on which consumers can share their locations. Originally, the channels were designed for users to communicate with each other (Forbes, 2011). Applications such as Facebook and Foursquare are used to share persons activities at a specific location. For example, when consumers check in at the Bijenkorf on Facebook, this is free advertising for the Bijenkorf, because friends of the consumer can see that the consumers is at the Bijenkorf. 2.2 OMNI CHANNEL SHOPPING BEHAVIOUR To take advantage of the growing internet access of households, retailers have developed strategies for communicating with their customers. Consumers use various channels to satisfy their needs and wants for products and services. Retailers can profit from operating on several channels because cross channel shoppers are more profitable than single channel shoppers. Cross channel shoppers spend more money and are more loyal than single channel shoppers (Lihra & Graf, 2007; Accenture, 2010). In 2010, seventy per cent of Dutch inhabitants were engaged in some way in cross channel shopping. These cross channel shoppers base their choices on customer service, convenience and buying experience. Dutch consumers wish to shop more cross channel; over forty percent of the customers wants to pick up their online ordered products at the shop. On the other hand, thirty five percent of the customers want to buy products in the shop and have them delivered at home. Twenty percent wants to return online bought items in the shop (Accenture, 2010). Retailers have strategies for reaching their customers through one or more channels. They can have a single channel, multi channel, cross channel or omni channel strategy. Figure 3 makes these different strategies clear. Single channel retailers focus on one channel only and, these channels are branded by traditional media or online. Some single channel retailers have a brick and mortar shop and also a website, but this website is not used for commercial activities. The customer is primarily managed in the shop. Those single channel retailers can be small retailers with brick and mortar shops as well as big retailers. Retailers which are involved in several channels for their commercial activities can be multi channel, cross channel or omni channel retailers. The difference between those types of retailers is in their strategy. In multi channel retailing, every channel has its own strategy. The branding of each channel is managed separately. Cross channels retailers and omni channel retailers have one strategy for all channels. The customer is the central object in their strategy. In omni channel retailing, consumers switch continuously between channels. The experience of customers is managed in all channels and improved by a cross channel dialogue. Omni channel shopping gives consumers a blended online and offline experience during their customer journey. The organization is completely focused on customers wants and needs. Omni channel customers use different channels but see them as one. They switch constantly between channels and this switching is seamless and happens (almost) automatically (Unic, 2012). Customers of the Bijenkorf use the Bijenkorf app while shopping in the Bijenkorf, for instance, to compare prices. Consumers can switch between channels, these channels feel as one complete channel. Dutch retailer the Bijenkorf has developed all its channels with the same strategy (Figure 4). This omni channel shopping experience can be focused directly and indirectly on digital and physical channels. The shopping experience of customers is very dynamic because new channels arise continuously. The challenge for omni-channel retailers is to manage customers' experiences on the best possible way across all channels. 18

19 SINGLE CHANNEL MULTI CHANNEL CROSS CHANNEL OMNI CHANNEL FIGURE 3, CHANNEL STRATEGIES (UNIC, 2012) FIGURE 4, BIJENKORF OMNI CHANNEL SHOPPING ( 2.3 THE CUSTOMER JOURNEY Consumers decision making process has changed due to online shopping behaviour. A considerable amount of literature has been published on the consumer decision making process. Steinfield, Bouwman & Adelaar (2002) describe the purchase process in three phases. The first phase is the pre-purchase stage, followed by the purchase stage and the process ends with a post-purchase stage (Kollmann et al., 2012). Suominen (2005) divides the buying process in five phases: activate, browse, configure, decide and purchase. Solomon, Bamossy, Askegaard & Hogg (2002) define the following phases: problem recognition, information search, evaluation of alternatives and last, product choice (Lihra & Graf, 2007). According to Engel et al. (1995), consumer decision making process has seven phases: need recognition, search for information, pre-purchase alternative evaluation, purchase, consumption, post-purchase alternative evaluation and divestment. The problem recognition phase of Solomon et al. (2002) shows similarities with the activation phase of Suominen (2005) and the need recognition of Engel et al., (1995). When consumers show significant differences between the current state and their desired state, then a problem is recognized: the consumer needs a product. Problem/need recognition is a natural process, which can be stimulated by marketing. After consumers have recognized the problem, they want to gain information. This information can be found in the consumers memory (internal search) or it can be acquired from the environment (external search). As the purchase is more expensive, a consumer generally wants to obtain more information about the product. The scope of information search can differ regarding the product type and consumers characteristics such as age, education level and gender (Lihra & Graf, 2007; Engel et al., 1995). When the information about possible products is obtained, the customer can evaluate the alternatives. Consumers may carefully evaluate different products based on the expected benefit or make a routine decision depending on the product category. When relevant options are known, the customer must choose between the product options. Consumers can be influenced in making their decision by experience, available information and attachment to the brand. There are also stages after the product purchase, as described by Engel et al. (1995): consumption, post-purchase 19

