UPSELLING Clarifying the concept of upselling and finding drivers

(1)

1

UPSELLING

Clarifying the concept of upselling and finding drivers

by SIMONE EILANDER MSc BA Marketing Master thesis 8 July 2012 Nieuwstadsweg 14 8081 AB Elburg 06-20826765 s1930680@student.rug.nl Student number S1930680

(2)

2

SUMMARY

There is a growing realization within firms that it is cheaper to retain the current customers, rather than acquiring new ones. Because of this understanding, firms realize that they need to build a strong relationship with their customers in order to retain them. There are two important strategies to retain customers in the field of customer relationship management (CRM); i.e. cross-selling and upselling (Kamakura, 2007; McNally, 2007). Despite the realization of the importance of both strategies, the difference of both strategies and the drivers of upselling have been underinvested.

(3)

3

PREFACE

After I completed my Bachelor Marketing Management, I really wanted to gather more knowledge about Marketing and go into the depth of the matter. So I decided to register for the Master Marketing Management at the Rijksuniversiteit Groningen.

In front of you lies my Master Thesis. This thesis is the completion of my Master Marketing Management. The past few years I experienced as interesting and fun. I feel that I have grown in knowledge, but also in the way of approaching different situations. During my education, I gained a lot of knowledge and I am looking forward to apply this in ‘real’ life. This Master degree is a good step in the right direction, but I am aware that I still have got to learn so many things during my career.

I would like to express my most sincere gratitude and appreciation to my supervisor Dr. Liane Voerman, for her support and guidance during the establishment of this thesis. Thanks for your help, encouragement, flexibility, and patience throughout the whole process. You have been a great help and I learnt much from you. In addition, I would like to thank my second supervisor, Dr. Ir. Maarten Gijsenberg, for his feedback for finalizing this thesis. Furthermore, I would like to thank the Rijksuniversiteit Groningen for research assistance.

At last, but certainly not at least, I would like to thank my husband, my parents and my family and friends, for their support, encouragement, and their never ending trust and faith in me. Thank you!

Simone Stremmelaar-Eilander

(4)

4

1. INTRODUCTION

Nowadays there is a growing realization within firms that it is more profitable to retain the current customers, rather than acquiring new customers (e.g. Reichheld and Sasser, 1990; Zeithaml, 2000; Reinartz and Kumar, 2000, 2002; Kumar et al., 2008; Kamakura, 2007).

Increasing profits by customer retention works twofold. On one hand, when a firm is able to achieve that customers stay longer at the company, the revenues of the company will rise. Not only the cumulative profits per customer will rise, but the customers also will be increasingly profitable with every year they stay with the company, because loyal customers tend to buy more products or services (Reichheld and Sasser, 1990). Reichheld and Sasser (1990) also state, that retaining 5% more customers will lead to an increase in profits of 25% up to almost 100%. Thus, a relatively small increase in retention can be very profitable.

On the other hand, with a customer retention focus, firms are also able to save on acquisition costs. According to Zeithaml (2000), it costs about five times more to acquire a new customer, than to keep an existing customer and, selling costs for present customers are about 20% lower than selling costs for new customers. So, by focusing on present customers (where the firm already invested in, and thus, where the acquisition costs are already sunk costs), firms are able to gain relatively more profit as they do when they focus on acquiring new ones.

Furthermore, in addition to the increase in revenues and the acquisition cost reduction when consumers retain longer with the firm, companies also learn to know the customer, they gain experience with them, and are therefore able to serve them more efficiently and effectively (Reichheld and Sasser, 1990). On the consumer side, improved perceived service quality and customer satisfaction will lead to the willingness of customers to pay price premiums, more positive-word of-mouth, and thus to attraction of new customers, which leads to higher revenues, a greater market share, and to more profitability (Rust et al.,1995; Zeithaml, 2000).

1.1 Strategy for retention: CRM

(6)

6 customers (i.e. Customer Relationship Management (CRM)), as Reinartz et al. (2004) mention. The result is that CRM has become one of the most familiar concepts in the relationship marketing over the last years (c.f. Aksoy et al., 2007; Shah and Kumar, 2008): ‘Relationship marketing is based on the main premise that customer retention is much more cost effective compared to customer acquisition’ (Aksoy et al., 2007); c.f. ‘The CRM process entails the systematic and proactive management of customer relationships as they move from beginning (initiation) to end (termination), with execution across the various customer-facing contact channels’ (Reinartz et al., 2004).

Reinartz et al. (2004) distinguish three different stages in the relationship that a customer has with a firm; initiation, maintenance, and termination stage. The initiation stage is about customer acquisition, the maintenance stage has the focus on customer retention, and the termination stage is associated with managing the ending of customer relationship. Authors found that CRM has the strongest effect on firm performance in the maintenance stage, followed by the initiation stage. Thus, maintenance activities (such as cross-selling and upselling strategies) have the strongest effect on the performance of a company.

In sum, the main goal of CRM is to broaden and deepen the scope of the relationship with the current customer, making customers stay longer involved in the relationship they have with the firm, and thus enabling the firm to increase customer retention, lower customer acquisition costs, and increase profits. But how could firms increase customer retention with CRM?

1.2 How to increase customer retention with CRM?

(7)

7 Response rates reported from cross-selling efforts are several times greater than cold sales (Kamakura, 2007). Furthermore, cross-selling leads to a broader relationship with the customer (Bolton et al., 2004) and to more knowledge of their needs and preferences, which will lead to a greater share of customer wallet, more familiarity of the customer with the firm, and in turn to even more cross-sell and upsell opportunities (Kamakura, 2007; McNally, 2007). In addition, cross-selling is likely to enlarge barriers for the customers to switch, increase both actual and mental costs of switching, which in turn will improve customer retention (Kamakura, 2007; McNally, 2007).

1.3 Problem statement

Despite the importance of both strategies to retain customers, little research has been done regarding the strategy of upselling. There is quite some literature on cross-selling, but in particular, the concept of upselling is underinvested.

There are two key issues that have received little or no attention in past literature: First of all, although the concept of cross-selling has been researched more than upselling, it is not clear what exactly or upselling is and what the difference is between both. Is cross-selling a form of upcross-selling? Or the other way around, is upcross-selling a form of cross-cross-selling? How are these concepts related? Do both concepts have an overlap, or are they totally different from each other?

Secondly, despite the realized importance to retain customers, little research has been done to identify what drives upselling. There has been quite some research about the antecedents of cross-selling, but the drivers of upselling strategy seem to be almost neglected by researchers. As well as with cross-selling, knowing what drives upselling, allows managers to increase their upselling potential and thus the profits made from upselling.

