The relationship between marketing metrics and consumers’ true willingness to pay for a brand at the point of purchase : what kind of relationship exists between marketing metrics and consumers’ true willingness to pay

(1)

Master Thesis – MSc Business Administration – Marketing Track

The Relationship between Marketing Metrics and Consumers’ True Willingness

to Pay for a Brand at the Point of Purchase

What kind of Relationship exists between Marketing Metrics and Consumers’

True Willingness to Pay for a Brand at the Point of Purchase and how strong is

this Relationship?

Study: MSc Business Administration – Marketing Track Institution: Amsterdam Business School / University of Amsterdam Thesis Supervisor: F. Slisser

Student: Nienke Noorlag Student number: 10872116

(2)

Statement of originality

This document is written by Student Nienke Noorlag, 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

Many different marketing metrics have been used to predict future consumer behavior. The hypothetical bias, which refers to differences in reaction between cases where the

consequences are hypothetical or where they are real, makes it difficult to estimate the demand for a brand and in designing optimal price schedules. In the present study, the relationship between different marketing metrics and the true willingness to pay (WTP) of consumers for a brand is investigated. A Vickrey Auction was used as research method to study this relationship. The results revealed that all tested marketing metrics individually have a significant linear relationship with consumers’ true WTP. However, by putting these

marketing metrics together in a multiple regression model, only the marketing metrics ‘behavioral customer loyalty’ and ‘income’ had a significant relationship with consumers’ true WTP. This indicates an important new insight, namely the clear difference between relationships of objective and factually proven marketing metrics with consumers’ true WTP and relationships of hypothetical marketing metrics with consumers’ true WTP. However, of both behavioral customer loyalty and income, the Beta values and the total variance explained by the model were low, which indicate that there have to be more factors which influence consumers’ true WTP. So, according to the results of this study, given that the hypothetical bias also plays a major role in answering questions related to marketing metrics, managers should scientifically test these marketing metrics before they accept them as predictors of consumers’ true WTP.

Keywords: Willingness to Pay, Hypothetical Bias, Marketing Metrics, Customer Satisfaction, Customer Loyalty, Net Promoter Score, Vickrey Auction, Gift Card, Brand

(6)

1. Introduction

In the current literature many different marketing metrics are used to predict future consumer behavior (e.g., Morgan and Rego, 2006; Keiningham, Cooil, Andreassen, and Aksoy, 2007). These different marketing metrics have been widely embraced and adopted by managers in firms around the world. However, recent studies have shown that managers should quit customer satisfaction monitoring and the fact that they focus solely on customer

recommendation metrics are misleading and potentially harmful for the firm (Morgan & Rego, 2006). The study of Keiningham et al. (2007) has demonstrated that it is necessary to strictly and scientifically test different marketing metrics before their widespread adoption.

Furthermore, another common problem is the hypothetical bias, which refers to differences in reaction between situations wherein the consequences are hypothetical or wherein they are real (Harrison, 2006). Stated-preference approaches may lead to overpriced products and services compared to consumers’ true willingness to pay (hereinafter WTP). Revealing consumers’ true WTP is a significant value for firms in estimating demand for specific products of a brand and in designing optimal price schedules (Wertenroch & Skiera, 2002).

The true WTP of a specific brand could be revealed with the use of the experimental Vickrey (1961) auction. Vickrey auction is a second-price, sealed-bid auction metric in which the purchase price is determined by the other participants’ bids (Wertenbroch & Skiera, 2002). In a second-price auction the highest bidder will gain the object, but will pay the second-highest bid price (Hoffman, Menkhaus, Chakravarti, Field, and Whipple, 1993). So, the winning bidder’s price will be determined by the offer of the second-highest bidder.

By adding different marketing metrics to the experimental Vickrey auction, the kind of relationship and the strength of the relationship between the marketing metrics and the true WTP will be disclosed. This leads to an understanding of which marketing metric(s) should

(7)

be used for revealing consumers’ true WTP for a specific brand and could help firms in their pricing strategy.

This leads to the following research question: ‘What kind of relationship exists between marketing metrics and consumers’ true willingness to pay for a brand at the point of purchase and how strong is this relationship?’

To answer this research question, the following sub-questions arise:

• Which differences arise between consumers’ hypothetical WTP and consumers’ true WTP?

• Which different marketing metrics have been used in existing literature for revealing consumers’ WTP?

• What kind of relationship exists between the marketing metrics and consumers’ true WTP?

• How strong is the relationship between the different marketing metrics and consumers’ true WTP?

• Which marketing metrics have, next to the relationship with consumers’ true WTP, a strong relationship with each other?

• Are there also consumer characteristics which have an influence on consumers’ true WTP and how strong is this influence?

(8)

1.1 Contribution

The academic contribution is providing a new way of looking at an old problem, namely investigating what kind of relationship exists between different marketing metrics and consumers’ true WTP for a brand and the strength of this relationship by using experimental Vickrey auction. This leads to an understanding of which marketing metric(s) can best be used to reveal consumers’ true WTP.

Looking at the managerial contribution, to date companies do not know which specific marketing metric(s) to use for revealing consumers’ true WTP. Knowing consumers’ true WTP is crucial in estimating the demand for a brand and in designing optimal pricing schedules (Wertenbroch & Skiera, 2002). By investigating which marketing metric(s) have the strongest relationship with the true WTP, firm managers could adopt these marketing metric(s) and use them for the pricing theory of their products in order to determine a better price.

1.2 Structure of Thesis

The structure of the thesis is as follows. The first section of this master thesis consists of a review of the current literature, whereafter the literature gap and the conceptual model are presented. Subsequently the methodology and the results of the research are explained. Finally, the discussion, conclusion, implications and recommendations for future research are given.

(9)

2. Literature Review

This chapter starts with an explanation of the concept of Willingness to Pay (WTP), which is the main focus and the dependent variable during this research. Subsequently, the

hypothetical bias, which refers to differences in response between situations in which

consequences are hypothetical or actual, will be discussed. This is one of the starting points of the research. Thereafter the different independent variables will be described. First, the

influence of consumer characteristics on WTP will be explained. Furthermore, the commonly used marketing metrics (Customer Satisfaction, Customer Loyalty and Net Promoter Score) will be illustrated. Thereafter the experimental auction, and especially the Vickrey Auction, will be clarified, which is the main marketing research technique used. The literature review will be concluded with the literature gap and the conceptual model.

