The power of personalization : the impact of personalization on perceived quality, when a discount is offered

(1)

The power of personalization

The impact of personalization on perceived quality, when a discount is offered.

Annelies Lubbe (10982248)

Dhr. Frank Slisser (supervisor)

University of Amsterdam

MSc. in Business Administration – Marketing

March 13, 2016

(2)

Statement of originality

This document is written by Annelies Lubbe, who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

Abstract

Prices are an important cue for customers to judge quality buying the product or service. On the same time, the importance of interaction and personalization of services is increasing because of competition and specific consumer needs and demands. This study examined the influence of level of service personalization on perceived quality, and the interaction effect of price promotions. Since personalization is a continuum, four different service scenarios from different personalization levels (low and highly personalized services), were used for this research. With regard to price promotions, three different levels of discounts were used in both percentage and absolute framings.

Data was collected from 132 respondents, divided into four groups. All groups were exposed to two service scenarios to evaluate the level of experienced interaction and personalization, and to evaluate the perceived quality based on the given service information and price (full price or including a 30% or 55% discount, framed as an €-off or %-off discount).

Results showed a high level of experienced personalization when the level of interaction is evaluated as high. In addition, a higher level of personalization leads to higher perceived quality than low personalized services. This is different for price promotions, where the highest discount (55%) resulted in the lowest perceived quality. The highest discount level differed significantly from no discount, but there was no significant difference with the 30% discount level. And no differences were found between €-off and %-off framings. Lastly, no interaction effect was found for price promotions. So there was no change in the simple effect of level of personalization on perceived quality, over different levels of price promotions. Different implications of this study are discussed.

(4)

List of tables and figures

Table 1 Survey procedure ... 27

Table 2 Level of Personalization - Descriptive statistics four services ... 31

Table 3 Level of Interaction - Descriptive statistics four services ... 33

Table 4 Perceived quality - Descriptive statistics four services (full price scenarios) ... 36

Table 5 Perceived quality - Descriptive statistics four services (all price scenarios) ... 37

Table 6 Perceived quality - Descriptive statistics all discount levels (No difference between service types was made) ... 40

Table 7 Perceived quality - Descriptive statistics all discount levels, all services ... 40

Table 8 Perceived quality - Descriptive statistics all discount levels, highly and low personalized services ... 42

Table 9 Perceived quality - Descriptive statistics all discount framings, highly and low personalized services... 43

Table 10 Perceived quality - Descriptive statistics interaction effect discounts, highly and low personalized services... 44

Figure 1 Conceptual model ... 22

Figure 2 Level of Personalization - Graphical representation of the difference ... 32

Figure 3 Level of Personalization (left) and Level of Interaction (right) - Graphical representation of the difference and comparison of graphs ... 34

(8)

1. Introduction

Imagine, you want to lose body weight. Summer is coming and after years you want to show your trained body on the beach. You followed some food programs and tried different sports, but the desired effect was not achieved. Probably because it was not the right program or exercise for you to do. Therefore you want someone to make a real personalized health program to finally reach your goals. When contacting a personal trainer, you get offered a service which includes everything to achieve your goal. You have to pay a high price but it sounds it is worth it. The high price means quality; this personal trainer will realize the body you are thinking about. Option two: you contact a personal trainer who offers a deal including an absolute discount. What do you think: no doubt, this is a good deal because saving some money, or do you have doubts about the quality of this personal trainer because there is a discount offered immediately. Maybe you would not even choose this personal trainer if you would see this discount before calling since it decreases your perception of quality too much? And what happens with your quality perception when the discount is offered as a percentage of the total amount?

Services and products are fundamentally different - services differ from goods along the four main characteristics of intangibility, inseparability, heterogeneity and perishability (Zeithaml et al., 1985). Hoffman et al. (2002) state that consumer perceptions of service prices and price strategies should be different from those of goods. In his article, Pizam (2011) mentions that the four P’s – price is one of them – are product oriented and thus not applicable to marketing fields like relationship marketing and service marketing. However, the service sector now creates approximately 78 percent of the US gross national product and employs almost 80 percent of its workforce (Eur-Lex Europe, 2015; Kinard et al., 2006).

(9)

To make the marketing mix suitable for service marketing, one element needs to be added to the four P’s: ‘People’ (Pizam, 2011). The article of Pizam (2011) says that the human aspect emphasizes the personal nature of the service marketing; the service suppliers are an influential tool of customer persuasion and a key factor in affecting the customer’s perception of the provided service quality.

The innovation of technologies accessible for business and customers results in more choices for customers, global access of products or services and different options to address personal and particular desires (Constantinides, 2006). Following Constantinides (2006) this situation makes the service and personalized customer approach essential, and so the article says that more interactive and individual marketing, quality of the relationship between supplier and customer, and loyal customers will be the foundation of commercial performance.

Next to Pizam (2011) and Constantinides (2006), a research by Deloitte Accountants BV (Schilder et al., 2014) shows that consumers experience less differences in travel, hospitality and the leisure industry, due to commoditization. Following the Deloitte Accountants BV’s article (Schilder et al., 2014), consumers are willing to pay higher prices for their services and could be more loyal when a fifth P is activated: the P of personalization. Following the article consumers are willing to pay higher prices for personalized services.

Although consumers are willing to pay higher prices in some situations, there are situations where sellers choose to lower the price by using price promotions. In service industries price promotions are often used as a sales promotion (Nusair et al., 2010). However, price is a central cue for consumers when evaluating service quality before buying. (Hoffman et al., 2002)

Because most pricing research is about product pricing (Hoffman et al., 2002), and because most choices between commoditized, not personalized, services are based on prices (Schilder et al., 2014), the question becomes what the impact is of price on personalized services. Especially, what is the impact of personalization and discounts related to perceived quality. Although services,

(10)

personalization and price promotions are studied before for many times, it seems that there was no research on price promotions linked to the relationship between personalized services and perceived quality. Therefore the following research question is formulated:

What is the effect of personalized service on perceived quality, and how will this be affected by price promotions and different levels of price promotions?

- What is personalization?

- What is the role of personalization with respect to service? - What are price promotions?

- What is the effect of personalized services on perceived quality?