(8)

8 Because the lack of research on drivers of upselling, this paper will draw an analogy with the current literature on the drivers of cross-selling, in order to find possible drivers of upselling.

Of all the potential drivers, I am particularly interested in the influence of the customer in the success of upselling, because the customer is the central key to a successful the retention strategy. Therefore, it is of paramount importance to learn to understand the consumer. Only when we know what drives the consumer to go for a cross- or upsell proposition, a firm is able to successfully retain its customers and subsequently improve its profits.

1.4 Contribution

This paper will contribute to the literature firstly by clearly identifying the areas of both retention strategies. This paper takes a look to what exactly has been defined in the literature as cross-selling and what as upselling, and what the difference is between the both strategies.

Secondly, when both concepts are clarified, unanimous definitions of the both strategies are drawn that can be used in further research.

Thirdly, this paper makes a first attempt to find the drivers of upselling. Research, although limited, has been done on finding drivers of cross-selling. The drivers of upselling, however, have been almost neglected by researchers. This paper draws an analogy between the past research on the antecedents of cross-selling and the concept of upselling in order to find drivers of upselling.

By identifying what drives upselling, the effectiveness of upselling can be improved significantly. The contribution of this research is especially important for managers. In order to be able to improve upselling within a firm, it is crucial for managers to know what drives successful upselling and which drivers need to receive the most attention. When managers are able to (re)allocate their scarce and costly resources to the success factors of upselling better, they are able to improve their upselling and to earn profits more efficiently.

(9)

9

2. LITERATURE REVIEW

Having outlined the background and the motivation of this research, this paper continues with the literature review. The literature review begins with an elaboration about the differences in definitions of cross-selling and upselling in the literature, followed by conclusive definitions of cross-selling and upselling as they are defined for this research and as they could be used in further research. This chapter continues with an overview and elaboration of the prior research on drivers of upselling and cross-selling. From this prior research on cross-selling, an analogy between the consumer related drivers of cross-selling and upselling is drawn and hypotheses are formed. This chapter ends with a conceptual model.

2.1 Cross-selling and upselling

As mentioned earlier, the definitions of both up- and cross-selling strategies have not been distinguished unanimously throughout the literature and there is a gray area between both concepts. Cross-selling is, generally speaking, selling more products to the consumer. But throughout the literature it is not always clear if cross-selling is about selling related or non-related additional products, thus selling additional products from the same category as the initial product (within category), or from another product category (cross-category). And what exactly is upselling and how does upselling relate to cross-selling? Is upselling a form of cross-selling, or is it something totally different? And is upselling within or cross-category?

This chapter elaborates on both cross-selling and upselling strategies, trying to find answers on questions above and trying to make unanimous definitions of both concepts, so that it is clear what exactly cross-selling is and what upselling is.

2.1.1 Cross-selling

(10)

10 selling additional and related products to a customer’. According to Kamakura (2007), cross-selling is the sale of extra products related (or sometimes not related) to a product that is purchased in the past by the consumer. And Bolton et al. (2004) emphasize that cross-selling is not necessarily related to the additional purchased good: ‘Customers can add services to their portfolio that have little, if any, connection with the current consumed services’.

Kumar et al. (2008) study the drivers of cross-buying in a non-contractual setting and define cross-buying as ‘the total number of different product categories that a customer has purchased from a firm from the time of first purchase’. Reinartz et al. (2008) also define the concept as ‘the purchase of products from multiple categories’.

Considering the literature above, I define cross-selling as: The practice of selling additional product(s) that are related or unrelated to the product category of the initial product. Thus, the additional product(s) could be within category, as well as cross-category.

2.1.2 Upselling

Like cross-selling, upselling is also defined in various ways throughout the literature. Aydin and Ziya (2008) state, very broad, that upselling is offering an additional product to a customer who just made a purchase (c.f. Squires et al, 2007). The difficulty with this broad definition is, that it is hard to distinguish the difference between cross-selling and upselling, because it is not clear if this additional product is the same item as the initial product, nor if the additional product is within or cross-category. Therefore the question remains, what exactly is upselling and what is the difference between cross-selling and upselling?

Generally speaking, there are two ‘kinds’ of upselling to distinguish throughout the literature; a more quantitative form and a more qualitative form of upselling. Kim and Kim (1999) define upselling quantitatively as selling more of the same services. On the other hand, according to Wilkie et al. (1998), upselling is ‘when a sales person attempts to persuade a customer to purchase a higher priced unit’ and ‘when the consumer responds positively to an in-store promotion for the substitute “switch” brand’, which is a more qualitative definition of upselling.

(11)

11 sales of more units of the same purchased item, or the upgrading into a more expensive version of the purchased item’.

According to the definitions above, upselling is thus always within the same category, because selling more of the same products is within the category, and selling an upgrade product (which is a more expensive version of the product) is also within category.

In this paper, the quantitative form of upselling will not be investigated. It is rather easy to sell more of the same products with a marketing action. The real challenge is to persuade consumers to buy an upgraded version of a product. Therefore, the focus of this paper will be on the qualitative form of upselling; the increase of order volume by persuading a customer to buy an upgraded, or more expensive version, of the initial product that the consumer wanted to buy. This upgraded version has a higher margin than the initial product that the consumer wanted to buy, and thus is yielding more profit for the firm (and hence an increase in order volume).

2.1.3 Conclusion cross-selling versus upselling

In sum, cross-selling is to sell additional product(s) that are either related or unrelated (thus within category or cross-category) to the product category of the initial product that the consumer wanted to buy or has bought in the past. This related or unrelated product is per definition not exactly the same product as the product that the consumer initially wanted to buy (as is the case with qualitative upselling).

Upselling can be defined in a more quantitative or qualitative manner. Quantitative upselling is to sell more of same products. Thus, buying more of the same product is upselling and not cross-selling. For example, if one buys two exactly the same items (e.g. ‘second product 50% sale’ actions), it is quantitative upselling. But, if one buys an additional product from the same product category (but not exactly the same product), then it is (related) cross-selling (for example strawberries and apples, which is both from the fruit category, but not the same product).

(12)

12 than he/she wanted to buy (thus, a television of Philips with a bigger screen and more options on it for example).

Because upselling is about selling more of the same product or an upgrade version of a product, upselling is always in the same product category, and thus per definition always within category, unlike cross-selling, which can be within category and cross-category.