20 alternative evaluation and divestment. Consumption is the phase in which the product is used. The postpurchase evaluation is the evaluation of the satisfaction about the product bought. The last phase of the consumer decision making process is the divestment phase, in which the unconsumed product or its residue is disposed. For this research, the buying process of a customer is called 'the customer journey'. This customer journey is more enhanced than the buying process of Steinfield et al. (2002) and comparable with the consumer decision making process of Solomon et al. (2002). The customer journey for this research consists of five phases; stimulation, search for information, purchase, delivery and after sales service. The phases are shown in Figure 5 and described below. These phases are the most relevant phases for retailers. The delivery phase is added in the customer journey because consumers have (among others through online shopping) several options for the delivery of their products. Stimulation Search for information Purchase Delivery After sales service FIGURE 5, THE CUSTOMER JOURNEY (ADAPTED FROM ENGEL ET AL., 1995) Description of the customer journey Phase 1 - Stimulation: During the first phase customers become inspired by a product. This phase is not always visible; consumers can get stimulated to buy a product in their subconscious mind. For instance, when consumers see an advertisement of the Hema on Facebook. Then the consumer can get attracted to buy the product. Phase 2 - Search for information: During the search phase, customers know what they are looking for because they are already inspired. Consumers are searching for information about products and/or services and their providers. This process can take a long time, but it can also be done in a short time, it depends on the type of product. There is a distinction between high involvement and low involvement products. Grocery is in general a low involvement product; a house for instance is a high involvement product. Phase 3 - Purchase: When all the information is collected, a purchase decision can be made. During this phase the consumer determines which product or service is actually purchased for what price and from which supplier. Phase 4 - Delivery: When the product is bought, there are few options depending on the channel where the product is bought; take the product home directly, delivery at home or at work, collect from a pick-up point or pick-up later in the shop. Phase 5 - After sales service: Then there is the after sales service. There are several channels through which consumers can reach companies offline as well as online. The importance of after sales service should not be underestimated; companies with a good after sales program acquire customer loyalty. Every phase in the customer journey requires the right information in order to go to the next phase. As result of an increasing number of digital touch points, the customer journey becomes more and more complex. Touch points are expressions of a retailer to contact the customer. On each touch point, the consumer is going through an experience, and this experience can be positive as well as negative (Kloet & Lenssen, 2012). 20

21 2.4 CONCLUSION In this chapter insight is gained in the concepts omni channel shopping behaviour and the customer journey. It appears that consumer shopping behaviour has four main aspects, for this research two aspects are most interesting communication behaviour and purchase behaviour. The use of internet has changed consumers shopping behaviour. Mainly the number of channels used during the customer journey is increased. Nowadays, the number of consumers that shop omni channel increases. Omni channel shopping is continuously switching between several channels online as well as offline. In addition, retailers with an omni channel strategy offer the same experience in every channel; this gives the consumer a seamless experience through all channels. Omni channel customers use different channels but experience them as one. The omni channel shopping behaviour for this research is regarded during the customer journey. The customer journey consists of five phases; stimulation, search for information, purchase, delivery and after sales service. Through omni channel shopping, the customer journey became complicated. Consumers use and can use many channels during their customer journey. In addition, consumers are well informed about products because of all these channels. This makes consumers more critical about their purchases. They know what they can buy, for what price and where. Whether consumers use several channels during their customer journey depends on consumers channel knowledge, demographics, perceived channel utility and shopping orientations. Factors which estimate online and offline buying behaviour are channel risks, price-search intentions, search effort, evaluation effort and delivery time. Consumers like to shop in brick and mortar shops because they consider experiencing products, touch and feel and service as important factors. 21

22 22

23 3 EVALUATION OF CONSUMER SEGMENTATION This chapter provides an insight in consumer segmentation for the purpose of analysing consumers shopping behaviour regarding the customer journey and omni channel shopping behaviour. This chapter provides insights by answering the following sub question: How can consumer segmentation be used for analysing omni channel shopping behaviour during the customer journey? As stated before, it is necessary to gain insight in consumer behaviour; who the consumers are, what they buy, why they buy it and where they buy it. This information can be used to acquire new customers, but it can also be used to increase customer s involvement and loyalty. Segmentation techniques can discover information about consumers omni channel shopping behaviour during the customer journey. First, the definition of consumer segmentation is given, and on the basis of this definition, the process of segmentation is explained (section 3.1). Next, several segmentation methods (section 3.2) and techniques (section 3.3) are described in order to select the best applicable method for segmenting consumers based on their omni channel shopping behaviour during the customer journey. In addition, a study is conducted on segmentation in literature (section 3.4). This chapter finishes by answering the sub question in the conclusion (section 3.5). 3.1 CONSUMER SEGMENTATION The best method to study consumer behaviour is consumer segmentation (Gilboa, 2009). Consumer segmentation was introduced in 1956 by Wendell Smith. Differences between groups of consumers were seen as opportunities for the market (Raaij & Verhallen, 1994). Consumer segmentation assumes that consumers exhibit heterogeneity in product preferences and buying habitats (Dibb, 1998). Segmentation is the process of dividing a heterogenic market into groups of customers based on similar characteristics, wishes, needs, buying habits or reactions to marketing activities. Companies can choose specific segments they would like to serve. When companies choose to serve all customer types, than they have an undefined marketing strategy. Google is a company which has chosen for an undefined market; their search machine has been developed for everybody. When a company chooses to serve different specific segments, it is called defined marketing. Car brands use defined marketing; they have different models for different types of customers. Another way of marketing is niche marketing, which means that a company is focussed on one specific segment. Prenatal has such a marketing method; it is focussed on pregnant women. In the nineties, a scientific discussion about the concept of segmentation took place. Lifestyles became more and more segmented and probably did not offer a solid basis for effective and efficient market segmentation. In addition, there was criticism on segmentation based on social class, age and gender. The general consensus was that these variables gave less useful information than originally thought, with respect to predicting specific behaviour of consumers (Quinn, 2009). For example, when segmentation takes place based on a person s age, persons with the same age are often placed in the same segment: however, persons with the same age can show totally different behaviour. Therefore, demographic variables are often used for profiling the segment, not for forming segments (Raaij & Verhallen, 1994). In addition, improvements to data management technologies by companies led to more opportunities for segmentation based on product-related factors such as benefits and usage patterns. Through these improvements, companies gained better access to an increasing number of segmentation variables. Especially segments based on behavioural data are valuable, behavioural data offers an appropriate basis for segmentation (Wedel & Kamakura, 2000; Dibb, 2001). Especially, segmentation based on product-specific behavioural variables give reliable formats. Because people who are grouped on the basis of similarities in certain behaviours, exhibit a high degree of internal homogeneity. Internal homogeneity is a condition for effective segmentation. As discussed earlier, segmentation based on global variables violate the condition of internal homogeneity; the chance that people form a segment on the basis of product-specific behaviour will respond to the deployment of a particular marketing program. Another condition of effective segmentation is hereby mentioned, namely predictability of the segments (Thomas, 1980). 23