2.1 Willingness to Pay (WTP)

The Willingness to pay (WTP) is an extensively discussed topic in the economy, psychology and marketing areas. WTP denotes the maximum amount of money someone is willing to pay, trade or sacrifice in order to obtain a product or avoid an unpleasant product (Shogren, Shin, Hayes, & Kliebenstein, 1994). WTP is often discussed in combination with the concept of willingness to accept, which refers to an amount someone is willing to gain in order to sell a product or experience something displeasing (Shogren, Shin, Hayes, & Kliebenstein, 1994). According to Wertenbroch and Skiera (2002, p. 228), WTP is thus “the maximum price a buyer is willing to pay for a given quantity of a good”. It is a ratio-scaled measure of the subjective value the buyer assigns to that quantity of a good of a specific brand. The customer chooses that specific item from a set of alternatives for which his or her WTP exceeds

(10)

Consumers’ WTP could be valuable information for companies in estimating demand for products and in designing price schedules (Wertenbroch & Skiera, 2002). However, Wertenbroch and Skiera (2002) argue whether existing market research techniques (e.g. surveys) provide an incentive to consumers to reveal their true WTP and whether they simulate an actual point-of-purchase context. According to Auger and Devinney (2007), the simple survey questions in a traditional survey of ‘intention to purchase’, which estimates individuals WTP, are too ‘noisy’ to provide significant information and leads to exaggerating intentions.

This leads to the fact that WTP measuring methods can be divided in hypothetical and actual WTP. A survey, as mentioned above, is an example of hypothetical WTP. By

measuring the actual WTP, the participants are often incentivized to actually purchase the product. As an alternative to the survey as a market research technique, Wertenbroch and Skiera (2002) discuss the use of experimental Vickrey (1961) auctions. This market research technique will be explained at the end of the literature review.

2.2 Hypothetical bias

One of the primary reasons experimental (Vickrey) auctions have emerged as a useful tool in marketing research is the increasing recognition of the shortcomings associated with purchase intention questions and hypothetical WTP (Lusk, 2010). People tend to significantly overstate the amount of money they are willing to pay for goods in a hypothetical setting as compared to when real purchases are made (Little and Berrens, 2004, List and Gallet, 2001, and Murphy et al., 2005). Besides, it has also been recognized by Morgan and Rego (2006) that stated purchase intensions fail to correctly predict future business performance.

An example of the importance of hypothetical bias, which refers to differences in response between settings in which the consequences are hypothetical or real, is found in one

(11)

of the key results across the three studies of Wertenbroch and Skiera (2002). Wertenbroch and Skiera (2002) found that under Becker, DeGroot, and Marschak’s (also known as BDM) (1964), a well-known procedure to measure consumers’ WTP in market research, consumers stated substantially lower WTP than under hypothetical response formats. This suggests that stated-preference methods may lead managers to overprice their products compared with consumers’ true WTP. (Harrison and Rutström, 2002, in Wertenbroch and Skiera, 2002).

2.3 Consumer characteristics and Willingness to Pay (WTP)

For products to be successful in the marketplace, it is useful to identify the motivations behind people’s preferences and why certain people bid more for one product than for another (Lusk, 2010). The results of the study of Lusk (2010) show that there is a small relationship between bidding behavior and most of the demographic variables, such as gender, age, having children and college degree. The exception to this statement is the income of the participants.

Participants with a higher income tended to bid higher for a certain product than participants who have a lower income. However, this was only the case for one of the used products, which implies that there are also other factors, rather than income, that could explain why people bid high or low on a particular product (Lusk, 2010).

2.4 Customer Satisfaction and Willingness to Pay (WTP)

Customer satisfaction (CS) has become an important focus of corporate strategy (Homburg, Koschate, & Hoyer, 2005). CS is a post hoc evaluation of consumption experience (Oliver, 1980). A CS evaluation can be completely specific in nature, for example an experience of a single transaction or particular attribute. It may also be cumulative, which means based on all previous experiences with a product or service of a specific brand (Anderson and Fornell, 1994, in Anderson, 1996). Research supports the idea about the positive relationship between

(12)

CS and financial performance (e.g., Anderson, Fornell, and Rust, 1997; Rust and Zahorik, 1993). Anderson, Fornell and Lehmann (1994, p. 63) investigated this relationship on data collected from the Swedish Customer Satisfaction Index, and they found that “firms that actually achieve high customer satisfaction also enjoy superior economic returns.”

However, according to Homburg, Koschate and Hoyer (2005) an important question is whether CS also affects the consumers’ WTP. They argue that this is an important

relationship, because price is an essential factor in the profit comparison and is therefore directly linked to profitability. Homburg, Koschate and Hoyer (2005) have researched in their study whether there is, at the individual level, a (positive) relationship between CS and WTP. This has not been investigated before in this context. Nevertheless, they argue that this information can have important implications for pricing practices. The findings of the study reveal the existence of a strong, positive impact of CS on WTP (Homburg, Koschate, & Hoyer, 2005).

The most commonly used marketing metric for customer satisfaction is a firm’s “Top 2 Box” satisfaction score, which is “the proportion of customers rating their overall

satisfaction on the two highest-scoring points on the most commonly used five-point scale” (Morgan and Rego, 2006, p. 426).

2.5 Customer Satisfaction, Customer Loyalty and Predicting Business

Performance

Early views of brand loyalty focused on repeat purchase behavior (Srinivasan, Anderson, and Ponnavolu, 2002). There are many different definitions of brand loyalty in the last decades. According to Assael (1992, p. 87), brand loyalty is “a favorable attitude toward a brand resulting in consistent purchase of the brand over time.” This definition of brand loyalty was also supported by Keller (1993, in Srinivasan, Anderson, and Ponnavolu, 2002), who

(13)

suggested that loyalty is present when favorable attitudes for a brand are demonstrated in repeated buying behavior. Gremler (1995, in Srinivasan, Anderson, & Ponnavolu, 2002) stressed that both the attitudinal as well as behavioral dimensions need to be incorporated in any measurement of customer loyalty. In other words, loyalty will always be measured in two ways, because there are clear differences between attitudinal and behavioral customer loyalty.

A common mistake in the literature is the mix-up between customer satisfaction and customer loyalty. Customer satisfaction and customer loyalty are not surrogates for each other (Bloemen & Kasper, 1995; Oliver, 1999, in Shankar, Smith, & Rangaswamy, 2002). It is possible for a customer to be loyal without being highly satisfied (e.g., when there are just a few choices) and to be highly satisfied and but not being loyal (e.g., when many alternatives are available). Customer loyalty is an important marketing metric for predicting the future profitability of a firm (Srinivasan, Anderson, & Ponnavolu, 2002).

Managers observe performance on marketing metrics that they assume to be superior indicators of future consumer behavior (e.g. Hauser et al., 1994; Ittner & Larcker, 1998, in Morgan & Rego, 2006). Firms collect feedback data via customer surveys using measures of overall satisfaction, behavioral loyalty intentions such as repurchase likelihood and likelihood to recommend, and actual loyalty behaviors such as making recommendations (e.g., Griffin et al., 2012; Morgan et al., 2005).