2.2 What drives upselling and cross-selling?

As mentioned before, almost no research has been done to identify drivers of upselling. The only exceptions are the work of Wiesman (2006) and the subsequent work of Squires et al. (2007). In order to find additional drivers of upselling, I will also examine the drivers of cross-selling, which have been researched more extensively. By drawing an analogy between cross-selling and upselling, potential drivers of upselling can be identified. Before a light will be shed on the drivers of cross-selling in subparagraph 2.2.2, I will first elaborate on the past research on drivers of upselling in the following subparagraph.

2.2.1 Drivers of upselling

Only two studies attempted to find drivers of upselling (i.e. Wiesman, 2006; Squires et al., 2007). Both studies researched drivers of upselling from the perspective of employees; which adjustments employees or managers could make in order to increase upselling among customers.

In his study of 2006, Wiesman analyzed the effect of performance feedback and social reinforcement (by managers to their employees) on up-selling behavior of employees of a fast food restaurant. The conclusion was that both factors were positively related to the performance of the employees with respect to improving the percentage of up-selling. Furthermore, the variability in performance decreased, making the performance of the employees more stable.

(13)

13 So, upselling can be improved by a better performance of employees who are stimulated by performance feedback, social reinforcement, task clarification, and visual prompt.

Although a step in the right direction, the drivers of upselling are under invested. In order to find more (potential) drivers of upselling, the past literature on drivers of cross-selling has been examined. By researching the drivers of cross-selling, an analogy between drivers of cross-selling and upselling can be drawn and potential drivers of upselling can be identified.

2.2.2 Past literature on drivers of cross-selling

Within the literature on drivers of cross-selling, a distinction can be made between; internal drivers (drivers that a firm is able to manage directly) and the external drivers (drivers that cannot directly be influenced by the firm). The internal drivers can be further subdivided into corporate related and marketing related drivers, and the external drivers can be divided into competitor related and consumer related drivers (see Table 2.1 for an overview of all literature on the drivers of cross-selling).

TABLE 2.1 – DRIVERS CROSS-SELLING

Drivers researched Cross-selling Upselling

INTERNAL DRIVERS

Corporate

Corporate reputations Jeng (2011) Interpersonal relationships Jeng (2011)

Marketing

Marketing instruments/efforts Verhoef et al. (2001) Bolton et al. (2004) Kumar et al. (2008) Loyalty programs Verhoef (2003)

Direct mailings Verhoef (2003)

Acquisition channels Verhoef and Donkers (2005)

EXTERNAL DRIVERS

Competitor

Competitor’s price Jeng (2011) Competitor’s product variety Jeng (2011)

Consumer

Payment equity / customers price perceptions

Verhoef et al. (2001) Verhoef (2003) Bolton et al. (2004) Satisfaction Verhoef et al. (2001)

Verhoef (2003) Bolton et al. (2004) Affective commitment Verhoef (2003)

Bolton et al. (2004)

Image conflicts Ngobo (2004)

(14)

14 Ratio of product returns Kumar et al. (2008)

Focused buying Kumar et al. (2008)

Customer characteristics Kumar et al. (2008) Category of first purchase Kumar et al. (2008)

Researching all the drivers of cross-selling stated in the table above in relationship with upselling, would be a great contribution to the literature of upselling, but also an immense job to do and far beyond the scope of this paper. In order to carry out a good research, from which well-founded conclusions can be drawn, this paper will focus on the customer related drivers.

As said, cross-selling and upselling are the key components of CRM (McNally, 2007) and these strategies are very important strategies for customer retention (Kamakura, 2007). Through these strategies, firms try to sell additional products or services and thereby broaden and/or deepen the relationship with the customers, making customers stay longer involved in the relationship they have with the firm (Reinartz et al, 2004). So, with customer retention, the customer is the central (key) component and one of the most important factors in the success of both strategies. When the consumer is not attracted to a cross-selling or upselling proposition, for any reason whatsoever, the attempt to retain customers will hopelessly fail. Therefore, it is of paramount importance to that we (firstly) learn to understand the consumer and to find what drives the consumer to cross- or upbuy a product, in order for a firm to be successful in retaining customers. Because of this customer centricity and the realization of the importance to understand the consumer, this paper will focus on the consumer related drivers of upselling.

In the following paragraph you will find an elaboration on the past literature about the consumer related drivers of cross-selling, the suggested relationship of these drivers with upselling, and the derived hypotheses. In addition to the drivers found in the cross-selling literature, I expect impulse buying tendency to also be a driver of upselling. The reasoning behind this, as well as the suggested relationship and the derived hypothesis is explained in the last subparagraph of this paragraph.

2.3 Consumer related drivers

(15)

15 convenience (Ngobo, 2004), average inter-purchase time, ratio of product returns, focused buying, household income, age, and category of first purchase (Kumar et al., 2008). These consumer related drivers will be explained in more detail in the subparagraphs below. In addition to the consumer related drivers found in the literature of cross-selling, impulse buying tendency will also be explained as potential driver of upselling. At first the drivers will be individually explained, followed by the conclusions of the research of those drivers, and ending with the hypotheses I derive from the past research.

2.3.1 Payment equity or price perception

Payment equity is the perceived fairness of the price that the consumer paid for a product of service of a particular firm and is strongly related to the price perception of consumers (Verhoef, 2003). Payment equity is affected by a firm’s own pricing policy, the competitor’s pricing policy, and the quality of the services or products that the company offers. When the quality and/or price of the service or product of the focal firm is better in the perception of the consumer, the payment equity of the focal firm will be higher, then the competitors.

Effect on cross-selling. Verhoef et al. (2001) researched the effect of payment equity on cross-buying. Contrary to their hypothesis, their results show that perceived equity does not affect cross-buying directly. However, their study showed that companies that are perceived as offering low prices (in comparison with the competition) are more likely to cross-sell. In line with these results, in a subsequent study Verhoef (2003) finds also no main effect of payment equity on cross-buying in a later study.

In contrary to the research above, Bolton et al. (2004) find a positive effect of customers price perceptions on the breadth of the customer relationship (the number of different products of services a customer buys from the business, in other words, the degree of cross-buying), when the pricing policy is consistent with the previous bought services.

(16)

16 likely to succeed if the firm offers noteworthy benefits to the customer, has the right qualities, and when these qualities are noticed by the customers.

Effect on upselling. Although the evidence is weak, it seems rather logical that the perceived price (payment equity) is very important to a consumer, when considering buying a more expensive upsell product. When a consumer perceives the firm as expensive, and thus when the consumer is convinced that he/she is able to get the same product somewhere else (far) cheaper, this will most likely influence the upselling intention negatively. Therefore, I hypothesize;

H1: The more positive payment equity, the higher the upselling intention.