24 3.2 SEGMENTATION METHODS A large number of different segmentation methods can be applied to find consumer segments within a dataset. Firstly, these methods can be divided in four approaches. A distinction can be made between a priori (before) and post hoc (after) methods. In a priori segmentation methods, segments are defined by the researcher. Researchers choose a variable or a set of variables, and based on this set of variables consumers are segmented. After the segments are formed, segments are described based on differences in personal characteristics (Green, 1977; Kriellaars, 2011). Post hoc segmentation is segmentation of consumers after data collection. The number and types of segments are not known before the data-analysis, they depend on the data which is gained from respondents. The data shows similarities by grouping selected variables (Wind 1978; Kriellaars, 2011). Secondly, segmentation methods can be divided based on whether statistic methods give descriptive or predictive information (Table 1). Predictive methods distinguish dependent (i.e. buying behaviour) and independent (i.e. age, gender or income) variables. Dependent variables are predicted by independent variables. For instance, buying behaviour can be predicted by consumers age, gender or income. Descriptive methods do not use this distinction, they analyse relations between variables. Several segmentation methods are applicable, these methods can be divided in four categories (Wedel & Kamakura, 2000; Kriellaars, 2011): TABLE 1, CLASSIFICATION OF SEGMENTATION METHODS (WEDEL & KAMAKURA, 2000) Descriptive Predictive A priori Cross tabs Log linear models Regression Discriminant analysis Logit framework Post-hoc Cluster analysis Latent class analysis (LCA) Latent class analysis (LCA) The goal of this research is to find similarities and differences between customers. For instance, whether there are similarities in omni channel buying behaviour between groups of consumers. Within this research it is not necessary to gain insight in why some groups of consumers react differently to specific stimuli for buying products. Therefore, descriptive segmentation methods are the best applicable for this research. The character of this research is descriptive. Similarities and differences between customers in a set of variables are analysed (omni buying behaviour during the customer journey). The segments are not determined prior to the research; they are formed on the basis of the data. Therefore, cluster analysis or Latent Class Analysis are appropriate methods for analyzing data for this research. Latent Class Analyses are often used in combination with conjoint analyses. However, data within this research is not appropriate for LCA analysis. Therefore, a cluster analysis is used for this research. 3.3 CLUSTER ANALYSES IN GENERAL Cluster analysis is the name of a group of multivariate techniques for grouping consumers based on their characteristics and behaviour (see table 1, cluster analyses) (Hair, Anderson, Tatham & Black, 1998). Every cluster contains a group of consumers with similar characteristics, thus within the cluster the characteristics of consumers are homogeneous. Between clusters there are differences in characteristics of consumers, these clusters are thus externally different. The most popular approaches for clustering data are hierarchical, partitioning and TwoStep cluster analyses. Similarities between consumers can be estimated by calculating the distance between consumers. Consumers with a smaller distance towards each other are more similar; consumers with larger distances are more dissimilar (Mooi & Sarstedt, 2011). Mooi & Sarstedt (2011) made a stepwise plan for clustering data (Figure 6). These steps are described in more detail in section 3.3.1, 3.3.2, 3.3.3, and