- What is the effect of (different types of) price promotions on perceived quality? - What is the relationship between personalized services and price promotions?

(11)

1.1 Theoretical relevance

This thesis will provide useful insights and suggestions for service managers because it will show the effect of price promotions on perceptions of service quality. Although we know how important price promotions are to be competitive, there could be a negative side of pricing as well (Lee et al., 2014). In addition, we know about the upcoming importance of customization and personalization in services (Logman, 1997; Constantinides, 2006). As a result, consumers are willing to pay higher prices but not in all cases (Candi et al., 2013; Pine et al., 1998). By offering a clear view of the impact of price promotions on the service and its quality perception, new knowledge will be provided

1.2 Managerial relevance

Since the importance of personalization increased, whereby, as a result, people are willing to pay more, it will be interesting for the service managers to know in which way price promotions increase or decrease the perceived quality of the service (Coelho et al., 2012). In some cases price discounts could decrease the value and quality perception and therefore higher prices could be more positive. But although higher prices are generally related to better quality, higher prices are not always better. (Andrew et al., 2010; Hoffman et al., 2002; Lee et al., 2014). This study will contribute to the managers’ knowledge of price settings in a context of service experiences and its quality perceptions.

(12)

2. Literature review

This literature review elaborates on personalization, services and price promotions in the next paragraphs, all with a focus on perceived quality.

Following Logman (1997), mass marketing is shifting to a customized, one-on-one method of reaching individual customers. And because of the bigger concern about customers’ real needs, wants, and demands, the information flow between customers and firms becomes more important (Kinard et al., 2006). Kinard et al. (2006) mentions the usefulness of relationship marketing in the marketing of services, specifically when relating provider-customer relationships to perceived service quality. Coelho et al. (2012) emphasizes that the normative marketing goal should be customization, especially since he found evidence that customization could be a competitive advantage. In addition, Pine et al. (1998) indicates that customization becomes more usual. Price promotions could be a competitive advantage as well. However, they could have a negative side also. (Choi et al., 2012; Nusair et al., 2010; Yoon et al, 2010).

2.1 Personalization

Modern personalization seems to have different types of meanings: from location analysis, matching the visual layout of the message to data terminal equipment, to modifying the content of the message, and tailoring the product (Vesanen, 2007). With this, marketers try to serve customers’ expectations and prevent spam reactions. There is a growing acknowledgement among theorists and experts that product and service differentiation represents a form of competitive advantage. As a more extreme form of differentiation, the idea of customization – which means the level to which the firm’s offering is tailored to meet heterogeneous customers’ needs – has met increasing popularity among firms. (Coelho et al., 2012)

(13)

However, following Hanson (2000) customization is just one of the different levels which together make the personalization continuum. At the one side, far left, is no personalization but homogeneity. Moving to the right, customer needs are more pleased by product diversity and personalized guidance. Moving further to the right, there are different products customized for individual tastes - customization. On the far right, the other side, there is relationship building, with cooperative participation happening over time. Miceli et al. (2007) define the personalization continuum as ranging from product versioning to mass customization, to one-to-one personalization, to co-creation and in the end there is reverse marketing. The level of the role of product diversity and interactional flexibility becomes higher when moving closer to reverse marketing. With respect to the design process, the main role is shifted from seller to buyer. In fact, one-to-one personalization aims at increasing collaboration and close communication between customers and the firm, as well as giving personalized value in terms of services, information, and support. Because of interaction in the early beginning of product development, customers can co-create their own personalized product. Reverse marketing involves in letting the customer entirely shape his or her product to let him or her engaging in the designing process. The firm will produce the completed design.

Like Miceli et al. (2007), Kevoe-Feldman (2015) emphasizes the importance of communication behaviors for personalization in a service context, which is connected with identifying specific needs, offering alternatives and providing assistance. The definition used by Kevoe-Feldman (2015) for personalization - tailored service, or service that attempts to address the unique needs of individual customers - is very close to the definition used for customization by Coelho et al. (2012) (see above). This is exactly what Vesanen (2007) noticed: the literature uses several terms for labeling personalization. For this research the following definition is used: Personalization is a special form of

product or service differentiation, in which a solution is tailored for an specific individual. (Hanson,

2000)

Some early empirical evidence for a positive influence of customization on customer relationships comes from research based on the American Customer Satisfaction Index model (Gronholdt et al.,

(14)

2000). Service customization and personalization contributes to the realization of customer satisfaction, although its impact is slightly lower than that of service quality. Some customers experience added value from interaction with firms and product co-creation (Coelho et al., 2012; Miceli et al., 2007; Vesanen, 2007). However, all forms of personalization does not automatically involve a benefit for the customer. Customers’ demand for customized products and services may vary or even be non-existent – that is consumers do not regard customization as favorable for them, or they link it to high efforts. The benefits of customization may well be compensated by the monetary and non-monetary costs that a customer encounters, such as the higher price of customized products or services, privacy risks, the delay in delivery, and the need for customers to spend time in identifying their preferences before the service can be brought. The great variety may result to annoying complexity and confusion. (Coelho et al., 2012; Miceli et al., 2007; Vesanen, 2007) Simply, when benefits go beyond costs, personalization creates value for customer. If customer’ costs exceed the benefits, the market is not ready to embrace personalization. (Vesanen, 2007)

2.2 Service

Active participation is an essential character of service settings, like getting a new haircut, depositing cash at the bank, or using your phone contract for making a phone call. Therefore customer encounters offer service providers a chance to develop a complex and personal relationship with a customer. Although these relationships can precipitate in all interactional circumstances, personal face- to-face interaction may be more prone to such bonding effects (Kinard et al., 2006). Kinard et al. (2006) show that consumers involved in a high interaction and customized service (face-to-face communication and offering unique services, like hairdressers) have greater relational advantages than consumers engaged in a standardized, moderate contact service (less face-to-face contact and hard to see a difference between providers’ offerings, like a fast food restaurant). As a result, service providers who are able to be more close with the customer, are more likely to fulfill unique wishes.