2.3.2 Satisfaction

Whereas the consumer price perceptions are reflected in payment equity, the quality perceptions of product/services are reflected in satisfaction of the consumer (Verhoef et al., 2001). Satisfaction is ‘a customer’s cumulative evaluation of the purchase and consumption experience’ (Bolton et al., 2004); the emotional state that occurs as a result of a customer’s interactions with the firm over time (Verhoef, 2003).

Effect on cross-selling. In addition to payment equity, Verhoef et al. (2001) also researched the effect of satisfaction with focal supplier on cross-buying. Contrary to their hypothesis, their initial results show that satisfaction does not affect cross-buying directly. Realizing that their dependent variable, cross-selling, consists of positive and negative data (consumers who did cross-buy more, and consumers who cross-buy less), authors decided to reanalyze the data with two different groups; the group of people who bought more, and a group who bought less. Subsequently, they found support for the positive effect of satisfaction on cross-buying, when the relationship is longer. In a subsequent study Verhoef (2003) finds no main effect of satisfaction on cross-buying.

Bolton et al. (2004) find, contrary to Verhoef et al. (2001) and Verhoef (2003), that customer satisfaction has a positive effect on cross-buying, but only if customers perceive the services of the firm to be similar in attributes or benefits.

(17)

17 Effect on upselling. Again, as with perceived value, the evidence is weak, but it seems rather logical that the satisfaction with the company is very important to a consumer, when considering buying a more expensive upsell product. When a consumer is dissatisfied with the firm’s services or offerings, he/she will go and buy the product somewhere else. Of course, the latter depends also on possible lock-in effects like switching costs. Regarding satisfaction, I hypothesize;

H2: The more satisfaction of the consumer, the higher the upselling intention.

2.3.3 Affective commitment

Affective commitment is a desire of the consumer to stay in a relationship with the firm (Bolton et al., 2004). Verhoef (2003) describes affective commitment as ‘the psychological attachment of consumers, based on loyalty and affiliation, to the firm’.

Effect on cross-selling. Verhoef (2003) explored the effects of customer relationship perceptions and relationship marketing instruments on customer retention and customer share development. Of the customer relationship perceptions concepts, affective commitment positively influences customer share development and retention, confirming that commitment is a significant variable in relationship development. This is in line with the results of Bolton et al. (2004), who also state that affective commitment will have a positive relationship with cross-buying. In addition, Ngobo (2004) concludes in his article that consumer’s loyalty to the firm does reduce the attractiveness of the offers of the competition, but the influence is weak.

Effect on upselling. Reflecting this to upselling, I expect that affective commitment will have the same positive effect on upselling as it has on cross-selling, because when a consumer has a desire to keep a relationship with the firm, he/she feels connected to the firm and is more likely to buy an upsell product of the firm, then when he/she is not affective committed to the firm. Consequently, I hypothesize;

H3: The more affective commitment, the higher the upselling intention.

2.3.4 Image conflicts

(18)

18 conflicts could appear when the added product or service does not fit with the firm’s past assortment. Image conflicts are often the result of blending different service management systems, like different service concepts, delivery systems, customers, and employee behaviors, where consumers will base their association with the firm on, and is also often the case with unrelated diversification (Ngobo, 2004). For example, when C&A (mostly perceived as low budget clothes) starts selling designer handmade clothes (which is mostly perceived as expensive and not C&A-like). When this happens, a consumer could get image conflicts about the abilities of the firm to deliver this high quality service.

Effect on cross-buying. Research about image conflicts shows that image conflicts about the abilities of the firm to deliver high quality services is strongly negatively related to the customer’s cross-buying intentions (Ngobo, 2004).

Effect on upselling. Looking at upselling, image conflicts could also certainly have a negative influence on upselling. When a consumer is not convinced of the abilities of a firm to sell/deliver an upgrade product, thus an upsell product, the consumer will be resistant to buying the upsell product, and this will negatively influence the consumer’s upsell intention. However, upselling is always within category, thus the upselling product is just slightly different from the initial product. Therefore, I expect that image conflicts will be far less likely to appear with upselling, as with cross-selling. But when image conflict does appear, I hypothesize;

H4: The more image conflicts about the firm’s capabilities, the lower the upselling intention.

2.3.5 Perceived convenience

Perceived convenience is the ease for consumers to do their all their shopping at one place. Consumers nowadays are often busy and in a hurry to do shopping. Therefore, they will economize their time and try to save effort by buying at one store.

(19)

19 Effect on upselling. Regarding the perceived convenience of a one-stop-provider, I hypothesize the same effect for upselling as for cross-buying, because a consumer could certainly be more attracted to an upselling proposition, when he does not have to look further for the same offer (or a good substitute). In other words, I think it is important that the consumer has the idea that he is able to choose from a wide range of products (and thus a wide range of upselling products) and that he subsequently has the feeling that does not need to search further for a better (upselling) offer at the competition. I expect that when a consumer perceives that he is able to choose from a wide range of upselling propositions, the upselling intention increases, and therefore, the convenience of a one-stop-provider improves upselling.

H5: The more perceived convenience of buying an upsell product from the same provider as the initial product, the higher the upselling intention.

2.3.6 Exchange characteristics

In their study of 2008, Kumar et al. identified drivers of cross-buying by identifying variables obtainable from a database of a large catalogue retailer with seven different product categories. These variables, which act as drivers of cross-buying, can be used to select and influence potential profitable customers. They examined different consumer characteristics as drivers of cross-selling. Exchange characteristics categorized by Kumar et al. (2008) are the average inter-purchase time (the average time in months between the purchases with the catalog retailer), the ratio of product returns (how often a consumer returns a product), and the degree of focused buying (or the depth of the purchase) which was measured as the total number of orders placed in a specific category.

Effect on cross-buying. With respect to the exchange characteristics, Kumar et al. (2008) found that average inter-purchase time, has an inverted U relationship with cross-buying. Thus, as the time between purchases increases, the chance of cross-buying increases as well, up to a certain point, where the chance of cross-buying decreases again. In other words, customers who buy at intermediate intervals stay longer with the firm and make more frequent purchases, which translate in a lower perceived risk and a higher amount of cross-buying.

(20)

20 Because, when consumers buy very frequently at a store (thus a low average inter-purchase time), they also have to spend frequently money on their products and will be therefore less alluring to buy an upsell product (which is more expensive). On the other hand, when the time between purchases is very long, thus a high average inter-purchase time, the customer is not used to buying at the store, which could result in unfamiliarity with the firm and a higher perceived risk when buying at the firm, and thus a lower chance of upselling.