Morgan and Rego (2006) empirically addressed the key question whether the most commonly used and widely supported customer satisfaction and loyalty metrics are the most valuable in revealing consumers’ WTP. They found that different customer satisfaction

metrics, the percentage of customers complaining and the repurchase likelihood loyalty metric are valuable marketing metrics. However, they also found that two widely supported

marketing metrics, which use recommendation behavior data, may have little or no predictive value. These results gave new empirical insights into the relationship between customer

(14)

satisfaction, customer loyalty and future consumer behavior (Morgan & Rego, 2006). However, these latest results are in contrast with earlier results about the different marketing metrics.

2.6 Word of Mouth (WOM) and Net Promoter Score

Another commonly used marketing metric is the Net Promoter Score. Most managers became aware of the marketing metric Net Promoter after the release of a 2003 Harvard Business Review article titled "The One Number You Need to Grow." (Keiningham, Cooil,

Andreassen, & Aksoy, 2007). The overall message of the article is that measurement of customer satisfaction and customer loyalty does not help firms to predict future consumer behavior and growth of the firm. Instead, according to the article, word of mouth (WOM) is the metric that is linked to growth. However, the WOM metric must be constructed in a specific way to calculate a Net Promoter Score (Reichheld, 2003). First, the survey respondents are asked to rate their likelihood of recommending a company. Next, the proportion of respondents rating the firm a 6 or less, called "detractors", are subtracted from the proportion of respondents rating the firm a 9 or 10, called "promoters". This difference represents a firms’ Net Promoter Score (Keiningham et al.,2007).

Recently there has been a discussion about whether this extensively embraced and adopted Net Promoter metric is superior. Keiningham et al. (2007) found no support for the claim that Net Promoter is the “single most reliable metric of a company’s ability to grow” (Keiningham et al., 2007, p. 45). Instead, they found that the Net Promoter Score does not perform better for the data under investigation than the American Customer Satisfaction Index (ACSI), a metric that was investigated and found not to correlate with growth (Reichheld, 2004, in Keiningham et al., 2007). They believe that their research demonstrates the necessity of scientifically testing different marketing metrics before their widespread adoption as

(15)

predictors of consumers’ WTP (Keiningham, Vavra, and Aksoy, 2006, in Keiningham et al., 2007).

2.7 Experimental Auction

The use of experimental auction is a marketing metric that has gained importance in recent years of studying consumer preferences (Lusk, 2010). In an experimental auction, people bid to buy real products using real money in a setting that provide incentives for people to truthfully reveal their WTP. The bidding environment stimulated individual thinking on their real value for the goods. The bids obtained in the experimental auctions are interpreted as the maximum amount people are willing to pay for a specific product of a brand, and as such, experimental auctions can be combined with traditional marketing metrics to yield

measurements of the desirability of products using a money metric (Lusk, 2010).

Experimental auctions are different than other marketing metrics. They are distinctive in the fact that they are not hypothetical (Lusk, 2010). Questions like ‘which brand will consumers like the best?’ and ‘which new brand or line extension will be most profitable?’ are questions that arise because companies want to know what would happen in a market and the future business performance they could expect. Traditional marketing metrics have been trying to address these questions by constructing hypothetical markets. As mentioned before, people tend to significantly overstate the amount of money they are willing to pay for goods in a hypothetical setting. By contrast, experimental auctions are attempting to answer these questions by revealing consumers’ true WTP. Responses to purchase intention questions provide an indication of stated preferences, but bids in experimental auctions will reveal true preferences (Lusk, 2010).

(16)

2.8 Vickrey Auction

As described above, market researchers need metrics that both are applicable at the point of purchase as well as providing incentive-compatible estimates of WTP derived from real transactions (Wertenbroch & Skiera, 2002). To obtain this goal, Hoffman et al. (1993) advocate the use of experimental Vickrey auctions. Vickrey auction is a second-price, sealed-bid auction in which the purchase price is determined by the other participants’ sealed-bids

(Wertenbroch and Skiera, 2002). In a second-price auction the highest bidder gains the

specific product, but pays the second-highest bid price (Hoffman et al., 1993). Vickrey (1961) suggests that incentive compatibility is ensured if a given bid determines only whether the bidder has the right to buy the good that is auctioned off (Wertenbroch & Skiera, 2002).

To explain why this auction is incentive compatible, you have to consider the costs and benefits of misrepresenting your value for a specific product (Hoffman et al., 1993). When your bid is higher than the value you gave to the product, it could happen that you would win and the second-highest price is still less than your true WTP. In this case, you neither gain nor lose, because you will always pay the second-highest price. However, if you win and the second-highest price is more than your true WTP, you have to pay more than your true WTP. In the opposite situation, when you bid less than your true WTP, it could happen that you lose and the second-highest price is more than your true WTP. In this case you neither gain nor lose. However, if you lose and the second-highest price is less than or equal than your value of the product, you have lost the opportunity to purchase the object at a price you are willing to pay.

In conclusion, you never gain but you can lose, by not bidding exactly your true WTP (Hoffman et al., 1993). So, unlike other marketing metrics based on stated preference data, Vickrey auctions provide bidders with an incentive to reveal their true WTP (Wertenbroch & Skiera, 2002).

(17)

2.9 Literature Gap

In the current literature many different marketing metrics are used to predict future consumer behavior (e.g., Morgan & Rego, 2006; Keiningham et al., 2007). These different marketing metrics are extensively incorporated and accepted by managers around the world. However, many studies, e.g. the study of Morgan and Rego (2006) and the study of Keiningham et al. (2007), have shown that managers should not rely on the investigation of marketing metrics such as customer satisfaction and customer loyalty. Besides, the fact that they focus merely on customer recommendation metrics in predicting future consumer behavior is misleading and potentially harmful for the company (Morgan & Rego, 2006). The study of Keiningham et al. (2007) has demonstrated that it is necessary to thoroughly test different marketing metrics before their widespread adoption.

Another common problem mentioned in the literature is the hypothetical bias. Stated-preference methods may lead managers to overprice their products and services compared to the consumers’ true WTP. Revealing consumers’ true WTP is a significant value for

companies in estimating demand for a brand and in designing an optimal pricing strategy (Wertenroch and Skiera, 2002).

By using experimental Vickrey auction with a gift card as product, the true WTP of consumers for a specific brand will be revealed, which means the true behavior of consumers, compared with the hypothetical behavior. By adding different marketing metrics to the

experimental Vickrey auction, the kind of relationship and the strength of this relationship between marketing metrics and consumers’ true WTP will be disclosed. This leads to an understanding of which marketing metric(s) should best be used for revealing consumers’ true WTP for a specific brand and could thus help firms in their pricing strategy.

So, the research objective is to identify what kind of relationship exist between the different used marketing metrics and consumers’ true WTP for a specific brand at the point of

(18)

purchase and the strength of this relationship. This leads to the following research question: ‘What kind of relationship exists between marketing metrics and consumers’ true willingness to pay for a brand at the point of purchase and how strong is this relationship?’