25 Decide on the clustering variables (section 3.3.1) Decide on the clustering procedure (section 3.3.2) Hierarchical methods Partitioning methods TwoStep clustering Select a measure of similarity or dissimilarity Choose a clustering algorithm Decide on the number of clusters (section 3.3.3) Select a measure of similarity or dissimilarity Sel Validate and interpret the cluster solution (section 3.3.4) FIGURE 6, STEPS WITHIN A CLUSTER ANALYSIS (MOOI & SARSTEDT, 2011) CLUSTERING VARIABLES The cluster process starts with selecting appropriate clustering variables. This can be done based on a mixture of intuition, data and the research question. Mistakes in assumptions can result in inappropriate segments, and thus in wrong strategies. There is a distinction between general and specific clustering variables on the one hand and observable and unobservable variables on the other hand. General clustering variables are independent of products, services or circumstances. They describe general characteristics of a person, such as psychographics and culture. Specific variables are related to the consumer and the product, service and/or particular circumstances. Specific variables describe situation specific characteristics, such as benefits sought in a product (smartphone) or usage frequency (Mooi & Sarstedt, 2011). Table 2 shows the four types of clustering variables. TABLE 2, TYPES AND EXAMPLES OF CLUSTERING VARIABLES (MOOI & SARSTEDT, 2011) Observable Unobservable General Cultural, geographic, demographic, socio-economic Psychographics, values, personality, lifestyle Specific User status, usage frequency, shop and brand loyalty Benefits, perceptions, attitudes, intentions, preferences. Segments are formed depending on the selected variables. Differences and similarities between input variables are the basis for forming clusters. It is important to select variables which show differentiation between segments regarding the objective of the segmentation analysis (Mooi & Sartedt, 2011). There should be significant differences between the dependent variables across the clusters. For example, when differences are found in the buying behaviour of consumers, then consumers can be grouped based on the differences in buying behaviour. When there is just one type of buying behaviour, for example, all consumers use only the shop during the customer journey, than only one group will be found based on the buying behaviour of the respondents. 25

26 3.3.2 CLUSTERING PROCEDURE Each clustering method has its own way of forming clusters, however, it always involves optimizing a criterion, for instance making the variance within the cluster as small as possible or making the distance between clusters as big as possible. Within cluster analyses in practice, there is a distinction between three methods; Hierarchical, partitioning and TwoStep clustering (Mooi & Starstedt, 2011). In the next sections hierarchical, partitioning and TwoStep clustering methods are discussed (Figure 6). Hierarchical clustering Hierarchical clustering is a stepwise clustering method with a tree structure. There is a distinction between agglomerative and divisive clustering within hierarchical clustering methods. The agglomerative clustering starts with a cluster for each consumer, and then the consumers (clusters) with the nearest distance (the most similar consumers) are merged and linked to a higher hierarchy. In subsequent steps, this process continues. Divisive clustering methods perform this analysis vice versa. In the first place, all consumers are assigned to one single cluster, and this cluster is then gradually split up. When a consumer is assigned to a cluster, it cannot be assigned to another cluster anymore. Divisive methods are rarely used in segmentation studies. Because this method is complex to compute, there are 2 N-1-1 possible divisions for a cluster. For example, when we need to cluster a dataset of 2000 respondents (N=2000) then there are clusters to be analysed, which is a lot. Therefore, agglomerative hierarchical clustering methods are used more often (Xu & Wunsch, 2009; Mooi & Sartstedt, 2011). Agglomerative hierarchical clustering There are several methods for measuring similarities between consumers. A straightforward method for measuring the most similar consumers is to draw a straight line between two consumers (Mooi & Startstedt, 2011). The distance between two consumers x and y is the square root of the sum of the squared differences (Mooi & Starstedt, 2011). D(x, y) = (x 1 y 1 ) 2 + (x 2 y 2 ) (x p y p ) 2 (Euclidean distance) D is the distance between two consumers x and y. The variables are identified by index p are the variables. The distance is defined as the square root of the sum of squares of differences between the corresponding coordinates of the points. This distance is the length of the line which connects consumer x and y. Sometimes data needs to be standardized because when variables are measured on different scales they do not contribute equally to the analysis. This standardization can be conducted through several methods. For instance, z-standardization, this method rescales every variable to a mean of 0 and a standard deviation of 1. Another way to standardize data is by range. In addition, data can be standardized by looking at the correlation between variables. When the method for measuring distance or similarity is chosen, then a cluster algorithm can be selected. Below, there is a list of the most often used agglomerative hierarchical cluster algorithms. These algorithms differ in the way in which the distance is defined. Within this research four hierarchical cluster algorithms are considered. These algorithms are single linkage, complete linkage, average linkage and centroid linkage. The differences between these algorithms are explained and clarified by Figure 7. Single linkage (nearest neighbour): The distance between a pair of clusters is determined by the two closest consumers to the different clusters. This clustering method inclines to make long-drawn-out clusters. Because consumers within two clusters with the smallest distance connect those clusters. This gives a chaining effect (Everitt et al., 2001). Because of this, two clusters with very different characteristics can be connected to each other because of noise. But, when the clusters are far separated from each other, then the single linkage method works well. Complete linkage: The distance between the clusters is determined by the distance between the two farthest consumers, each in another cluster. This technique ensures a maximum distance between the consumers in a cluster. 26