But, when a consumer buys at an intermediate frequency at the store, the consumer knows the firm, is able to spend money on a product that needs to last for a while, and thus would be more likely to be seduced to buy upgrade version of the initial product/service he/she wanted to buy. So then the upselling intention will be the highest. Therefore, I hypothesize;

H6: The more average inter-purchase time between purchases, the higher the upselling, up to a certain point, where the upselling intention will be lower as the average inter-purchase time increases (inverted U-shape).

Regarding the ratio of product returns, I also expect an inverted U-shaped relationship with upselling, because when one returns a product, the product does not fulfill the needs that the customer is facing, therefore the temptation to buy an upgrade version of the same product would be high, and thus the chance of upselling bigger. But, on the other hand, if one needs to return its products too often, it will decrease the customer’s satisfaction and quality perceptions with the products and the firm, and the customer will be, therefore, less likely to buy an upsell product at this firm. Hence, I hypothesize;

H7: The more product returns, the higher the upselling intention, up to a certain point, where the upselling intention will be lower as the ratio of product returns increases (inverted U-shape).

(21)

21 specific category during a certain period will be lower, when upselling increases. Therefore, I hypothesize;

H8: The more focused buying, the lower the upselling intention.

2.3.7 Product characteristics

The third category drivers distinguished by Kumar et al. (2008) are the product characteristics, which were measured with the category of the first purchase of the consumer at the firm.

Effect on cross-buying. Kumar et al. (2008) show that the category of the first purchase is strongly related with cross-buying.

Effect on upselling. Kumar et al. (2008) concluded that the category of first purchase has a significant relationship with cross-buying. Unfortunately, he lacked to give an explanation for this relationship. With respect to the subject of upselling, I expect no relationship between the category of first purchase and upselling intentions. Within every category there are upsell products, and I cannot think of a reason why this should be different between categories.

In addition to the drivers above found in the cross-selling literature, I expect impulse buying tendency also to be a strong predictor of upselling. In the following subparagraph I will elaborate what impulse buying tendency is, why I expect that impulse buying tendency could be a driver of upselling, and which relationship I expect.

2.3.8 Impulse buying tendency

According to Rook (1987) impulse buying is often defined as unplanned purchases. Impulse buying is ‘not consciously planned, but arises immediately upon confrontation with a certain stimulus. Once triggered, an impulse encourages immediate action’ (Rook 1987). Furthermore, he states that these purchase decisions are made after the consumer enters a retail environment. In his article, Rook (1987) defines impulse buying as; ‘Impulse buying occurs when a consumer experiences a sudden, often powerful and persistent urge to buy something immediately…’.

(22)

22 rational, is perceived more bad than good, and is likely to come with an out-of-control feeling Rook (1987).

According to Jones et al. (2003), impulse buying tendency is ‘the degree to which an individual is likely to make unintended, immediate and unreflective purchases (i.e. impulse purchases)’. Unintended, because the decision is made while shopping, unreflective because the purchase is done without an extended decision making process, and immediate because there is little time between seeing the product and buying it.

Both researchers are convinced that impulse buying tendency varies among individuals (Jones et al., 2003) and that it is a consumer personality trait (Rook, 1987), and that therefore, individuals could be segmented based on their impulse buying trait.

Looking more closely to the concept of upselling, upselling is in fact to take distance of your initial decision to buy a product or service, when you are confronted with an upgrade version of the product that you initially wanted to buy. You are confronted with this upgrade product on the last moment before buying the initial product, when standing in the store, for example. The decision to buy this upgrade product is thus on the last moment and kind of impulsively. Upselling has thus, very much the same characteristics as impulse buying. It is also not consciously planned and thus unintended, because the upsell decision is made in the retail environment while shopping, after confronted with the upselling stimulus. Furthermore, it also disrupts the consumer’s behavior stream, is spontaneous and a fast experience, because of the little time between seeing, deciding and purchasing (= short decision making process).

Because the tendency of consumers to buy impulsively can be different per person, I expect that individuals with a high impulsive buying tendency will be more likely to be seduced to buy an upsell product, than individuals with a low impulsive buying tendency. Hence, I hypothesize;

(23)

23 H1 + H2 + H3 + H4 – H5 + H6 Inv. U H7 Inv. U H8 – H9 + 2.4 Conceptual model

Paragraph 2.3 elaborated on the present research about the drivers of cross-selling. By drawing an analogy with the research on the drivers of cross-selling, hypotheses are formed. These propositions are summarized and visualized in a conceptual model (see Figure 2.2).

FIGURE 2.2 – CONCEPTUAL MODEL

Upselling on the right side of the model is the dependent variable. The nine concepts on the left side of the model depict different potential drivers of upselling, derived from the literature on cross-selling. Payment equity, satisfaction, and affective commitment will have a positive influence on upselling, whereas image conflicts about the abilities of the firm to deliver a higher quality product or service and focused buying will be negatively related to upselling. Perceived convenience of a one-stop-provider is proposed to have a positive effect on upselling. Furthermore, the average inter-purchase time and the ratio of product returns is expected to have an inverted U-shape relationship with upselling, meaning that the relationship is positive, up to a certain point, where it becomes negative.

Moreover, in addition to the potential drivers that are found in the cross-selling literature, the impulse buying tendency of customers is predicted to have a positive relationship with upselling. Payment equity Satisfaction Affective commitment

UPSELLING

Image conflicts Perceived convenience Average inter-purchase time

(24)

24

3. RESEARCH DESIGN

The main purpose of this paper is to identify drivers of upselling. Using the literature on drivers of cross-selling, some hypotheses regarding potential drivers of upselling where proposed in chapter 2. This chapter begins with describing which research design has been used in this study and how the variables are operationalized. It proceeds with how the data is collected, it elaborates on the population and sample, and explains the procedure of the research. This chapter ends with the plan of analysis to analyze the data gathered.

3.1 Experimental research design

From the literature review, nine independent variables are extracted to be potential drivers of the dependent variable upselling. To research these potential drivers of upselling, quantitative research has been done, by using an extension of a true experimental research design: a factorial design (Blumberg, 2008, p. 413).

3.1.1 Factorial design

A factor is commonly used to refer to an independent variable and factors could be divided into treatment levels, which represent various subgroups (Blumberg, 2008, p.410). This research had two independent variables or factors (satisfaction and payment equity), with two treatment levels each (positive or negative, and low or high), that were manipulated and completely crossed. Consequently, four different conditions emerged 2 (satisfaction: positive vs. negative) x 2 (payment equity: high vs. low). The experimental design is visualized in Figure 3.1.