To answer this research question, there are formulated a number of hypotheses:

Based on the studies of Wertenbroch and Skiera (2002), the first hypothesis is: * H1: The hypothetical WTP of consumers will be higher than the true WTP of consumers.

The results of the study of Lusk (2010) leads to the second hypothesis: * H2: Consumers with a higher income will have a higher WTP.

In accordance with the study of Homburg, Koschate and Hoyer (2005), the third hypothesis is:

* H3: Customer satisfaction will have a positive relationship with consumers’ true WTP. According to the study of Srinivasan, Anderson, and Ponnavolu (2002) and the study of Morgan and Rego (2006), the following two hypotheses are:

* H4a: Behavioral customer loyalty will have a positive relationship with consumers’ true WTP.

* H4b: Attitudinal customer loyalty will have a positive relationship with consumers’ true WTP.

As indicated above, the current literature is ambiguous about the performance of the Net Promoter Score. To investigate this performance, a position has to be taken. In line with the study of Reichheld (2003), the fifth hypothesis is as follows:

(19)

2.10 Conceptual Model

Figure 1 – Conceptual Model

All relations between the variables are visualized in Figure 1. All those hypotheses (H2, H3, H4a, H4b, H5), as described in this chapter, will be examined during this research.

In addition, the difference between consumers’ hypothetical WTP and consumers’ true WTP for a specific brand, hypothesis 1 (H1), will be investigated, as this is the starting point of the research. Bid on Gift Card (DV) (True WTP, Vickrey Auction) Marketing Metric: Consumer characteristic (Income) (IV) H2 Marketing Metric: Customer Satisfaction (IV) H3 Marketing Metric: Behavioral Customer Loyalty (IV) H4a Marketing Metric: Attitudinal Customer Loyalty (IV) H4b Marketing Metric: Net Promoter Score (IV) H5

(20)

3. Methodology

After thoroughly discussing the literature and presenting the hypotheses, the research design and the methods used for the analytical strategy will be explained. First the methods will be discussed, followed by a brief discussion about the used product for this research. Hereafter the research design and different variables will be described. Finally, the experiment

procedure, the sample frame and the data coding and categorization will be explained.

3.1 Methods

The research of this thesis is both of exploratory and explanatory nature. This study is a multi-method quantitative research that uses primary source data. The main data will be collected through an experiment with a Vickrey auction, which will be conducted using the online auction site Veylinx. In this second-price, sealed-bid auction, participants will not know what others bid for the product. However, the participant with the highest bid will only pay the second highest price. So, the winning bidder’s price will be determined by the bid of the second-highest bidder.

As discussed in the literature review, the true WTP of consumers is the dependent variable of this research. Using a Vickrey auction will help to reveal consumers’ true WTP for products by giving them an incentive to bid their true WTP, assuring that if they win, they will pay less than their maximum price. This experiment is preferred over a research survey. During a survey, people often tend to answer what they think is most idealistic, i.e. how they expect to act in a certain situation. However, when they are actually in the situation, most people diverge from what they say they would do, which refers to the hypothetical bias. So, using a Vickrey auction makes it possible to measure consumers’ true WTP by placing them in a real buying situation instead of just giving an answer about their hypothetical WTP.

(21)

However, in order to test whether the hypothetical bias is also the case for the product used in this study, the hypothetical WTP will be measured too. Other respondents than those who participated in the online action will be asked about their hypothetical WTP for the used product. The question about their hypothetical WTP will be asked face-to-face, to ensure they give their actual hypothetical WTP without discussing their answers with others.

3.2 Product Information: Gift Card

The total sales of gift cards has increased since 1997, exceeding $60 billion in 2005 and around $80 billion in 2007 (Offenberg, 2007). Gift cards used to be provided by large chain store companies, but also local businesses increasingly offer these specific products. Gift cards are an intermediate option between two alternatives: purchasing a physical gift, which could possibly be returned or exchanged, versus giving cash.

According to Offenberg (2007), gift cards have some specific features. The more potential features a gift card has, for example if a gift card belongs to a store that offers more general use items or has a larger product variety, the less welfare loss it should create. With welfare loss Offenberg (2007) meant the difference between the average face value of the gift card being sold and the average sale price, divided by the average face value. “The face value of the gift cards is the amount of the store’s credit currently stored on the card” (Offenberg, 2007, p. 232). Another specific feature is the size of gift cards. According to Offenberg (2007), a smaller amount in the form of a gift card is less likely to create a welfare loss. However, Offenberg (2007) couldn’t reveal the highest bidder’s willingness to pay, because she used an ascending auction.

Chiou and Pate (2010) investigate the pricing of retail gift cards on eBay. They found substantially less price dispersion for gift cards than previously has been documented in online markets for consumer goods. Compared to online prices for consumer products (e.g.

(22)

books and CD’s), in which the difference between the minimum and maximum price is between 25-40% of the mean price (Ellison and Ellison, 2005; Brynjolfsson and Smith, 2000, in Chiou and Pate, 2010), the price range for gift cards is between 4-8% of the overall mean price across the different categories of cards (Chiou and Pate, 2010). The results of the study of Chiou and Pate (2010) suggest that the low level of price dispersion of the gift cards may be due to the fact that gift cards represent a truly homogenous good, as the characteristics of the gift card are identical and known to all consumers of the cards.

During the experimental Vickrey auction, a €100 gift card of Bol.com will be used as the product. The major reason for using gift cards is the investigation of the relationship between the different marketing metrics and the true WTP of a brand, not of a specific product. This will give more insight in the popularity of a brand and the different marketing metrics that are able to reveal this, instead of just the need for a specific product. Another important reason for the use of gift cards is the substantially less price dispersion for gift cards in comparison with the price dispersion for other consumer goods (Chiou and Pate, 2010).

(23)

3.3 Research Design & Variables

In this paragraph, the research design and the different variables of this research will be explained.

Figure 2 - Research Design

During the experimental Vickrey auction, people first have to give their bid for the product, a €100 Bol.com gift card, which will reveal their true WTP. The €100 Bol.com gift card which is shown to the participants of the auction is presented in Appendix 1.

As mentioned before, the true WTP of the participants is the dependent variable during this research. The participants have six minutes to give their bid for the product, after which they have to answer five different questions (see Figure 2 – Research Design). Each question will test another hypothesis, and thus refer to another marketing metric (see Table 1 –

Research Design). Bid on Gift Card (DV) (True WTP) Q1: Consumer characteristic (income) (IV) _Q2: Customer Satisfaction (IV) Q3: Behavioral Customer Loyalty (IV) Q4: Attitudinal Customer Loyalty (IV) Q5: Net Promoter Score (IV)

(24)

Marketing Metric (IV)

Source Question Scale type

Q1: Consumer Characteristic (Income) *1 Written by researcher using literature on consumer characteristics and consumers’ true WTP (Lusk, 2010).