(15)

In addition, Kinard et al. (2006) provide evidence that there is a significant difference between perceived relational benefits between involvement groups for a high contact service; whereas there are no significant differences in perceived benefits between involvement groups for a standardized, moderate contact service. Service firms creating high contact service offerings are accomplished to instill a perceived interpersonal advantage for highly involved consumers versus less involved consumers, and they can specifically target relational activities associated with confidence benefits to highly involved consumers (Kinard et al. 2006). As a result it produces relational performance results. For example, hairdressers could increase consumer loyalty, satisfaction and positive word-of-mouth by focusing on people care a lot about their look, versus people who see it as less important.

2.3 Relation personalized service and perceived quality

Perceived value is the consumer’s overall assessment of the utility of a product based on perceptions of what is received and what is given (Gupta et al., 2006). Zeithaml (1988) defines value as “all factors, both qualitative and quantitative, subjective and objective, that make up the complete shopping experience”. Following Miceli et al. (2007) customer perceived value depends on how the interaction between seller and customer is tailored to the exclusive wishes of customers and product categories. This is like mentioned earlier: different levels of personalization concern both interaction and content issues.

However, the definition of perceived value is too broad so that it is too difficult to measure the variable with respect to validity and reliability, confirmed by Gupta et al. (2006). Therefore the meaning of perceived quality is given here as well. Perceived service quality is “the degree and direction of discrepancy between customers’ service perceptions and expectations”. (Gupta et al., 2006)

Following Coelho et al. (2012), the more the service is customized, the higher the level of perceived quality. Since the definition of Coelho et al. (2012) of customization is very close to definitions of

(16)

personalization (mentioned earlier), the suggestion could be made that personalization has a positive impact on perceived value as well as on perceived quality.

Looking at the analysis of Coelho et al. (2012), the outcome of customization on perceived quality is bigger for banking industry than for cable TV, probably because there are more aspects that could be customized in the banking industry. In addition, this research shows that customization and quality are the main predictors for satisfaction (Coelho et al., 2012). Kevoe-Feltman (2015) mentions the positive correlation between personalization and customer satisfaction as well, and states that the perception of perceived attention while realizing the service is generally the foundation of level of satisfaction. Here we see the link between personalized service and getting attention as well, which means interaction between seller and customer. Solomon et al. (1985) gives an explanation provided by the role theory: by changing their behavior for different customers (personalizing their actions), frontline employees can better meet individual consumer’s expectations and so the service provider realizes customer satisfaction.

Although Kevoe-Feltman (2015) and Tax et al. (1998) say that there is a positive correlation between personalization and interaction, and satisfaction on the other side, Kevoe-Feltman (2015) says that customers link personalized service to quality and that quality has a direct relationship with customer satisfaction. So the conclusion can be made that the higher the customers’ perceived attention, the higher the perceived quality, which results in higher satisfaction. This mediating effect of quality is confirmed by Coelho et al. (2012) and indicated by Gupta et al. (2006), who say that service quality can be seen as the influencer of financial performance, but as also as the antecedent of customer satisfaction. Also Kinard et al. (2006) confirm this relationship and mention the positive link between provider-customer relationships and perceived quality of services, since it gives evaluative criteria for customers to value factors like trust, commitment and reliability.

(17)

industry, especially when explaining perceived quality and customer loyalty. As mentioned earlier, Coelho et al. (2012) explain the different effects by the opportunities to customize services: there are more personalization opportunities in the banking industry. Moreover, it is probably easier to use customization to create a sense of exclusivity and satisfaction of needs in banking by more personalized interaction, which is emphasized by Kinard et al. (2006); service providers using high levels of interaction with the customer create greater benefits for customer experience, compared to providers with moderate to low levels of customer-provider interaction. So, this emphasizes the suggestion that the higher the interaction, the more the service is personalized, and the more the service is personalized, the higher the perceived quality is. Therefore the following hypothesizes are formulated:

H1: There is a positive relationship between interaction with one specific individual when realizing the service, and level of experienced personalization.

H2: There is a positive relationship between level of experienced personalization of the service and perceived quality.

2.4 Price promotions

Following the decision theory (Monroe, 1973), consumers use cues that are most simply accessible in the alternative judgment process to evaluate product quality. Because of the importance of its role as a cue for services, price should be a central sign for customers to evaluate service quality before buying. (Hoffman et al., 2002; Yoon et al. 2010, Andrews et al., 2010). Since price is an important factor to evaluate the product or service, marketers investigate a lot each year to promote service bundles by emphasizing convenient payment methods, free gifts and discounts. (Andrews et al., 2010). A price discount can be defined as a “short-term reduction in the list price of a product or service” (Yoon et al., 2010).

(18)

Next to the reduction which is short-term, the positive brand effects of a price discount are short-term related as well, like motivating and entertain shoppers, boost specific sales groups, realizing changing brand choices, and demoralize new coming competitors. On the other hand there are some negative effects of discounts which are linked to the long-term, like consumer loyalty and lowering the perceived product value. (Lee et al., 2014; Gupta et al., 2006; Nusair et al., 2010)

While Lee et al. (2014) mention that price promotions could result in boosting brand switching, Andrews et al. (2010) suggest that no change in switching or search plans can be assigned to the incentives’ value. Andrews et al.(2010) emphasize, however, that the value of savings in economic terms can improve perceptions of value, which is a central aspect for consumers when searching for extra choice alternatives.

Even from consumer perspective there are some positive and negative effects. On the one hand consumers are given the idea of ‘made a good deal’ which results in a positive shopper mood, and at the same time the competitive products which do not include any discount, are evaluated in a more negative way. In addition, there may be a reduction of ‘pain of payment’ as a result of the discount, and so consumers like consuming the product more. (Lee et al., 2014; Gupta et al., 2006; Nusair et al., 2010) However, Lee at al. (2014) also show some negative sides of the discounts, like decreasing attention when consuming and therefore the pleasure is lower, and the decrease of sunk-cost considerations or the inner need to validate the costs. There is also a negative discount effect on consumers’ perceptions, like the perception of quality – higher prices are linked to higher quality –, brand expectations and uncertainty about the brands’ value (Nusair et al., 2010).

Lee et al. (2014) confirms the price-quality relationship. They mention that lowering the price of an ordinary unbranded shirt there will be a more negative effect for perceived quality, compared to a product like Coke since this its quality is less ambiguous than quality of an ordinary unbranded shirt. In this research it is about services. And since it is difficult to know about the quality of service before

(19)

consuming, we could conclude that the quality is more ambiguous and therefore lowering the price has a negative impact (regardless the way a price is lowered, e.g. all types of discount).