27 Average linkage: The distance between two clusters is the average distance between all pairs of consumers within the clusters. Centroid (geometric center) linkage: Two clusters are merged based on the distance of their centroid. FIGURE 7, CLUSTER ALGORITHMS (MOOI & SARSTEDT, 2011) Partitional clustering Partitional clustering assigns, contrary to hierarchical clustering, consumers to clusters without a hierarchical structure. The consumers are assigned to non-overlapping clusters. The best known method for partitional clustering is the k-means algorithm (Xu & Wunsch, 2009; Berkhin, 2006). The k-means method represents each cluster by the mean. K-means seeks an optimal partition of the data by minimizing the sum-of-squared-error. K- means clustering minimizes the sum-of-squared-error with an iterative optimization procedure, see the objective function below. E(C) = x i C j 2 j=1:k x i Cj Where x i C j 2 is the chosen distance function, x i is the consumer and C j is the cluster center (centroid). k is the number of clusters (C j ). The k-means clustering procedure is thus a procedure which minimizes the distance (difference) between the cluster centre and the point which represents the consumer. As a result, optimal clusters are formed in which every consumer within a cluster is (almost) similar to the average consumer within the cluster. k-means clustering has two versions. The first version is the EM (Expectation- Maximization) algorithm, also called the Forgy s algorithm, which consists of two-step major iterations which reassign all consumers to the nearest cluster based on the centroids (geometric center). Centroids of newly assembled clusters are recomputed. This process iterates until a stopping criterion is achieved (Xu & Wunsch, 2009). Another version of k-means clustering is iterative optimization, which reassigns consumers based on a detailed analysis to another cluster. This analysis shows the effect on the objective function of moving a consumer from a current cluster to any other cluster. When this move has a positive effect then the consumer will be relocated and the cluster centroids need to be computed again. According to Xu & Wunsch (2009), the k-means algorithm owes its popularity to the comprehensibility, easy application and solid basis of analysis of variance. But they also indicate some disadvantages of the k-means algorithm. First, the results depend mainly on the initial guess of centroids. Thereby, calculated local optimums can be different from the global optimums. In addition, it is not clear how the number of clusters can be chosen. Further, the process is sensitive to outliers. The basic algorithm is not scalable. Only numerical attributes are covered and resulting clusters can be unbalanced. 27

28 TwoStep clustering TwoStep cluster analysis is a method to find natural groupings within a dataset, this algorithm differs from traditional clustering methods. In the first place, a mixture between categorical and continuous variables can be used within this analysis, while k-means and hierarchical clustering only allow ratio/interval scaled variables. The method treats variables as independent variables, a combined multinomial normal distribution can be located on both categorical and continuous variables. Secondly, this procedure selects the best number of clusters automatically. The most optimal number of clusters can be selected by comparing the values of the model-choice criterion of the cluster solutions. Thirdly, by building a Cluster Function (CF) tree the records can be summarized. This makes it possible for this analysis to analyze large datasets, just like the k-means clustering procedure (Moiseeva, 2013). This method is developed in 2001 by Chiu, Fang, Chen, Wang & Jeris and works especially for SPSS statistics. The TwoStep cluster technique consists of two steps, pre-clustering and clustering. Firstly, the consumers are classified into pre-clusters. These pre-clusters are clustered within the second step with a hierarchical procedure (Moiseeva, 2013). Within this step a CF tree is used which consists of ranks of nodes. Every node has a number of inputs. A node can be an internal node or a leaf-node. Leaf-nodes are final sub-clusters, while internal nodes are used to guide a new consumer (record) into the correct leaf-node. When a consumer reaches a leaf-node, then the algorithm finds the closest leaf entry in the leaf node, then the CF three is updated. The consumer will only be assigned to a cluster when the consumer is within a threshold distance of the nearest leaf entry, otherwise this consumer starts a new leaf node. When there is not enough space within an existing leaf node then this leaf node will split up. When the CF tree becomes too big, the threshold distance must increase and the CF tree will be rebuild. This is an iterative procedure which is finished when all consumers are assigned to a cluster. Distance measure The distance between the clusters is calculated with a log-likelihood or Euclidean measure. The Euclidean distances measure is within this cluster analysis used for continuous variables. The distance between two cluster centers is measured. In comparison to the Euclidean distance measure, the log-likelihood distance measure can handle continuous as well as categorical variables. However, it is assumed that continuous variables are normal distributed, categorical variables are assumed to be multinomial. Thereby, all variables must be independent from each other (Mooiseva, 2013). The log-likelihood distance measure is based on probability. The following equation shows the definition of the distance between cluster 1 and 2: D(1,2) = Log L1 + Log L2 Log L < 1,2 > Within this formula D(1,2) is the distance between cluster 1 and 2; Log L1 is the log-likelihood for the cluster 1, consequently Log L2 is the log-likelihood for cluster 2; < 1,2 > is an index that represents the cluster is formed by combining clusters 1 and 2. The measure Log L for cluster s is defined as: Log L = N s ( k=1 log(σ 2 k 2 + σ 2 sk ) + Ê sk K A 1 K B k=1 ) Where Ns is the total number of consumers in cluster s; K A is the total number of continuous variables; K B is 2 the total number of categorical variables; σ k is the estimated variance of the continuous variable k, for the 2 entire dataset; σ sk is the estimated variance of the continuous variable k, in cluster s. Ê sk is calculated in the following equation. L k Ê sk = N s ( N skl l=1 N s log N skl N s ) Within this equation N skl is the number of consumers in a cluster s whose categorical variable k takes l category. 28