Income

What is your annual gross income? Multiple-choice: 1) Less than €10.000 2) Between €10.001 - €25.000 3) Between €25.001 - €50.000 4) Between €50.001 - €75.000 5) More than €75.000

6) Prefer not to say Q2: Customer

Satisfaction

Adapted from research Morgan and Rego (2006).

Customer Satisfaction All in all, how satisfied or unsatisfied are you with Bol.com?

Five-point Likert:

Very unsatisfied (1) to very satisfied (5) and a 6th option: ‘Never bought at Bol.com’ Q3: Behavioral Customer Loyalty Written by researcher using literature on customer loyalty (Kamakura et al., 2002; Morgan and Rego, 2006).

Behavioral Customer Loyalty

When did you last purchased a product from Bol.com?

Multiple-choice: 1) Never

2) More than 1 year ago 3) Between 6 - 12 months ago 4) Less than 6 months ago

Q4: Attitudinal Customer Loyalty

Written by researcher using literature on customer loyalty (Morgan and Rego, 2006).

Attitudinal Customer Loyalty

How likely are you to repurchase at Bol.com in the future (besides the possible purchase of this gift card)?

Five-point Likert:

Very unlikely (1) to very likely (5)

Q5: Net Promoter Score

Adapted from research Keiningham et al. (2007).

Net Promoter Score How likely is it that you would recommend Bol.com to a friend or colleague?

Eleven-point Likert:

Very unlikely (0) to very likely (10)

Table 1 - Research Design

*1: Other consumer characteristics of the respondents, e.g. demographic variables such as age, gender and college degree, have already been measured with the online auction site Veylinx. These additional data will also be used during the investigation.

(25)

Due to the fact that the bid on the gift card (which reveals the true WTP of consumers) and the additional five questions (which give information about the five different marketing metrics) are linked to the same respondent, it is possible to investigate the relationship between the true WTP of the participants and the particular marketing metric.

As mentioned in paragraph 3.1, besides the online auction, the hypothetical WTP will also be measured. This variable is measured during a face-to-face interview with 60

respondents. Using literature on hypothetical WTP (Wertenbroch and Skiera, 2002), the following question will be asked to the respondents: ‘What is the maximum amount of money you are willing to pay for a €100 Bol.com Gift Card?’. By comparing the mean of the

answers on this intention question (consumers’ hypothetical WTP) with the mean of the bid amount on the €100, - Bol.com Gift Card (consumers’ true WTP), the hypothetical bias could be measured.

3.4 Experiment Procedure

As described before, the data for this research is collected through the use of the online auction named ‘Veylinx’. Veylinx is a Dutch online SPSB experimental Vickrey auction platform which currently has over 9.000 members. Veylinx bases its results on real buying situations, which makes it possible to measure consumes’ true WTP instead of their hypothetical WPT in hypothetical situations. By being concrete and using real money, Veylinx studies the target audience and tests consumers’ true WTP and could thus predict future consumer behavior.

The experiment procedure is as follows. Firstly, the panelists of the platform are recruited on a voluntary basis. Once registered with name, age, gender and e-mail address, the panelists receive an e-mail with an invitation to participate in an online auction. Since the product will be displayed only when the panelist participates, self-selection is not possible.

(26)

The panelists are asked to provide a legally obligated bid that is equal to their maximum willingness to pay. To all members the legal obligation to pay has been shown, which forces them to accept the auctions’ terms and conditions and makes them attentive that the bids they give are binding. After the bid, the participants of the auction have to answer the different multiple choice questions, related to the auction. Once the auction finishes, all participants receive an e-mail revealing the winning and the second highest bid. The second highest bid is mentioned because the participant who has won pays the price of the second highest bid. The table summary of the experiment procedure is presented in Appendix 2.

3.5 Sample Frame

As mentioned above, a total number of about 9.000 people enrolled in the online auction site Veylinx. Of these Veylinx members, around 5.000 are active, meaning they bid regularly, approximately once a month. The member base of Veylinx is represented by Dutch society in terms of gender, age and socio-economic profile. During this study, 1172 members

participated in the auction (N = 1172).

However, firstly 14 participants were excluded because they only bid on the gift card and didn’t answer (all) five questions. Only cases which had no missing data in any variable were analyzed. Besides, 2 participants were excluded because they had an age of 109 and 116 years old, which is very unbelievable. So, the final sample size of this study is N = 1156.

The respondents consisted of 558 men (48.3%) and 598 women (51.7%) and the age ranged from 15 to 83 years old, with an average of 41.57 years. Regarding their education, 831 of the 1156 participants had answered a question about their education in previous auctions. 0.6% of these 831 participants had primary school, 5.0% had secondary education, 17.0% had tertiary education, 24.7% had a bachelor’s degree, and the remaining 24.6% had a master’s degree. So, almost 50% of those participants had a bachelor’s or master’s degree.

(27)

Finally, 683 of the 1156 participants replied in previous auctions the question whether they were a student or not. Of those 683 participants, 10.7% were students.

3.6 Data coding and categorization

Besides excluding cases with missing values, some data of the dataset must be recoded and some values in the variables have to be converted into categories. Below an overview of the adjustments for each variable to correct the data for the analysis:

- Bid amount (DV): The variable ‘bid amount’ was already correct to use for the analysis (continuous data).

- Consumer characteristic (income) (IV): Value 6 (prefer not to say) has been coded as ‘missing value’. Furthermore, different values are categorized: Old value 1 (less than

€10.000) and old value 2 (between €10.001 - €25.000) are categorized into the new category ‘low income’ (value 0); Old value 3 (between €25.001 - €50.000) is coded into the new category ‘average income’ (value 1); Old value 4 (between €50.001 - €75.000) and old value 5 (more than €75.000) are categorized into the new category ‘high income’ (value 2). These new categories have been chosen for a better distribution of respondents across categories (Low income N = 391; Average income N = 307; High income N = 198). In total 260

respondents have answered ‘prefer not to say’, so they were coded as missing in this variable. - Customer satisfaction (IV): Value 6 (Never bought at Bol.com) has been coded as missing value. In total 50 respondents have answered ‘never bought at Bol.com’, so they were coded as missing in this variable. The 5pt Likert scale used to measure ‘customer satisfaction’ was already correct to use for the analysis (ordinal data).

- Behavioral Customer Loyalty (IV): The classification of categories used to measure ‘behavioral customer loyalty’ was already correct to use for the analysis (ordinal data).

(28)

- Attitudinal Customer Loyalty (IV): The 5pt Likert scale used to measure ‘attitudinal customer loyalty’ was already correct to use for the analysis (ordinal data).