2.5 The effect of different price discount framings

Since there could be discounts on different levels, research shows that different discount levels have different impact on perceived quality (Nusair et al., 2010; Yoon et al. (2010). Research of Nusair et al., who investigated absolute and percentage discount forms, (2010) show that increasing trends in service quality and purchase intention stopped at 60% discount for quick service restaurants (QSR) and outlet malls. For budget hotels the turn to negative affect on perceived quality is on the 40% discount level, whereas the 20% discount level results in the highest quality expectation for mailing service. Yoon et al. (2010) indicate that perceived quality of high-end services decreased at a 50% discount level, and that there is no significant difference in perceived quality results between no discount and a 30% discount level. The explanation for different effects from different discount levels, given by Nusair et al. (2010), has to do with the higher risk and unrecoverable losses. So we could say that different discount levels affect consumers’ perceived quality, and that the effect varies across types of service.

Chen et al. (1998) and Darke et al. (2005) state that discounts in general – no difference in discount method or frame was made – can lead to more negative consumer perceptions by a lowered quality perception of the discounted item. The self-perception theory (Bem, 1973) - buyers link their purchases to the price promotion (“I bought it because there was a discount”), rather than to the purchase quality (“This is a perfect product and therefore I bought it.”) - gives an explanation for the unwanted effects of discounts. Darke et al. (2005) emphasize that discount effects were low when there is no quality guarantee, which is consistent with the suggestion that a negative price-quality relationship can decrease the perceived value of a discounted purchase. Rao et al. (1989) states the bigger the difference in the price dealings, the larger the probability that individuals perceive

(20)

differences in prices and begin to suggest about purchase’ quality. For Rao et al. (1989) this could be a reason for the weak significant price main effect in some price-perceived quality studies.

Next to different discount levels, there could be different discount frames to promote services. The way a price-promotion is framed is expected to influence consumers’ perceptions of price, quality, and purchase intentions. (Nusair et al., 2010) In this research we focus on two formats: dollar off discounts and percentage off discounts. Format, or framing, effects will be observable when the same price discounts are offered in different formats. Price discounts related to a high-priced products show a more significant effect when framed as an absolute discount, whereas a relative discount is more significant when linked to a low-priced product. (Nusair et al., 2010). A given explanation by Nusair et al. (2010) is that the dollar off discount is mostly seen to be of a bigger size than the relative difference.

In addition, consumers evaluate the amount of a price promotion in absolute terms when the original price is the seller’s own price, whereas buyers evaluate it in relative terms when the reference point is a competitor’s price (Choi et al., 2012). Next to that, the percentage format is preferred over the dollar format for quality assessment. (Nusair et al., 2010)

It seems logically to link more service to higher prices and therefore prefer dollar off discounts when promoting this high-end services. Research of Yoon et al. (2010) confirms this thought. Their results indicate that the dollar format is preferred for high-end services, while the percentage discount framing is favorite for low-end services. In addition the research shows that the absolute discount framing was more significant for hospitality industries (hotels and restaurants). Since quality of the service provided by employees has been acknowledged as the most persuasive factor in determining customers’ overall satisfaction and the probability of make a reservation for the hotel again, it seems that hotel customers today not just demand for basic services and facilities provided by a hotel. They are looking for a high level of being served personally. (Choi et al, 2001) Therefore we could state that high end services and hospitality services are comparable to personalized service and therefore,

(21)

following the theory, dollar off framings would be more significant for personalized services. Therefore, we propose the following hypotheses:

H3: There is a negative relationship between price discounts and consumer’s perceived quality of services.

H3a: The influence of price discounts in a €-off framing will be greater for highly personalized services, than for low personalized services.

H3b: The influence of price discounts in a %-off framing will be greater for low personalized services, than for highly personalized services.

Since the two variables - personalized service and price discounts - have different impact on perceived quality, a negative moderating effect of price discounts on the relationship between personalized service and perceived quality is expected. (H4)

(22)

2.6 Conceptual model

H1: There is a positive relationship between interaction with one specific individual when realizing the service, and level of experienced personalization.

H2: There is a positive relationship between level of experienced personalization of the service and perceived quality.

H3: There is a negative relationship between price discounts and consumer’s perceived quality of services.

H3a: The influence of price discounts in a €-off framing will be greater for highly personalized services, than for low personalized services.

H3b: The influence of price discounts in a %-off framing will be greater for low personalized services, than for highly personalized services.

H4: The positive relationship between personalized service and perceived quality is moderated by price discounts, so that the relationship is weaker when the personalized service includes a price discount. (negative moderating effect of price discounts)

(23)

3.0 Methodology

In order to test the conceptual model (figure 1) the first step in the research was to find appropriate personalized services. Since personalization is a continuum, two different levels - mass customization and reverse marketing - and two examples of each, are used and compared to see differences.

3.1 Pretest

To know which examples needed to be used, and whether these examples were the right services to test, a pretest was developed. For this research four services were selected to test the level of personalization: flights (airline ticket), mobile phone contract, hairdresser and personal trainer. There was an assumption that flights and mobile phone contract are seen as low personalized services, and hairdresser and personal trainer are seen as highly personalized services.

To measure the level of service personalization, a description of four independent services was given. (See appendix 1 for all four service descriptions.) Respondents evaluated four different services by several statements (same statements for every service scenario to make comparison possible). There was searched for a personalization measurement scale, but no appropriate measurement scale for service personalization was found. Therefore the statements, based on articles of Miceli et al. (2007) and Wind et al. (2001), resulted in a new measurement scale which was using 7-point Likert scales.

As said, the developed scale was based on articles of Miceli et al. (2007) and Wind et al. (2001). Both articles provide a continuum of the changing way of doing marketing, in which criteria made clear the differences between different levels of personalization. Wind et al. (2001) shows differences between mass & segmented marketing, and customerization on the other side. Miceli et al. (2007) show a continuum from product versioning to mass customization, to one-to-one personalization, to co-creation and in the end there is reverse marketing. Both articles mention different but comparable

(24)

criteria to split the levels. A combination of both resulted in the measurement scales, used for the pretest (see appendix 2).