29 Auto-clustering procedure The auto-clustering procedure makes the TwoStep clustering interesting. This analyse technique can determine the number of clusters automatically. However, researcher have the ability to specify a number of clusters. For automatically selecting the number of clusters this clustering technique uses two steps. Within the first step, the Bayesian Information Criterion (BIC) or Akaike s Information Criterion (AIC) is calculated for every number of clusters, to find an optimal number of clusters. AIC is a measure of the goodness of fit of any estimated statistical model. AIC can in special find unknown models that have high dimensional reality. BIC is designed to find the most probable model given the data (Mooiseva, 2013). BIC k = 2L k + r k log n (BIC) AIC k = 2L k + 2m k (AIC) Where n is the sample size, L k is the log-likelihood and m k is the number of free parameters to be estimated. According to studies comparing the performance of AIC and BIC, AIC performs well in small samples, but it is inconsistent and it does not improve in performance in large samples. In contrast, BIC is consistent and improves in performance in large sample sizes (Mooiseva, 2013). m k is computed according to the following equation: K B m k = J {2K A + (L k 1) } L k is the number of categories for categorical variable number k. The second step in this clustering procedure refines the estimation of the first step. The following equation is used to calculate the change in distance between the two closest clusters. R(k) = k=1 d min (C k ) d min (C k+1 ) R(k) is the cluster model containing k clusters and d min (C k ) is the minimum inter-cluster distance for model C k. Consequently, C k+1 is the next larger model that contains one cluster more and d min (C k+1 ) is the minimum inter-cluster distance for cluster model C k+1. For every solution SPSS calculates models with different number of clusters (based on the maximum number of clusters) and shows for every solution the change of BIC and ratio distance. Selecting the clustering procedure Now all clustering techniques are discussed, we can make a good choice for selecting a cluster analysis technique for this research. Within this research we are probably dealing with a large dataset, therefore partitional clustering methods are not applicable. For measuring buying behaviour and consumer characteristics, it is probably necessary to use nominal and ordinal scales. K-means and hierarchical clustering techniques require datasets with only interval and ratio variables. Variables like gender (nominal) and education level (ordinal) cannot be used within K-means and hierarchical clustering techniques. TwoStep clustering techniques do allow continuous as well as categorical variables. Therefore, the TwoStep clustering technique is the best method to form clusters for this research NUMBER OF CLUSTERS Within TwoStep cluster analysis technique the number of clusters can be selected automatically. In addition, it is possible to determine the maximum number of clusters within the TwoStep cluster algorithm. The maximum number of clusters which can be formed by the algorithm can be determined based on the objectives of the research, the number of consumers and the number of variables. On basis of this information, the researcher can interpret what the maximum number of clusters may be. 29

30 3.3.4 CLUSTER SOLUTION INTERPRETATION AND VALIDATION For the validation of the results, the validation criterion needs to be determined. These validation criterions are different for theoretical research than for marketing research. Criterion variables in theoretical research are related to cluster variables, but these are not included in the analysis. Criterion variables in marketing research are commonly related to managerial outcomes. When the criterion variables show significant differences, then the clusters are distinct groups with criterion validity. In order to make an evaluation choice for a cluster solution, the following criteria are helpful (Dibb, 1999; Mooi & Sarstedt, 2011): Substantial: segments are profitable enough and large enough. Every cluster needs to be clear and include at least 100 respondents. Accessible: segments can be reached and served effectively; this requires them to be characterized by means of observable variables. There are variables (male/female, age) which do not give significant differences within the cluster solution but which are important for characterizing the clusters. Differentiable: the clusters can be distinguished conceptually. When consumers within several clusters need to be reached by advertisement leaflets, it is important that when there are differences within the response of consumers on advertisement leaflets, that they do occur within the clusters. Actionable: effective programs can be formulated to attract and serve the clusters. Every shopper typology has an own approach to be attracted and served while shopping. Stable: clusters should be stable over time, to be successful for marketing strategies. When clusters are formulated it is important that these clusters do not change every month. Parsimonious: to be managerially meaningful, only a small set of substantial clusters should be identified. It is important that the formed clusters are usable in practice; therefore the maximum number of manageable clusters needs to be estimated. Familiar: for managerial implications, the clusters should be comprehensive. The name, description and information about a cluster should be clear. Relevant: clusters should be relevant in respect of the companies competencies and objectives. In this study Wereldhave wants to gain insight in the omni channel buying behaviour of consumers during the customer journey. In order to be relevant the cluster should give this information. Compactness: clusters exhibit a high degree of within-segment homogeneity and between-segment heterogeneity. Compatibility: clustering results must meet other managerial function requirements. Cluster analyses end up describing a cluster. Therefore, centroids of clusters needs to be examined. These are the average characteristics of all consumers in a certain cluster. Only when clusters have significantly different means, then they are distinguishable. This can be determined by comparing the correlation of consumers characteristics with clusters. With that information the cluster can gain a name or label. Variables which are used for clustering are often unobservable (buying behaviour); this makes it hard to assign a new consumer to an existing cluster. However, observable (age, education level) variables reflect the partition of the consumers in a cluster very well. 30