- Net Promoter Score (IV): The Net Promoter Score is officially a 11pt Likert scale (values 0-10). However, within this scale there are three different levels/categories: ‘detractors’

(proportion of respondents rating the firm a 6 or less), ‘passives’ (proportion of respondents rating the firm a 7 or 8) and ‘promoters’ (proportion of respondents rating the firm a 9 or 10). So, the 11 values were categorized in three new categories: Value 0 to value 6 are categorized in the new category ‘detractors’ (new value 0), value 7 and value 8 are categorized in the new category ‘passives’ (new value 1) and value 9 and value 10 are categorized in the new

category ‘promoters’ (new value 2).

Furthermore, the data have been checked for any errors in the data set. For the dependent variable ‘Bid amount’ (true WTP), no outliers and errors were detected. This is because participants were not able to participate in the online auction without giving their bid on the gift card.

(29)

4. Results

In the previous chapter the methodology of the research was explained. This section provides an overview of the results of the data and the various analyzes that are made throughout the study. First the descriptive statistics are discussed, second the hypotheses are tested and finally a multiple regression analysis is performed. SPSS is used for analyzing the data.

4.1 Descriptive Statistics

Firstly, the descriptive statistics of the dependent variable ‘bid amount’ are shown in Table 3. The average bid was €37.77 with a standard deviation of €26.89. Furthermore, given the values of the mean, median and mode, it can be concluded that the distribution of the data is slightly negatively skewed. This is also reflected in the values of the percentiles.

Variable Mean Median Mode SD Min. Max. Percentiles

25 50 75 Bid amount

(in euro’s)

37.77 40.00 50.00 26.886 0.00 110.00 11.07 40.000 55.00

Table 3 - Descriptive statistics bid amount

Further descriptive statistics are presented in Table 4. Spearman’s correlation coefficient is a non-parametric statistic based on ranked data. It measures the degree of association between two variables and does not make any assumptions about the distribution of the data. Spearman’s correlation coefficient should be used when the variables are

measured on a scale that is at least ordinal (Field, 2013). Cohen’s standard will be used to evaluate the correlation coefficient to determine the strength of the relationship. Coefficients between .10 and .29 represent a weak correlation, coefficients between .30 and .49 represent a moderate correlation and coefficients above .50 represent a strong correlation.

(30)

Variables Mean SD 1 2 3 4 5 6 7 8 1. Bid amount (euro’s) 37.77 26.886 -

2. Gender 0.52 0.500 -.131** - 3. Age 41.57 14.049 -.156** .009 - 4. Income .78 .782 .153** -.240** .242** - 5. Customer satisfaction 4.13 .854 .115** .045 -.091** -.026 - 6. Behavioral customer loyalty 3.39 .906 .272** -.005 -.118** .145** .363** - 7. Attitudinal customer loyalty 4.14 1.019 .210** .034 -.068* .069* .598** .558** -

8. Net Promoter Score 1.30 .726 .173** .100** -.090** .027 .616** .462** .618** - **. Correlation is significant at the 0.01 level (2-tailed).

*. Correlation is significant at the 0.05 level (2-tailed).

Table 4 – Means, Standard Deviations, Nonparametric Correlations (Spearman’s rho)

As shown in Table 4, the bid amount, which refers to consumers’ true WTP, has a significant relationship with all other variables. However, all those coefficients show a weak correlation. The dependent variable bid amount has a weak negative relationship with the variables gender and age, but a weak positive relationship with the independent variable income and all marketing metrics. In comparison with the other marketing metrics, bid

amount has the strongest positive relationship with behavioral customer loyalty,

rs

= .272, p < .01, followed by the positive relationship with attitudinal customer loyalty,

rs

= .210, p < .01.

Furthermore, the variables age and gender have both a weak relationship with income, where the relationship between income and gender is negative,

rs

= -.240, p < .01, and the relationship between income and age is positive,

rs

= .242, p < .01. Because of the weak correlation between the variables age and gender with the independent variable income, these two variables will be used as control variables for hypothesis testing of the relationship between the independent variable income and the dependent variable bid amount.

(31)

The strongest correlations exist between the different marketing metrics; customer satisfaction has a strong positive relationship with the Net Promoter Score,

rs

= .616, p < .01, and attitudinal customer loyalty,

rs

= .598, p < .01, and a moderate positive relationship with behavioral customer loyalty,

rs

= .363, p < .01. Behavioral customer loyalty has a strong positive relationship with attitudinal customer loyalty,

rs

= .558, p < .01, and a moderate positive relationship with the Net Promoter Score,

rs

= .462, p < .01. Finally, attitudinal customer loyalty has a strong positive relationship with the Net Promoter Score,

rs

= .618, p < .01, which is also the strongest correlation in this matrix.

4.2 Hypothesis testing

After outlining the descriptive statistics of both the dependent variable ‘bid amount’ as well as the different independent variables, the hypotheses are tested. For all hypotheses, the Durbin-Watson, Tolerance and VIF have been checked.

Hypothesis 1 was stated as follows: ‘The hypothetical WTP of consumers will be higher than the true WTP of consumers.’ As shown in Table 3, the average bid of the true WTP of

consumers was €37.77 with a standard deviation of €26.89. However, as mentioned in chapter 3, in order to test whether the hypothetical bias is also the case for the product used in this study, the hypothetical WTP has been measured too. 60 different respondents than those who participated in the online auction were asked about their hypothetical WTP for the used product, see Appendix 3.

The descriptive statistics of the hypothetical WTP are shown in Table 5. The average bid was €57.23 with a standard deviation of €22.88.

(32)

Variable Mean Median Mode SD Min. Max. Percentiles

25 50 75 Hypothetical WTP

(in euro’s)

57.23 60.00 50.00 22.880 5.00 100.00 50.00 60.000 75.00

Table 5 - Descriptive statistics hypothetical WTP

So, according to the descriptive statistics of the bid amount (true WTP) and the hypothetical WTP of consumers, the difference is substantial (see Table 6). However, to test whether this difference is significant, an independent samples t-test has been conducted (see Table 7). Looking at Levene’s Test for Equality of Variances, F = 8.982; p = .003 ≤ .10, the equality of variances is rejected. Furthermore, looking at the t-test for equality of means, P-value .000 < .05, there is a significant difference between the bid amount (true WTP) and the hypothetical WTP of consumers. In other words, the hypothetical WTP of consumers is significant higher than the true WTP of consumers, so hypothesis 1 is supported.

Groups N Mean Std. Deviation Std. Error Mean

Bid Hypothetical WTP (in euro’s) 60 57.23 22.880 2.954

Bid True WTP (in euro’s) 1156 37.77 26.886 .791

Table 6 – Group Statistics Hypothetical WTP and True WTP

Levene’s Test t-test for Equality of Means

F Sig. t df Sig. (2-tailed) Mean Difference Std. Error Difference 95% Conf. Int. of the Difference Lower Upper Equal variances assumed 8.982 .003 5.505 1214 .000 19.464 3.536 12.527 26.401 Equal variances not assumed 6.365 67.742 .000 19.464 3.058 13.362 25.566

(33)

Hypothesis 2 was stated as follows: ‘Consumers with a higher income will have a higher WTP.’ In Figure 3 is visualized how the independent variable income is related to the dependent variable bid amount:

Figure 3 - Relationship Income - Bid amount

Hereafter, a hierarchical regression was performed to investigate the ability of the consumer characteristic ‘income’ to predict consumers’ true WTP (bid amount), after

controlling for gender and age. Since a clear linear relationship is shown between the different levels of income, with approximately the same increase of bid amount per level, no dummy variables were used.