3.2 Results pretest

The level of personalization scale has reliability in all scenario’s, with Cronbach’s Alpha = .768 (personal trainer), Cronbach’s Alpha = .847 (flight), Cronbach’s Alpha = .874 (mobile phone contract), and Cronbach’s Alpha = .805 (hairstylist). Deleting two items increased the reliability.

The objective of the pretest was to assess the respondents’ perception of personalization of the four services given, in order to know whether there is a statistically significant difference between level of personalization of the services used in the survey (flight and mobile phone contract on the side of low level of personalization, personal trainer and hairstylist on the side of high level of personalization). To compare the four services – personal trainer, flight, mobile phone contract and hairstylist - a one way repeated measures ANOVA was conducted (N = 28, 2 cases were excluded, sampling method was convenience sampling).

The assumption was that personal trainer (M = 5.12, SD = 0.95) and hairstylist (M = 5.16, SD = 1.20) differ significantly in level of personalization from flight (M = 3.95, SD = 1.38) and mobile phone contract (M = 4.35, SD = 1.42). A significant effect was found in the multivariate test, Wilks’ Lambda = .62, F (3,25) = 5.013, p = .007. So, there is an overall difference in means.

Follow-up (pairwise) comparisons indicate where the differences occurred. There is a significant difference in level of personalization between personal trainer and flight (p = .002), between personal trainer and mobile phone contract (p = .037), between hairstylist and flight (p = .003), and between hairstylist and mobile phone contract (p = .037). These comparison results are like assumption before testing: flight and mobile phone contract differ from personal trainer and hairstylist.

(25)

The difference between personal trainer and hairstylist is not statistically significant (p = .881). This is also like assumption: both services were assumed to be of high level of personalization, and no difference was expected. In contrast to assumption, there is a significant difference between flight and mobile phone contract (p = .020). So, both services are not on the same level of personalization. Therefore, the text about mobile phone contract is changed a bit before using it in the survey, in a way that the mean of mobile phone contract is expected to be closer to the mean of flight. More concrete: the additional music service is deleted to make it a more homogeneous service (the text about flight service also does not contain additional services).

3.3 Survey

This section elaborates on the method to measure different constructs in the research model.

The strategy used in this research was an experimental design, executed by using surveys (self-completed). The impact of interaction on personalization, and the impact of personalization on quality is measured by evaluating different scale items. The moderation effect of discounts (level and frame) is tested by manipulating different scenario’s.

3.3.1 Measures

Personalization

Like in the pretest, a service description was shown to the respondents and they were asked to evaluate different items. So (level of) personalization was tested by the measurement scale used in the pretest. Since this measurement is very important and could differ on individual level, the same statements were evaluated in survey again.

Communication

Some statements (see bold statements in appendix 2) measured the impact of interaction, and therefore the relationship between interaction and level of personalization could be tested.

(26)

Perceived quality

To measure the effect of personalization on perceived service quality, respondents were asked to imagine they buy the presented service, given a presented price. Then respondents were asked to evaluate different statements about the perceived quality. The measurement scale for testing the perceived quality was a combination of the ACSI measurement scale (Fornell et al., 1996) and the measurement scale used by Coelho et al. (2012), based on the ECSI measurement scale, a European version of ACSI. Since the services were not really bought and experienced, the statements were ‘translated’ to a form of perceived service quality-expectation (see appendix 3).

Price discounts (level and framing)

To measure the moderation effect of price discounts (different levels and different frames), the effect of personalization on perceived service quality was evaluated in different scenarios: different service (4 types), with different discount levels (3 levels (0, 30%, 55%)) framed differently (2 framings (percentage and absolute)). This resulted in a 4 x 3 x 2 approach.

The selected discount levels were based on previous literature. Nusair et al. (2010) show different turning points to negative affect on perceived quality (60% for QSR, 40% for budget hotels and 20% for mailing service). Yoon et al. (2010) indicate that perceived quality of high-end services decreased at a 50% discount level, and that there is no significant difference between 0% - 30% discount level. The percentages are translated to absolute values for the absolute framed discounts.

Since it could be easy to see that the absolute value of 50% is half the price, there could be a possible result of no difference between percentage and absolute framing. Therefore the choice was made to set this discount on a level of 55% (and an absolute value of 55%).

So the chosen levels were no discount, 30% and 55% for testing the discount effect on perceived quality.

(27)

3.3.2 Procedure

Since the experimental design resulted in a 4 x 3 x 2 approach (service type x discount level x discount framing), 24 manipulated scenarios were developed. To keep respondents motivated and to reduce participants’ burden to respond to all scenarios, 24 scenarios were divided into four treatment groups: each group contained 6 scenarios.

Since we wanted to test the pretest measurement scale again in this survey, the chance of reduced motivation could increase. Therefore respondents were presented just two of the four services. This means just two scenarios with statements about level of personalization (like in the pretest), and six scenarios which contains just two service types, but combined with absolute or percentage discounts on different discount levels (and randomly mixed). (See table below: two services per group, different discount levels and different framings)

To reduce the text in this survey, questions about quality and discount were based on a picture which showed the service, service features, price and potential discount. (See appendix 4 for examples of pictures, see appendix 5 for example Survey Group 1)

Group 1 Group 2 Group 3 Group 4

PT text + pretest question TC text + pretest question HS text + pretest question FL text + pretest question TC text + pretest question HS text + pretest question FL text + pretest question PT text + pretest question

PT full price TC full price HS full price FL full price

PT 30% discount (%) TC 30% discount (%) HS 30% discount (%) FL 30% discount (%) PT 50% discount (€) TC 50% discount (€) HS 50% discount (€) FL 50% discount (€)

TC full price HS full price FL full price PT full price

TC 30% discount (€) HS 30% discount (€) FL 30% discount (€) PT 30% discount (€) TC 50% discount (%) HS 50% discount (%) FL 50% discount (%) PT 50% discount (%)

PT = Personal trainer TC = Telephone contract

HS = Hairstylist FL = Flight

(28)

3.3.3 Sample

The population for this study is everyone who would buy a personalized service. Since this population is very broad and large, the research sample was specified to students following a hospitality management study (all on same university EuroCollege Hogeschool). Although this was just a part of the population, this small sample made comparison best possible because of its homogeneity. Because of their background (service industry), there was a link to the research topic. In addition, the students are future service customers, which makes the study interesting for future business models.