FIGURE 3.1 – EXPERIMENTAL RESEARCH DESIGN

(25)

25 The reason of using an experiment is because, it ‘allows a researcher to alter systematically the variables of interest and observe what changes follow’ (Blumberg, 2008, p.395). By using an experimental research, causal relationships can be determined, because the respondents are manipulated to experience satisfaction (positive or negative) and payment equity (high or low). In order to assure that participants really perceive the manipulation in the correct way, a pretest has been done to rule out that participants do not experience the same situated conditions in a different way.

3.1.2 Factorial survey

The method of experimental research design was a (factorial) consumer survey (c.f. Verhoef et al., 2001; Verhoef 2003; Ngobo 2004). Implementing experimental design into a survey is called a factorial survey, also known as vignette research (Blumberg, 2008, p.414). In a factorial survey, the respondents are confronted with a hypothetical situation and are subsequently asked to assess the situation.

Two of the independent variables (satisfaction and payment equity) were manipulated in a sketched story in the beginning of the questionnaire. Thus, four different surveys were developed, one for every experimental condition (see also Figure 3.1). A quarter of the respondents received a questionnaire with high payment equity and a positive satisfaction with the focal company. Another quarter received a questionnaire with also high payment equity, but a negative satisfaction. The third and the fourth quarter of respondents received a questionnaire with low payment equity, the former in combination with a positive satisfaction and the latter with a negative satisfaction.

The reason to manipulate these two variables, and thus to use a hypothetical situation in the survey, is that this makes is easier to find respondents. Otherwise, if this variables were not manipulated, it would be an almost impossible task (if you do not have access to a database of a company) to find respondents that recently bought an upsell product, in the same category, who experienced positive and negative satisfaction, and high and low payment equity.

(26)

26 Due to the choice of this research design, the variables average inter-purchase time, ratio of product returns, and focused buying could not be measured with this research design, because they must be measured on base of existing sales data. Kumar et al. (2008) measured the average inter-purchase time by calculating the average time in months between the orders in January and the previous purchase. The ratio of product returns was calculated by the total dollar amount of the products returned during January until the previous purchase, divided by the total dollar amount of orders placed in that period. In addition, focused buying was measured by the total number of orders placed in a specific category. Because of the decision to use a hypothetical situation in the questionnaire, these variables could not be measured.

Control variables

The variables household income, gender and age where used as control variables. Household income is measured by asking the respondents what their average household income is. This involves the average amount income of the whole family, per (calendar) month after tax. In other words, the nominal amounts of income of the whole family they see appear on their bank account every month. The answer possibilities were 1 = “Less than modal”, 2 = “Between modal and 2x modal”, 3 = “Between 2x modal and 3x modal”, 4 = “Between 3x modal and 4x modal”, 5 = “4x modal or more”. Because, it may not be known to everyone what modal exactly refers to, the following explanation is given with the question.

The average income is €32.500 (before tax) per year (inclusive holiday surcharge of 8%)1. This is about €2500 (before tax) per month or about €580 (before tax) per week (exclusive holiday surcharge). After tax a modal income it is about €1800 per month and about €420 per week.

(27)

27 3.2 Operationalization variables

In order to measure the potential drivers of upselling, research has been done to develop proper statements and scales to measure the dependent and independent variables. By using scales from existing literature and adapting these slightly to the current research, where possible, the most appropriate scales emerged and were used in the survey.

Although most of the statements in the questionnaire were based on statements developed by researchers in existing literature, a reliability analysis of the scales was done after the survey to measure if the statements that were used to measure a particular independent variable, consistently reflect the construct that it is measuring (Field, 2009, p.673). ‘Reliability (analysis) is concerned with estimates of the degree to which a measurement is free of random or unstable order’ (Blumberg, 2008, p.455). In other words, the reliability of a scale is the degree to which the scale provides the same answers on a different point in time, all other things being equal.

The most common way to measure the reliability of a scale (Field, 2009, p.674), and the best way for multi-item scales at the interval level of measurement (Blumberg, 2008, p. 458), is the Cronbach’s alpha (denoted by α). The Cronbach’s alpha coefficient ranges from 0 to 1; a coefficient of 0 means an absolute unreliable scale, whereas a coefficient of 1 means a perfectly reliable scale. In order to state that a scale is reliable (internal consistent), and that it can be merged into one scale, the value of the alpha must be higher that .6 (Malhotra, 2010, p.319).

Note that the dependent variable upselling is measured with only one statement and therefore cannot be tested on its reliability with the Cronbach’s alpha.

3.2.1 Dependent variable ‘upselling’

(28)

28 Squires et al. (2007) and Wiesman (2006) performed a true experiment using an existing restaurant setting and its employees, so they could measure the products being actually sold. In the present study, it is not possible to conceptualize upselling in the form of the amount of sold products. Hence, participants’ buying intentions are asked instead of analyzing their actual buying behavior (cf. Ngobo, 2004). Respondents are asked how likely they are to would buy the upsell product, in this case the Philips 800HDI (“How likely is it that you will buy the Philips 800HDI, instead of the Philips 500HDI?” (0% = “Not likely at all”, to 100% = “Very likely”)).

3.2.2 Operationalization independent variables

Payment equity or price perception

To measure payment equity Verhoef et al. (2001) and Verhoef (2003) asked respondents to rate how satisfied they were about the insurance premium (1 = “Very dissatisfied” to 5 = “Very satisfied”), and what they thought of the price premium they paid (“Too high”, “High”, “Normal”, “Low”, “Too low”).

Because of the decision to use a hypothetical situation in the survey, payment equity of the focal company cannot be measured, for the simple reason that respondents do not have an opinion or previous experience with the company. Therefore, this independent variable cannot be measured with statements (c.f. Verhoef et al., 2001; Verhoef, 2003). Hence, payment equity is manipulated in the questionnaire. Respondents received a questionnaire with a high or low payment equity (prices of the focal company and one competitor were mentioned in the sketched situation) (see also 3.5 Procedure).

Satisfaction with focal supplier

Verhoef et al. (2001) and Verhoef (2003) also used scales to rate the satisfaction of the consumer with the focal supplier. Questions were asked about the personal attention of the salesman, the willingness of the salesman to explain procedures, the service quality, the response to claims, the expertise of employees, the respondents relationship, and the alertness of the company (1 = “Very dissatisfied” to 5 = “Very satisfied”).