In the first step of the hierarchical regression, two predictors were entered: gender and age (see Table 8). This model was statistically significant F (2,893) = 15.530; p < .001 and explained 3.4% of variance in bid amount. After entry of income at Step 2 the total variance explained by the model as a whole was 5.7% F (3,892) = 18.054; p < .001. The introduction of income explained additional 2.4% variance in bid amount, after controlling for gender and age (R2 Change = .024; F (1,892) = 22.358; p < .001). In the final model all three variables

€- €5,00 €10,00 €15,00 €20,00 €25,00 €30,00 €35,00 €40,00 €45,00 €50,00

low income average income high income

(34)

were statistically significant, where the unstandardized coefficient B of the independent variable income indicates that when your income level increases with 1 category, the bid amount will increase with €5.61. So, hypothesis 2 is supported. However, both the R2 as well as the Beta value of income are low (β = .162), which implies that income has a small impact on bid amount.

R R2 R2 Change B SE β t

Step 1 .183 .034**

Gender (0=male, 1=female) -6.439 1.785 -.119** -3.608

Age -.272 .062 -.143** -4.339

Step 2 .239 .057** .024

Gender (0=male, 1=female) -4.307 1.820 -.079* -2.366

Age -.330 .0063 -.173** -5.221

Income 5.610 1.186 .162** 4.728

Note. Statistical significance: *p <.05; **p <.001

Table 8 – Hierarchical Regression Model of Income – Bid amount

Hypothesis 3 was stated as follows: ‘Customer satisfaction will have a positive relationship with consumers’ true WTP.’ In Figure 4 is visualized how the independent variable customer satisfaction is related to the dependent variable bid amount:

Figure 4 – Relationship Customer satisfaction - Bid amount € 0,00 € 5,00 € 10,00 € 15,00 € 20,00 € 25,00 € 30,00 € 35,00 € 40,00 € 45,00 very unsatisfied

unsatisfied neutral satisfied very satisfied

(35)

A linear regression was performed to investigate the ability of customer satisfaction to predict consumers’ true WTP (bid amount). In Table 9 is shown that this model was

statistically significant F (1,1104) = 18.760; p < .001 and explained 1.7% of variance in bid amount. So, hypothesis 3 is supported. The unstandardized coefficient B indicates that when your customer satisfaction level increases with 1 category, the bid amount will increase with €4.07. However, both the R2 as well as the Beta value of customer satisfaction are really low (β = .129), which implies that customer satisfaction has a weak impact on bid amount.

R R2 B SE β t

Customer satisfaction .129 .017** 4.074 .941 .129** 4.331

Note. Statistical significance: **p <.001

Table 9 –Regression Model of Customer satisfaction – Bid amount

Hypothesis 4a was stated as follows: ‘Behavioral customer loyalty will have a positive relationship with consumers’ true WTP.’ In Figure 5 is visualized how the independent variable behavioral customer loyalty is related to the dependent variable bid amount:

Figure 5 – Relationship Behavioral customer loyalty - Bid amount € 0,00 € 5,00 € 10,00 € 15,00 € 20,00 € 25,00 € 30,00 € 35,00 € 40,00 € 45,00

never more than 1 year ago between 6 to 12 months ago

less than 6 months ago

(36)

A linear regression was performed to investigate the ability of behavioral customer loyalty to predict consumers’ true WTP (bid amount). Since a clear linear relationship is shown between the different levels of behavioral customer loyalty, no dummy variables were used. In Table 10 is shown that this model was statistically significant F (1,1154) = 87.369; p < .001 and explained 7.0% of variance in bid amount. So, hypothesis 4a is supported. The unstandardized coefficient B indicates that when your behavioral customer loyalty level increases with 1 category, the bid amount will increase with €7.88. However, both the R2 as well as the Beta value of behavioral customer loyalty are low (β = .265), which implies that behavioral customer loyalty has not a major impact on bid amount.

R R2 B SE β t

Behavioral customer loyalty .265 .070** 7.875 .843 .265** 9.347

Table 10 –Regression Model of Behavioral customer loyalty – Bid amount

Hypothesis 4b was stated as follows: ‘Attitudinal customer loyalty will have a positive relationship with consumers’ true WTP.’ In Figure 6 is visualized how the independent variable attitudinal customer loyalty is related to the dependent variable bid amount:

Figure 6 – Relationship Behavioral customer loyalty - Bid amount € 0,00 € 5,00 € 10,00 € 15,00 € 20,00 € 25,00 € 30,00 € 35,00 € 40,00 € 45,00

very unlikely unlikely neutral likely very likely

(37)

A linear regression was performed to investigate the ability of attitudinal customer loyalty to predict consumers’ true WTP (bid amount). In Table 11 is shown that this model was statistically significant F (1,1154) = 54.741; p < .001 and explained 4.5% of variance in bid amount. So, hypothesis 4b is supported. The unstandardized coefficient B indicates that when your attitudinal customer loyalty level increases with 1 category, the bid amount will increase with €5.62. However, both the R2 as well as the Beta value of behavioral customer loyalty are low (β = .213), which implies that attitudinal customer loyalty has a weak impact on bid amount.

R R2 B SE β t

Attitudinal customer loyalty .213 .045** 5.618 .759 .213** 7.399

Table 11 –Regression Model of Attitudinal customer loyalty – Bid amount

Hypothesis 5 was stated as follows: ‘The Net Promoter Score will have a positive relationship with consumers’ true WTP.’ In Figure 7 is visualized how the independent variable Net Promoter Score is related to the dependent variable bid amount:

Figure 7 – Relationship Net Promoter Score - Bid amount € 0,00 € 5,00 € 10,00 € 15,00 € 20,00 € 25,00 € 30,00 € 35,00 € 40,00 € 45,00

detractors passives promoters

(38)

A linear regression was performed to investigate the ability of the Net Promoter Score to predict consumers’ true WTP (bid amount). Since a clear linear relationship is shown between the different levels of the Net Promoter Score, no dummy variables were used. In Table 12 is shown that this model was statistically significant F (1,1154) = 37.267; p < .001 and explained 3.1% of variance in bid amount. So, hypothesis 5 is supported. The

unstandardized coefficient B indicates that when your NPS level increases with 1 category, the bid amount will increase with €6.55. However, both the R2 as well as the Beta value of the Net Promoter Score are really low (β = .177), which implies that the Net Promoter Score has a weak impact on bid amount.