A simple random sampling technique was used (within the university). The sample size had a minimum of 150 respondents (30 for the pretest, and 30 for each group in the main survey). Since the survey was send by university’s location manager, and since filling in the survey was not mandatory, it is hard to say wat the response rate is. However, it is known that not all students filled in the survey since some of researchers’ Facebook connections had to be asked to fill in the survey.

(29)

4. Results

To perform the statistical analysis, the Statistical software Package for Social Sciences (SPSS) was used.

4.1 Descriptive data of sample

At 12 January 2016 the survey was sent to potential respondents. The survey was distributed by e-mail to students of EuroCollege Hogeschool. As mentioned before, filling in the survey was not mandatory for students and therefore the author of this thesis sent the survey to Facebook contacts as well (preferably people with a management or marketing related background).

On 21 January 2016, closing date of survey, 175 unique respondents filled in the survey. To analyze the data, 43 respondents were deleted because of incomplete/unfinished surveys and unrealistic short response durations (e.g. 2m 8s). Since the author does not know exactly how many people were contacted to fill in the survey (e-mailing the students was done by a location manager of EuroCollege Hogeschool and a Facebook messages were shared by friends to their network (snowball sampling method to have enough respondents in a limited time period)), it is not possible to mention the response rate.

Of the 132 respondents, approximately 40% was male and almost 60% was female. Most respondents have an age between 21 and 30 years old: 85,6%. This overrepresentation makes sense because of survey distribution in a university, and although this overrepresentation does make comparison with the total population unreliable, it could be interesting for (future) business models focused on this specific age. Only three respondents have a nationality other than Dutch, and so there is a minimal risk that comparison of answers is not reliable because of cultural differences. More than 50% has a

(30)

background in marketing, business administration, management or economics. All detailed frequencies are shown in appendix 6.

As described in the methodology chapter, respondents were randomly assigned to one of the four groups which all four contains two services. All respondents were exposed to two different service explanations and were asked to evaluate different statements. Although evaluating the statements was the same as in pretest, respondents were exposed to only two services this time. In addition, respondents were exposed to six different pictures: two services, discount levels and discount framings mixed (see table 1 Survey procedure).

4.2 Variables and measurement

4.2.1 Reliability

In order to assess the consistency of the different scales that were used, a check on reliability was performed. The respondents were randomly divided over four different groups with similar questions about different, manipulated scenarios. Like in the pretest, data from the survey showed that the level of interaction and level of personalization were both reliable after deleting two items (Cronbach’s Alpha > .70). For the perceived quality variable reverse coded items were recoded. After deleting some items, the scale was reliable (Cronbach’s Alpha > .70).

4.2.2 Retesting the pretest

Although the level of interaction and level of personalization were already measured in the pretest, the variables were tested again since the text about the mobile phone contract service was changed in response to pretest results (deleting the information about the music service to make the overall service more homogeneous). Moreover, the number of respondents that filled in the survey was higher

(31)

than the number of respondents that filled in the pretest. Therefore analysis based on survey data was more reliable.

New variables for the level of personalization and interaction were computed for all of the four services by taking all scale items together. They are transformed into one overall variable for level of personalization and interaction, which means the average of all items about personalization or about interaction. Next an one way ANOVA was conducted to test the assumption about differences between the four services: is there a statistically significant difference in levels of personalization between mobile phone contract (M = 4.64, SD = .99) and flight (M = 4.34, SD = 1.15) on the one hand (low level of personalization), and personal trainer (M = 4.96, SD = .83) and hairstylist (M = 5.09, SD = .84) on the other hand (high level of personalization). Because all four services need to be compared, but only two – instead of all four – services were shown to a respondent in the survey, a one way ANOVA was performed (between subjects) instead of a repeated measures ANOVA like in the pretest (within subjects).

Service N Mean Std. Dev.

Hairstylist 67 5.09 .84

Personal trainer 65 4.96 .83

Mobile phone contract 65 4.64 .99

Flight 67 4.34 1.14

Dependent Variable: Personalization

Table 2 Level of Personalization - Descriptive statistics four services

The one way ANOVA showed a significant effect, F (3, 260) = 8.42, p < .05. However, the Levene’s test showed a significant outcome (p < .05), which means that variances of the services are significantly different. Even after checking for outliers, the test of homogeneity of variances was significant (p < .05). For the ANOVA, the Brown-Forsythe and Welch were performed as well. Both Welch (F (3, 143.507) = 7.723, p = .00) and Brown-Forsythe (F (3, 242.809) = 8.432), p = .00) are

(32)

both highly significant. So, there was a significant effect of service type on level of personalization, and therefore we move on to see where the statistically significant difference between the services lie. The Games-Howell procedure was selected as post hoc test since it does not rely on the assumption of equal variances. Follow-up comparisons show a significant difference between hairstylist and mobile phone contract (p = .03), between hairstylist and flight (p = .00), between personal trainer and mobile phone contract (p = .049), and between personal trainer and flight (p = .00). Differences between hairstylist and personal trainer (p = .79) and between mobile phone contract and flight (p = .35) were not significantly different. All results are like assumption. Therefore we could conclude that changing the mobile phone contract text had a positive effect. An additional planned contrasts test was executed to see whether there was a significant difference between personal trainer and hairstylist on the one side, and mobile phone contract and flight on the other side. This test shows an significant outcome, like assumed, p < .05.

As shown in the graph below, it is visible that hairstylist (1) and personal trainer (2) are closer to each other than mobile phone contract (3) and flight (4). In addition we could see the decreasing line, which also show the difference between hairstylist and personal trainer on the one side, and mobile phone contract and flight on the other side.

(33)

4.3 Hypothesis testing

The next session will test the hypotheses derived from the conceptual model, paragraph 2.6 – figure 1.

Hypothesis 1: There is a positive relationship between interaction with one specific individual when realizing the service, and the level of experienced personalization.

We know that there is a significant difference between the four services, see above. Therefore the first step here was to conduct an one way ANOVA to see whether there is a significant difference between interaction level of services as well.