(29)

29 Therefore, the independent variable satisfaction is manipulated in the questionnaire as well. Respondents received a questionnaire with a positive or negative past experience (and thus a positive or negative satisfaction) (see also 3.5 Procedure).

Affective commitment

Affective commitment of the participants is measured c.f. Verhoef (2003) using the statements2; “My loyalty to a firm determines where I buy my products”, “My strong attachment to a firm determines where I buy my products”, and “My strong sense of belongingness to a firm determines where I buy my products” (1 = “Strongly disagree” to 5 = “Strongly agree”).

The overall reliability of the scale to measure affective commitment is quite good (Cronbach’s α = .877) (see Table 3.2 for a total overview of the statements and the corresponding Cronbach’s alphas). All three statements used to measure affective commitment have high reliabilities. If one of the statements will be deleted, the Cronbach’s α will be lower than the overall Cronbach’s α, therefore these three statements consistently measure the same underlying construct and could be combined into one variable in the analysis.

Image conflicts

Image conflict is measured c.f. Ngobo (2004), although adapted to the present research setting, with the statements; “I trust that Store A will be able to deliver the Philips 800HDI” (1 = “Strongly disagree” to 5 = “Strongly agree”), “When I don’t trust a store to deliver a product, I will certainly not buy my product at this firm” (1 = “Strongly disagree” to 5 = “Strongly agree”), and “How important is your trust in the firm to deliver the product, when buying it?” (1 = “Very unimportant” to 5 = “Very important”).

The first statement is phrased reversed, because when one does trust store A (and thus, answers with strongly agree or agree) he/she is not experiencing image conflicts. Whereas the respondents that answer the second and the third statement with (strongly) agree or (very) important, does experience image conflicts. Therefore, the results of the questionnaire are adjusted for analysis.

The overall reliability of the scale is very unreliable (Cronbach’s α = .371), but deleting the first statement will increase the Cronbach’s α to .581. The explanation is quite logical. The

2

(30)

30 first statement (“I trust that Store A will be able to deliver the Philips 800HDI”) is state-dependent and is very likely to be answered differently by the different satisfaction groups. The positive satisfaction group has more satisfaction about the store and has most likely more trust in Store A and, therefore, will rate this statement higher than the negative satisfaction group.

In conclusion, for further analysis, the first statement will be deleted. This way, the Cronbach’s α rises to .581, which is officially not reliable. However, because the value is so close to .6, I consider this scale reliable enough.

Perceived convenience

The perceived convenience is measured with the statements; “I find it important that a store has a large assortment of televisions, when I buy a television” and “When a store has a large assortment of televisions, I am less inclined to browse at other stores” (1 = “Strongly disagree” to 5 = “Strongly agree”) (based on Ngobo, 2004).

The reliability analysis shows that these two statements are not reliable (Cronbach’s α = .269) and, thus, do not consistently reflect the construct that it is measuring and cannot be combined into one variable.

Recall that perceived convenience is the ease for consumers to do their all their shopping at one place, because shopping at one provider gives customers one-stop-benefits, namely time and effort savings (Ngobo, 2004). Perceived convenience has in essence more to do with the breadth of an assortment, then with the tendency of consumers to browse at other stores. Therefore, I think that the first statement (“I find it important that a store has a large assortment of televisions, when I buy a television”) measures perceived conveniences the best. Hence, this statement will be used to measure perceived convenience in further analyses.

Impulse buying tendency

(31)

31 which appeared to be more thorough than previous scales and more predictive of nature, and therefore, will be used in this research3.

This Impulse Buying Tendency Scale consists of 5 statements; “When I go shopping, I buy things that I had not intended to purchase” and “I am a person who makes unplanned purchases” (1 = “Very rarely” to 5 = “Very often”), and “When I see something that really interests me, I buy it without considering the consequences”, “It is fun to buy spontaneously”, “I avoid buying things that are not on my shopping list” (1 = “Strongly disagree” to 5 = “Strongly agree”).

The higher the scores on the first four statements, the more the respondent has a tendency to buy impulsively. The last statement, I avoid buying things that are not on my shopping list, is, in contrary to the other statements, stated the other way around (the higher the score, the lower the impulse buying tendency). Therefore, the answers to this statement are converted in the database.

As Weun et al. (1998) already showed in their research, the Impulse Buying Tendency Scale appears also in this research to be a very reliable scale (Cronbach’s α = .790). All five statements used to measure impulse buying tendency have high reliabilities and the overall Cronbach’s α will drop if one of the statements will be deleted. Therefore, all statements consistently measure the same underlying construct and are combined into one variable in the analysis.

On the next page you find an overview of the reliability analyses of the independent variables that were measured with multiple items. In the first column are the independent variable and its items stated. In the column ‘Cronbach’s α overall scale’ are the Cronbach’s alpha values over all the statements used in the questionnaire to measure the particular independent variable. In the latest column are the recalculated Cronbach’s alpha values stated (if possible) when one item was deleted in order to obtain the highest Cronbach’s alpha value.

3

(32)

32 TABLE 3.2 – PURIFYING SCALES

Scales Cronbach’s α overall scale (all statements) Cronbach’s α if one item deleted Affective commitment

- My loyalty to a firm determines where I buy my products.

- My strong attachment to a firm determines where I buy my products. - My strong sense of belongingness to a firm determines where I buy my products.

.877

Perceived convenience

- I find it important that a store has a large assortment of televisions, when I buy a television.

- When a store has a large assortment of televisions, I am less inclined to browse at other stores.*

.269 **

Image conflict

- I trust that Store A will be able to deliver the Philips 800HDI.*

- When I don’t trust a store to deliver a product, I will certainly not buy my product at this firm.

- How important is your trust in the firm to deliver the product, when buying it?

.371 .581

Impulse buying tendency

- When I go shopping, I buy things that I had not intended to purchase. - I am a person who makes unplanned purchases.

- When I see something that really interests me, I buy it without considering the consequences.

- It is fun to buy spontaneously.

- I avoid buying things that are not on my shopping list.

.790

* This statement is deleted in the analysis, because of the low reliability of the scale. ** Calculating the Cronbach’s alpha over one item is not possible.

3.3 Data collection

Before the factorial survey was distributed to the respondents, a pretest was necessary to verify if the intended manipulation works. A manipulation check was necessary in order to assure that participants in one experimental condition experience the manipulation all the same way (thus, for example, that the participants in the positive satisfaction condition, all experience a positive satisfaction). In order to measure the manipulation, a pretest was conducted (see Appendix A - Pretest).