R R2 B SE β t

Net Promoter Score .177 .031** 6.548 1.073 .177** 6.105

Table 12 –Regression Model of Net Promoter Score – Bid amount

4.3 Additional analysis

Besides the analyses that have been done for the investigation of the hypotheses, additional analyses are conducted.

A multiple regression has been done with all the independent variables (income, customer satisfaction, behavioral customer loyalty, attitudinal customer loyalty and Net Promoter Score) and the dependent variable bid amount. A stepwise backward method was used in the multiple regression.

In Table 13 the results of the multiple regression with the stepwise backward method are shown.

(39)

R R2 Adjusted R2 R2 Change B SE β t

Model 1

*All independent variables

.260 .068** .062

Income 3.830 1.161 .110* 3.297

Customer Satisfaction .275 1.369 .009 .201

Behavioral Customer Loyalty 5.827 1.410 .166** 4.133

Attitudinal Customer Loyalty 1.342 1.359 .046 .987

Net Promoter Score 1.647 1.733 .042 .950

Model 2

*Cust. Sat. excluded

.260 .068** .063 .000

Income 3.812 1.159 .110* 3.293

Attitudinal Customer Loyalty 1.423 1.298 .049 1.096

Net Promoter Score 1.776 1.608 .045 1.105

Model 3

*Att. Cust. Loy. excluded

.257 .066** .063 -.001

Income 3.844 1.159 .111* 3.317

Net Promoter Score 2.603 1.421 .067 1.832

Note. Statistical significance: *p <.01; **p <.001

Table 13 –Multiple Regression (stepwise backward method)

As shown in Table 13, in the backward method all the independent variables are placed as predictors in the model after which the contribution of each variable is calculated by looking at the significance value of the t-test for that variable. If an independent variable is not making a statistically significant contribution to how well the model predicts the dependent variable bid amount, it is removed from the model and the model is re-estimated for the remaining independent variables. The influence of the remaining independent variables is then reassessed (Field, 2013).

The multiple regression with the stepwise backward method was executed to investigate the relative contribution of each independent variable in this exploratory model where all the different independent variables are included. The backward method was used rather than the forward method to minimize suppressor effects, “which occur when a

(40)

predictor has a significant effect but only when another variable is held constant” (Field, 2013, p. 323).

The first model, where all the independent variables were included, was statistically significant F (5, 856) = 12.413; p < .001 and explained 6.8% of the variance in bid amount. However, as shown in Table 11, only the independent variables income and behavioral

customer loyalty were statistically significant (income: β = .110, p < .01; behavioral customer loyalty: β = .166, p < .001).

In model 2, the independent variable customer satisfaction was excluded (β= .009, p > .05). Also this second model was statistically significant F (4, 857) = 15.523; p < .001 and explained still 6.8% of the variance in bid amount. However, still, only the independent variables income and behavioral customer loyalty were statistically significant (income: β = .110, p < .01; behavioral customer loyalty: β = .167, p < .001).

In the final model, both independent variables customer satisfaction and attitudinal customer loyalty were excluded (customer satisfaction: β = .022, p > .05; attitudinal customer loyalty: β = .049, p > .05). This final model was also statistically significant F (3, 858) = 20.293; p < .001 and explained 6.6 % of the variance in bid amount. However, the adjusted R2 of model 3 is still .063, so the model explained still 6.3 % of the variance in bid amount if the model had been derived from the population from which the sample was taken. In the final model two out of three predictor variables were statistically significant, with behavioral customer loyalty recording a higher Beta value (β = .185, p < .001) than income (β = .111, p < .01).

(41)

5. Discussion

The findings of this study extend existing knowledge of the relationship between common used marketing metrics and consumers’ true willingness to pay (WTP). First the results of this study will be interpreted and they will be linked to the theory, followed by an explanation of both the theoretical and managerial contribution. Finally, the limitations and implications for future research are specified.

5.1 Findings

The purpose of this study was twofold: (1) to find out what kind of relationship exists between the different, often used marketing metrics and the true WTP of consumers and the strength of this relationship by using an online Vickrey auction; (2) to investigate the

relationship between consumers’ true WTP and gift cards, which will give more insight in the value of a brand instead of just the need for a specific product. By using a €100 Bol.com Gift card as product in the online Vickrey auction and adding five questions, each measuring a different marketing metric, these two goals have been examined.

In scientific marketing literature multiple marketing metrics are used to predict future consumer behavior, and those marketing metrics are widely embraced and adopted by

managers around the world (Morgan and Rego, 2006). However, several studies, including the study of Keiningham, Cooil, Andreassen, and Aksoy (2007), have showed that it is essential to strictly and systematically test marketing metrics before their widespread adoption.

The first hypothesis states that the hypothetical WTP of consumers will be higher than the true WTP of consumers. According to the studies of Little and Berrens (2004), List and Gallet (2001) and Murphy et al. (2005), people tend to significantly exaggerate the amount of money they are willing to pay for products in a hypothetical setting as compared to when real

(42)

purchases are made. However, this was only experienced for regular products, not for gift cards. The results of this study indicate that these hypothetical bias also applies to gift cards. The average bid on the €100 Bol.com gift card during the online auction, so the true WTP of consumers, was €37.77, while the average bid on the €100 Bol.com gift card in the

hypothetical setting, so the hypothetical WTP of consumers, was €57.23. In accordance with the conducted independent samples t-test, there is significant evidence to support that the hypothetical WTP of consumers is higher than the true WTP of consumers, even for gift cards.

The second hypothesis states that consumers with a higher income will have a higher WTP. According to the results of the study of Lusk (2010), participants with a higher income tended to bid higher for products than participants with a lower income. The results of this study indicate indeed a significant linear relationship between the income level of the participants and the bid amount. However, both the R2 as well as the Beta value of income were really low, which implies that the impact of income on bid amount is weak. In the multiple regression analysis with the stepwise backward method, the relationship between income and bid amount was still statistically significant in all the three models. This suggests the existence of a statistically significant relationship between income and bid amount, but given the low Beta value and R2, there are many other factors influencing the bid amount which were not captured in this model.

The third hypothesis states that customer satisfaction (CS) will have a positive relationship with consumers’ true WTP. According to the findings of the study of Homburg, Koschate and Hoyer (2005), there will be a strong, positive impact of CS on WTP. However, the results of this study show a different outcome. Looking at the linear regression that was performed to investigate the ability of customer satisfaction to predict consumers’ true WTP, this model is indeed statistically significant. However, both the R2 as well as the Beta value

The relationship between marketing metrics and consumers’ true willingness to pay for a brand at the point of purchase : what kind of relationship exists between marketing metrics and consumers’ true willingness to pay

Master Thesis – MSc Business Administration – Marketing Track