31 3.4 SEGMENTATION STUDIES In order to gain insight in applications of segmentation techniques, a literature research is conducted. A considerable amount of literature has been published on consumer segmentation. Table 3 compares relevant consumer segmentation studies. The studies are classified on subject (shopping centers yes/no), variables and whether multi channel shopping is integrated. Omni channel buying behaviour has not been implemented in segmentation research yet, multi-channel shopping was integrated in some segmentation methods in 2004 and The studies on consumer segmentation are discussed below. For an explanation about the segmentation methods see sections 3.2 and 3.3. TABLE 3, SEGMENTATION STUDIES AUTHOR (YEAR) AREA SHOPPING CENTER VARIABLES GEO GRAPHIC SOCIO- DEMOGRAPHIC PSYCHO GRAPHIC BEHAVIOUR BUYING PROCESS ACTIVITIES BENEFITS PRODUCT CATEGORIES Bloch et al. (1994) x x x x x x Jarret (1996) x x Frasquet et al. (2002) x x x Reynolds et al. (2002) x x x MALL ATTRIBUTES Bhatnagar & Ghose (2004) x x x x Keen et al. (2004) x x x x x Ruiz et al. (2004) x x x x x x x Konuş et al. (2008) x x x x x x Gilboa (2009) x x x MULTI CHANNEL In 1994, Bloch, Ridgway & Dawson published a paper in which they identified shopping center related shopping orientations by exploring differences in shopping center habitat activity patterns. 600 consumers were divided into groups based on their shopping behaviour. These groups were clustered using hierarchical as well as nonhierarchical cluster analyses. First, Ward s method is used to determine the number of clusters. Followed by forming clusters using a K-means cluster algorithm. With this research, four clusters of shopping center shoppers were formed, namely enthusiasts (higher than average value on every activity dimension), traditionalists (higher score than average on mall-focused activities and relatively high on product purchasing), grazers (high tendency to pass time in the mall browsing and eating) and minimalists (low participation in all activities). Significant clusters were found based on the intensity of mall shopping. A few years later in 1996, Jarret identified a set of variables that is relevant and appropriate for shopper segmentation. Through a telephone interview, information was collected from 931 consumers in three Australian trading areas. With cluster analysis, six shopper types were identified based on the consumers importance of the shopping offer (variety, price, quality, comparative, shopping and value), the shopping environment (progressive, exciting, clean, attractive and interest) and the shopping service (friendly, helpful, parking and information). These shopper typologies are: have to shoppers (low scores importance of shopping offer, service and environment), moderate shoppers (moderate scores importance of shopping offer, service and environment), service shoppers (moderate scores on importance of shopping offer and environment and high score on importance of shopping service), experiential shoppers (high scores on the importance of shopping offer, service and environment), practical shoppers (moderate scores on importance of shopping environment) and product focused shoppers (moderate score on importance of shopping offer and low scores on importance of shopping service and environment). These developed shopper typologies were used to develop a retail strategy. Another research was conducted by Frasquet, Gil & Mollá (2001), who used the consumer choice modelling method to analyse two objectives. The first objective was to analyse the perceived value on shopping-centre 31

32 selection. The second objective was to investigate benefits of adopting a segmentation approach in the study of consumer preference for a shopping center. A survey was conducted in shopping centers of Valencia. Segments were formed by analysing answers on twenty items of the value scale. In the first place, demographic segmentation criteria were chosen. This has two advantages. First, it produces segments which are easy to identify and measure, and second consumers wants and preferences are often linked to demographic characteristics. Ward s method was used to choose the number of clusters, followed by K-means clustering for clustering the final segments. Variables which were used to identify the segments were age, occupation and marital status. Reynolds, Ganesh & Luckett (2002) researched differences between factory outlets and traditional malls through consumer segmentation. The importance of mall attributes, mall essentials, entertainment and convenience were rated by respondents. Through a mall intercept study, data was collected from 1097 traditional mall shoppers and 827 outlet shoppers. A multistep-cluster analysis (first Ward s method, then K- means) was conducted to form clusters. Through this research, five shopper segments were found for the traditional mall as well as the factory outlet the enthusiasts, the basic, the apathetic, the destination and the serious. Unique for the factory outlet was the cluster brand seekers. Multi channel shopping behaviour was studied by Bhatnagar and Ghose (2004). They used a latent class modelling approach to segment online shoppers based on their purchase behaviour across several product categories. Data was collected from 1330 respondents through an online survey. In the online survey, respondents had to evaluate online shops in general on 11 attributes. Within this study consumers are segmented based on benefits, to gain insight in benefits that respondents perceive from online shopping. For describing the segments, variables such as age, education, gender, income, marital status and internet experience were used. A Latent Class modelling approach was used for the segmentation research. An important finding of this research was that web shoppers do not consider getting the lowest price as an important attribute. Respondents who did search online often did not buy online because of their perception about security and sensitive information. Another study on multi channel shopping behaviour was conducted by Keen, Wetzels, Ruyter & Feiberg (2004). The research was conducted to investigate the structure for consumer preferences in making product purchases. According to this research, the structure of the consumer decision-making process depends on the retail format and price of the desired product. Keen et al. (2004) analysed three channels: shop, catalogue and the internet. Two product categories were considered, CD s and personal computer. Data was collected from 281 shopping center shoppers in a suburb of Chicago with the shopping center intercept technique. Through conjoint analyses, the structure of the decision and the importance of attributes in the decision-making process were estimated. Clusters were formed through hierarchical as well as k-means clustering in two stages. Interesting is the segmentation study of Ruiz, Chebat & Hansen (2004). This study is based on a methodology developed by Bloch et al. (1994). Shoppers are segmented on the basis of their performed activities during their shopping center visit. Variables such as perception, emotions and motivations were used to extend the data. 889 questionnaires were collected in a shopping center in Eastern Canada. A series of Yes and No questions related to activities performed in the shopping center during their visit were answered through a questionnaire among mall customers. Visitors were also asked about the frequency of their visits and the number of purchases. The p-median model was used to find a structure in the dataset and to estimate the optimal number of clusters. Chi-square tests were conducted to test the significance of the segments. The base variables for the segments were differences in activity patterns (do exercise, talk with other customers, browse, take a snack, go to the bank, unplanned purchase, purchase). Descriptive variables were used to describe segments and these variables are classified into these groups: geographic (postal code), sociodemographic (age, mother tongue, sex, annual income, education level, number of children under eighteen at home and occupation), psychographic (perceptions, emotions, atmospheric variables, approach avoidance reactions, motivations, non-economic costs) and related benefits sought. Konuş et al. (2008) analysed the multi channel shopping behaviour of Dutch consumers through segmentation. The segmentation study of Konuş et al. (2008) can therefore give a good basis for researching omni channel shopping behaviour during the customer journey of Dutch consumers. The research focuses on two phases of 32