3.3.1 Pretest

Four different pretest surveys were developed; a positive satisfaction situation, a negative satisfaction situation, a high payment equity situation, and a low payment equity situation.

Four samples of 10 persons were asked to read the sketched story and were subsequently asked to rate several statements about how they perceived the manipulations regarding satisfaction or price equity. These participants were not included the main survey.

(33)

33 The first sample received a questionnaire with a positive satisfaction situation and the second sample group received a questionnaire with a negative satisfaction situation. The past experiences with the company (good or bad, which lead to positive or negative satisfaction) were imbedded in the sketched situation. After reading the sketched story, the satisfaction of the respondents was measured with the statements; “How satisfied are you about the personal attention of Shop A?”, “How satisfied are you about the alertness of Shop A?”, “How satisfied are you about the service quality of Shop A?”, and “How satisfied are you overall about Shop A?” (1 = “Very dissatisfied” to 5 = “Very satisfied”).

The third and fourth sample group received a questionnaire with a high and a low payment equity situation, respectively. Prices of the focal supplier and one competitor supplier were incorporated in the sketched story. The perceived payment equity was measured with the statements; “What do you think of the prices of Shop A compared to Shop B?” (1 = “Very expensive” to 5 = “Very cheap”), “How satisfied are you about the prices of Shop A?” (1 = “Very dissatisfied” to 5 = “Very satisfied”), “How important is the price of the competitors to you, when buying this television?” and “How important is the satisfaction with the firm to you, when making the decision to buy the television?” (1 = “Very unimportant” to 5 = “Very important”).

3.3.2 Manipulation check

In order to test if the intended manipulation indeed worked, two independent samples T-tests were conducted between the two satisfaction groups and between the two payment equity groups.

Satisfaction

(34)

34 TABLE 3.3 – MEAN SCORES AND SIGNIFICANCE OF SATISFACTION GROUP

Mean positive satisfaction

Mean negative

satisfaction Sig. (2-tailed)

How satisfied are you about the

personal attention of Shop A? 4,4 1,6 .000

alertness of Shop A? 4,2 2,2 .001

service quality of Shop A? 4,5 2,3 .000

How satisfied are you overall about

Shop A? 4,5 2,0 .000

TOTAL 4,4 2,0

Payment equity

Just like the satisfaction manipulation, the price equity manipulation also significantly worked (p = < .031). Respondents of the high price equity manipulation indeed experienced the focal store to be relatively fairly priced in comparison with the competition (so a high price equity) (average mean score of 3,9). Furthermore, the respondents in the low price equity situation indeed experienced the focal company as relatively expensive (average score mean 2,0) (see Table 3.4).

TABLE 3.4 – MEAN SCORES AND SIGNIFICANCE OF PRICE EQUITY GROUP

Mean high price equity

Mean low price

equity Sig. (2-tailed)

What do you think of the prices of

Shop A compared to Shop B? 4,3 1,6 .000

How important is the price of the competitors to you, when buying this television?

4,1 2,2 .002

prices of Shop A? 3,2 2,0 .031

How important is the satisfaction with the firm to you, when making the decision to buy the television?

4,0 2,0 .000

TOTAL 3,9 2,0

(35)

35 3.4 Population and sample

The survey was conducted in a period of 3 months in October, November and December of 2011, in The Netherlands. The questionnaires were self-completed questionnaires, distributed via the website Thesistools.com, via the email accounts of students at the Rijksuniversiteit Groningen, and via links in e-mails and social media.

One sample of 200 Dutch respondents was collected for this survey. Participants were randomly assigned to one of the four experimental conditions, until the sample size of 50 respondents per experimental condition were met.

Representativeness of the sample

The total sample size that has been used for this study is based on the average respondents of previous studies on drivers of cross-buying and impulse buying tendency.

According to Field (2009, p.222) about 10 to 15 respondents are necessary for each independent variable in order to obtain a representative sample for the regression analysis. This research uses 9 different independent variables (including the control variables) in the regression analysis, so around 90 to 135 respondents are required in order to get a reliable regression analysis. In addition, he mentions that there are 50 + 8k (k = number of predictors) respondents necessary to adequately measure the overall fit of the regression model and about 104 + 5k cases of data are needed to obtain trustworthy estimates of the individual predictors in the regression analysis (Field, 2009, p.222). In the light of this research, more than 122 (50 + 8 x 6) or 149 (104 + 5 x 6) respondents are needed to be able to adequately measure the overall fit and attain reliable estimates of the independent variables. To conclude, this research has a very sufficient sample size (200 respondents), making this sample very representative, so conclusions can be drawn for the whole population.

Validity of the sample

In the total sample size there were slightly more females (94 respondents, 47%) represented than males (106 respondents, 53%) and the ages of the respondents ranged from 16 to 99 years old.

(36)

Chi-36 Square value = 14,415, p = > .275). In other words, the average household income of the four experimental groups is not significantly different between the groups.

Furthermore, the differences between the number of men and woman in the groups were very small. The difference in group 4 was the largest (18 male and 32 female respondents). A Chi-Square test shows that there is no dependency between gender and the four experimental groups (Pearson Chi-Square value = 5.54, p = > .136). Therefore, the expected number of males (or females) does not differ significantly to the actual number of males (and females) in the experimental groups, so males are not more likely to be in one of the four groups, than females.

In addition, the average age of the respondents in the four groups was very similar, around 30 years old. A One-Way ANOVA test shows that the variance in the means of the four groups do not significantly differ from each other (F = .341, p = .795). So groups are equal in sampling variation, what basically means that the groups do not significantly differ from each other.

TABLE 3.5 - AGE AND GENDER DISTRIBUTION PER GROUP

Gender

Age Household

income*

Male Female Minimum Maximum Mean Average

Group 1

Positive satisfaction, high payment equity

56% 44% 21 57 31 1.68

Group 2

Negative satisfaction, high payment equity

36% 64% 22 56 29 1.58

Group 3

Positive satisfaction, low payment equity

54% 46% 19 64 29 1.94

Group 4

Negative satisfaction, low payment equity

42% 58% 16 99 30 1.74

* Where 1 represents an income between modal and 2x modal, and 2 represents and income between 2x modal and 3x modal.

3.5 Procedure

UPSELLING Clarifying the concept of upselling and finding drivers

UPSELLING

Clarifying the concept of upselling and finding drivers

SUMMARY

PREFACE

TABLE OF CONTENTS

1.

INTRODUCTION

2.

LITERATURE REVIEW

UPSELLING

3.

RESEARCH DESIGN