New variables for interaction were computed for all of the four services by taking all scale items together. The descriptive statistics of the variables are shown in the table below. They are comparable to the descriptives of level of personalization (decreasing values of mean and close means of hairstylist and personal trainer), which is shown in figure 3.

Service N Mean Std. Dev.

Hairstylist 67 5.07 .80

Personal trainer 65 5.07 .85

Mobile phone contract 65 4.70 .99

Flight 67 4.45 1.16

Dependent Variable: Interaction

(34)

Figure 3 Level of Personalization (left) and Level of Interaction (right) - Graphical representation of the difference and comparison of graphs

The one way ANOVA showed a significant effect, F (3, 260) = 6.74, p < .05. However, the Levene’s test showed a significant outcome (p < .05), which means that variances of the services are significantly different (same as in testing the level of personalization). Since Levene’s test was significant, the Games-Howell procedure was selected as post hoc test to see where the significant differences are.

Follow-up comparisons show a significant difference between hairstylist and mobile phone contract (p = .03), between hairstylist and flight (p = .00), between personal trainer and mobile phone contract (p = .03), and between personal trainer and flight (p = .00). Differences between hairstylist and personal trainer (p = 1.00) and between mobile phone contract and flight (p = .53) were not significantly different. All results are like assumption, like they were for level of personalization. And so results are comparable with the one way ANOVA for level of personalization.

To test this hypothesis the correlations between the two variables were checked: interaction and level of personalization. When looking at the total variables (all four services taken together), interaction

(35)

correlated significantly with level of personalization (r = .89, p = .000) (n = 264 since all respondents evaluated two service types).

To be sure there is a direct relationship, a multiple linear regression test was performed. Control variables age, gender, background and nationality were taken into account as well, to predict the variance of the dependent variable level of personalization. In step two the independent variable ‘interaction’ was added to the model.

The first step was not statistically significant. After entry of interaction the total variance of the model as a whole was 82%, F (5, 126) = 120.99, p = .00. Interaction was the only predictor variable that was statistically significant (Beta = .90, p = .00). Therefore, hypothesis 1 is accepted.

Appendix 7 shows the model summary and the regression plot. In addition the correlation data can be found in appendix 7.

Hypothesis 2: there is a positive relationship between level of experienced personalization of the service and perceived quality

A three-way ANOVA was conducted to see the main and interaction effects of all three variables (level of personalization, discount level and discount framing) on dependent variable perceived quality. Since the interaction effect of all three variables is significant (F (3, 772) = 3.981, p < .05), the simple main effects are not totally reliable for a test where one or two of the independent variables could be excluded (like hypothesis 2). Therefore new tests were done. The effects of the three way ANOVA can be found in appendix 8.

To test hypothesis 2 the one-way ANOVA was conducted to check upon the significance of difference between the mean scores. Personalized service was measured here by comparing the

(36)

perceived quality means for the four service scenarios: hairstylist (highly personalized), personal trainer (highly personalized), mobile phone contract (low personalized) and flight (low personalized). First, this hypothesis was tested for perceived quality outcomes of the ‘full price’ scenarios. Perceived quality values of scenarios including a discount were excluded in this first measurement. In this case, N = 264 since all four experimental groups (approximately 33 respondents per group) filled in two ‘full price’ scenarios. Descriptive statistics for each service, measured on a 7-point scale, are shown in table 4.

Service

N Means Std. Dev.

Hairstylist 67 4.642 1.15816

Personal trainer 65 4.589 1.22668

Mobile phone contract 65 3.873 1.01658

Flight 67 3.627 1.09611

Dependent Variable: Perceived quality

Table 4 Perceived quality - Descriptive statistics four services (full price scenarios)

Looking at the means, a decreasing structure could be followed. The sequence of services shown in the table, follows a structure of decreasing level of personalization. So, following the means, we could say that perceived quality is higher when the level of personalization of the service is higher. To see whether the differences are statistically significant, a one-way ANOVA is performed.

The Levene’s test for equality of variances was not statistically significant, F (3, 260) = .936, p = .42. Therefore, the outcomes of the one-way ANOVA test are reported, where equal variances were assumed. This test showed a statistically significant difference between means, F (3, 260) = 13.58, p = .00. Results of the Tukey post hoc tests show where the significant difference between the services lies.

(37)

The post hoc test revealed that the perceived quality was significantly higher for hairstylist and personal trainer, compared to mobile phone contract and flight. Differences between hairstylist on the one side, and mobile phone contract (p = .00) and flight (p = .00) on the other side were significantly different. Differences between personal trainer on the one side, and mobile phone contract (p = .00) and flight (p = .00) on the other side were significantly different as well. Because of the insignificant differences between hairstylist and personal trainer (p = .99), and between mobile phone contract and flight (p = .59), we can conclude that there is a difference between highly and low personalized services with respect to perceived quality. (See appendix 9)

In addition to these results where only perceived quality evaluations of ‘full price’ scenarios were included, the one-way ANOVA test was conducted for all scenarios (all prices, price discounts and framings included) where respondents were asked to evaluate perceived quality.

Service N Means Std. Dev.

Hairstylist 201 4.4117 1.06124

Personal trainer 195 4.3346 1.13458

Mobile phone contract 195 3.7872 1.01704

Flight 201 3.5522 1.04272

Table 5 Perceived quality - Descriptive statistics four services (all price scenarios)

Looking at the table, perceive quality means follow a decreasing structure when the services decrease in level of personalization. Because of this structure, the conclusion could be made that there is a difference in perceived quality when level of personalization changes, like could be done for the full price scenarios

The one-way ANOVA confirmed this assumption and showed that the effect of personalized service on perceived quality was statistically significant, F (3, 788) = 30.79, p = .00. Levene’s test was

(38)

insignificant, p = .56. Like before, the Tukey post hoc tests showed a significant difference between hairstylist and mobile phone contract (p = .00), between hairstylist and flight (p = .00), between personal trainer and mobile phone contract (p = .00), and between personal trainer and flight (p = .00). Hairstylist and personal trainer (p = .89) and mobile phone contract and flight (p = .13) did not significantly differ. (See appendix 10)

Even when taking into account both fixed factors of discount level and discount framing, the effect of service type on perceived quality was significant at a 0.05 significance level. The main effect of service type showed a F ratio of F (3, 772) = 30.25, where p = .00 (see appendix 8). The one-way ANOVA test (results were reported above) indicated where the significant difference occurred.