33 the buying process, the search and the purchase phase. Consumers are segmented based on their attitude towards several channels as search and purchase alternatives. A survey was conducted among 364 Dutch consumers in a research panel. Three types of channels (brick and mortar shop, the internet and catalogues) were evaluated by the consumers in terms of their appropriateness for the two phases. The latent-class analysis technique in combination with the Bayesian information criterion (BIC) was used and three segments were found. The multichannel enthusiasts, uninvolved shoppers and shop-focused consumers. Several descriptive variables (shopping enjoyment, loyalty and innovativeness) were used to predict which persons belong to which segments. The research did not find significant relationships with socio-demographics. This confirms prior findings that consumer behaviour is driven more by psychographics. The results demonstrate that segment membership is affected by hedonic and economic variables. In 2009, Gilboa identified four shopper types based on the shopping behaviour of Isreali shopping center visitors. Behaviours were divided into three categories: visiting patterns, motivations for trips to the mall and activities engaged in during the visit. This segmentation study labelled four types of customers: disloyal, family bonders, minimalists and mall enthusiasts. Data from 636 Israeli consumers was obtained in order to form these shopper types. Research variables for shopping center visits were motivation, activities performed during the visit, visiting patterns and personal details. To find out whether Israeli consumers can be divided into distinct groups of consumers, a TwoStep cluster analysis was conducted. This analysis combines the hierarchical analysis method of Ward with the non-hierarchical k-means clustering procedure in order to optimize the cluster solutions. The TwoStep cluster analysis was most suitable for this study because both categorical and continuous variables can be used. 3.5 CONCLUSION This chapter has investigated consumer segmentation, in order to answer the main question of this chapter: How can the customer journey and omni channel shopping be implemented in consumer segmentation? The customer journey and omni channel shopping are explained in chapter 2. Now it is time to implement omni channel shopping in combination with the customer journey in the segmentation strategy. In the recent years, there has been an increasing amount of literature on consumer segmentation. However, omni channel shopping behaviour is not researched (yet). In their study Konuş et al. (2008) found consumer typologies based on consumers channel orientation. This research was conducted over two phases of the shopping process, namely the purchase phase and the information search phase and they investigated three channels (brick and mortar shop, internet and catalogues). From this research we have learned how consumer segments can be formed based on channel orientation. Within our research, the customer journey has five phases; stimulation, search for information, purchase, delivery and after sales service. For this research, both online and offline channels must be selected. Even though some channel behaviour is easy to predict, it is interesting to gain information about channel usage during the customer journey of consumer target groups. It is assumed that consumers choose from a set of channels for every phase separately. Collected data can be analysed by several cluster analysis techniques, in order to discover omni channel consumer segments. There is no information about the shopper typologies before the segmentation study is conducted, therefore the segmentation base is post-hoc. First of all, variables for clustering need to be selected. From literature research we have learned that clustering consumers on socio-demographics does not result in interesting clusters. In addition, interesting cluster can be found by using information about consumers behaviour, activities and benefits. From studies we have learned that it is interesting to form clusters based on channel selection during several phases of the customer journey. TwoStep clustering is the most suitable method for analysing clusters in this research, because we probably need to handle a large dataset. Thereby, TwoStep clustering can handle several types of variables on different scales (ordinal, ratio and nominal values). In addition, it is not necessary to determine the number of clusters before cluster analysis take place. 33

34 34