Since we concluded that the main effect of service type on perceived quality is significant and since there is a difference in level of personalization between hairstylist and personal trainer on the one side (highly personalized service) and mobile phone contract and flight on the other side (low personalized service), we can conclude that there is a difference in perceived quality between different levels of personalization. And because of the increasing values of means when level of personalization increases, the conclusion can be made that there is a positive relationship between personalized service and perceived quality. Therefore, hypothesis 2 is accepted.

Hypothesis 3: there is a negative relationship between price discounts and consumer’s perceived quality of services.

To test hypothesis 3 a one-way ANOVA was performed to check upon the significance of difference between the mean scores of different discount levels. The influence of price discount on perceived quality was measured here by comparing the perceived quality means for three discount levels: no discount, 30% discount and 55% discount. For hypothesis 3, absolute and percentage framings were not differentiated.

(39)

Looking at table 6, means are decreasing when discounts are increasing. So, following these table we could say that perceived quality is higher when there is no discount. And the higher the discount, the lower the perceived quality. To see whether the differences are statistically significant, a one-way ANOVA is performed.

The Levene’s test for equality of variances shows a statistically significant effect, p = .03. However, the Robust tests of equality of means show significant effects as well (Welch: p = .01, Brown-Forsythe: p = .01). Therefore the ANOVA results can be followed and are reported. This test showed a statistically significant difference between means, F (2, 789) = 4.89, p = .01.

The Games-Howell post hoc test revealed whether the differences between means were significant. The test showed that no discount results in a significantly higher perceived quality of service than the 55% discount (no difference in framing was made for this analysis), p = 0.01. Differences between no discount and 30% discount, of 30% and 55% discount were not statistically significant (respectively p = .15 and p = .40). (See appendix 11)

However, since we know that the impact of service type on perceived quality is different, the one-way ANOVA for the effect of discount was performed for all services separately to exclude the impact of level of personalization

(40)

Discount level N Means Std. Dev.

No discount 264 4.1818 1.20522

30% discount 264 4.0009 1.02035

50% discount 264 3.8797 1.11972

Total 792 4.0208 1.12313

Table 6 Perceived quality - Descriptive statistics all discount levels (No difference between service types was made)

Table 7 Perceived quality - Descriptive statistics all discount levels, all services

Hairstylist Personal Trainer Mobile phone contract Flight

Discount level

N Means Std. Dev. N Means Std. Dev. N Means Std. Dev. N Means Std. Dev.

No 67 4,64 1,16 65 4,59 1,23 65 3,87 1,02 67 3,63 1,10 30% 67 4,32 ,945 65 4,41 1,01 65 3,78 ,956 67 3,50 ,905 50% 67 4,28 1,05 65 4,00 1,10 65 3,71 1,08 67 3,53 1,13 Total 201 4,41 1,06 195 4,34 1,14 195 3,79 1,02 201 3,55 1,04

(41)

Table 7 shows the means for each level of discount for all services. Means of hairstylist, personal trainer and mobile phone contract follow the structure where the value of perceived quality is decreasing when the discount is increasing. One contradictory outcome is for the flight service, where data show a higher mean of perceived quality for a 50% discount level than for the 30% discount level.

To see whether differences between means of discount levels for all services are significant, one-way ANOVA tests were done for the four services types separately. For all services the Levene’s tests of homogeneity of variances were insignificant. Therefore the ANOVA results are reported and the Tukey HSD post hoc test was done when the ANOVA result was significant.

First of all, the differences between means of hairstylist were statistically insignificant, F (2, 198) = 2.421, p = 0.91. The one-way ANOVA resulted in a statistically insignificant p-value for mobile phone contract as well, F (2, 192) = .429, p = .652. Also flight, the service where the 50% discount level showed a higher value of perceived quality than the 30% discount level, showed an insignificant result, F (2, 198) = .269, p = .764. Only the personal trainer service resulted in a statistically significant effect, F (2, 192) = 4.712, p = .010.

Because of the significant differences between means of perceived quality for the personal trainer service, the Tukey HSD post hoc tests were performed to see where the differences lie. Results of the multiple comparisons show a statistically significant difference between no discount and the 55% discount (p = .009). Differences between no discount and 30 % discount (p = .637), and 30% discount and 55% discount (p = .095) were insignificant, like the post hoc test results before (where services were not differentiated). (See appendix 12)

Although just one of the four services is significant, we can see from the descriptive statistics that hairstylist and personal trainer on the one side have means which are close to each other, and mobile phone contract and flight on the other side have values which are close to each other. And since this

(42)

research is differentiating between these two groups – highly and low personalized services – it could be more interesting for this research to do the one-way ANOVA for highly personalized and low personalized services.

Highly personalized services Low personalized services

Discount level N Means Std. Dev. N Means Std. Dev.

No 132 4.6155 1.18813 132 3.7481 1.06088 30% 132 4.3636 .97309 132 3.6383 .93710 50% 132 4.1420 1.07637 132 3.6174 1.10448 Total 396 4.3737 1.09725 396 3.6679 1.03555

Table 8 Perceived quality - Descriptive statistics all discount levels, highly and low personalized services

The means of highly personalized services (see table above) are decreasing when the discount level is increasing, for low personalized services there is no specific sequence for the means. In addition, the differences between means for highly personalized services are bigger than they are for low personalized services. Two one-way ANOVA’s show whether these differences between means are statistically significant.

The differences between means of the highly personalized services are statistically significant, F (2, 393) = 6.319, p = .002. Following the post hoc tests (Tukey HSD) it is clear that there is a significant difference between no discount and 55% discount ), p = .001. Differences between no discount and 30% discount (p = .143) and differences between 30% discount and 55% discount (p = .221) were not statistically significant. In line with the observation of the statistics, the one-way ANOVA does not show a significant difference between means for low personalized services, F (2, 393) = .606, p = .546. (See appendix 13)

The power of personalization : the impact of personalization on perceived quality, when a discount is offered