Does personalization add value? An experiment on the efficiency of personalization in a loyalty program
Final Version June 25, 2021
Faculty of Economics and Business University of Amsterdam
MSc Business Administration Track: Digital Marketing Sophie Klijn
Student number: 12683655 Supervisor: Dr. N. Bombaij
EBEC approval number: 20210330120300
Statement of Originality
This document is written by Sophie Klijn 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 the completion of the work, not for the contents.
Table of content
Abstract ... 5
1. Introduction ... 6
2. Literature review ... 9
2.1 Loyalty programs ... 9
2.1.1 Reasons for adopting a loyalty program ... 9
2.1.2 Loyalty program designs ... 10
2.1.3 Consumers heterogeneity ... 12
2.2 Personalization ... 13
2.2.1 Advantages and disadvantages of personalization ... 13
2.2.2 Dimensions of personalization ... 14
2.2.3 Consumers heterogeneity ... 15
2.3 Marketing communication channels ... 16
2.3.1 Advantages and disadvantages of online versus offline communication channels . 16 2.3.2 Different types of communication channels ... 18
2.3.3 Consumer heterogeneity ... 19
2.4 Sales promotions and promotion sensitivity ... 20
2.4.1 Reasons for adopting sales promotions ... 20
2.4.2 Different types of sales promotions ... 21
2.4.3 Promotion sensitiveness ... 22
3. Conceptual framework ... 23
3.1 Personalization versus no personalization ... 24
3.2 Individualized versus segmented personalization ... 26
3.3 Communication channels ... 27
3.4 Promotion sensitiveness ... 29
4. Methodology ... 31
4.1 Research design ... 31
4.2 Pretest ... 31
4.3 Variable operationalization ... 35
4.3.1 Stimuli ... 35
4.3.2 Measurements ... 35
4.3.3 Control variables ... 36
4.4 Procedure ... 36
4.5 Sample size and the sample ... 37
5. Results ... 38
5.1 Reliability, factor analysis & correlation ... 38
5.2 Main effects ... 38
5.3 Moderating effect communication channel ... 41
5.4 Moderating effect promotion sensitivity ... 42
5.5 Robustness check ... 45
6. Discussion, implications & limitations ... 46
6.1 Discussion ... 48
6.2 Academic implications ... 50
6.3 Managerial implications ... 51
6.4 Limitations and future research ... 52
Reference list ... 54
Appendix A: Pretest questionnaire ... 67
Appendix B: Visualizations outcomes pretest ... 74
Appendix C: Experiment questionnaire ... 75
Appendix D: Factor analysis ... 81
Nowadays, many retailers implement personalization into their loyalty programs with the hope to increase customer loyalty. However, implementations of personalization vary widely, and their relative effectiveness remains unknown. Therefore, the purpose of this research is to examine the effectiveness of personalization – individual, segmented, or none – within a loyalty program. Moreover, we want to know if this impact is contingent on other factors such as communication channels and promotion sensitivity. We use an experimental design with a sample of 317 respondents to test these effects. The findings show some support for a positive effect of personalization on loyalty, although it is not contingent on communication channels or promotion sensitivity. Managers can use these insights to develop an optimal
personalization strategy for their loyalty program.
Keywords: personalization, loyalty programs, individualized personalization, segmented personalization, communication channel, promotion sensitivity.
One of the biggest trends in customer loyalty is personalization (The Wise Marketer, 2021).
As such, firms increasingly try to create customer loyalty by developing a personalized loyalty program. Research shows that personalized loyalty programs have a positive influence on customer behavior (Venkatesan & Farris, 2012; Zhang & Wedel, 2009). Interestingly, retailers’ practices vary widely: the Dutch drugstore Kruidvat, for example, runs a
personalized loyalty program whereby they reward their members based on preferred topics indicated in advance (‘Jouw extra aanbieding’). These segmented rewards are communicated through email (Kruidvat, n.d.). Another example is ‘Mijn Bonus Box’ established by the Dutch supermarket Albert Heijn. This program offers individualized rewards that are based on past purchases which are communicated through a mobile app notification and email (Albert Heijn, n.d.). Both retailers do not take customer characteristics into account. This diversity of executing a personalized loyalty program stands in contrast with the academic literature because we do not know the effectiveness of these different practices. It is therefore not only valuable to know if personalization can add value, but also how retailers should personalize, how retailers should communicate those offers, and whether it differs among customers’ characteristics within a loyalty program.
Research shows that around 60% of all retailers run a loyalty program (Finaccord, 2013). Moreover, almost ninety percent of the population actively participates in some type of loyalty program (Sneed, 2005). Although extant research concludes that members of a loyalty program are more loyal in comparison to non-members (Bijmolt, Dorotic, & Verhoef, 2010), the satisfaction of loyalty programs decreases over years (Bond, 2020). Anticipating this, a firm can boost customers’ satisfaction and loyalty by increasing the attractiveness of the loyalty program and its rewards (Demoulin & Zidda, 2008; Steyn, Pitt, Strasheim, Boshoff &
Abratt, 2010; Wirtz, Mattila & Lwin, 2007). McCall and Voorhees (2010) state that
customers must see and identify the benefits of a program before it can become successful. As 75% of the customers would like to switch to a loyalty program that fits better their personal needs (Cizmeci, 2020), it is worthwhile to study the effectiveness of personalized loyalty programs.
This study relates to two research streams of literature, namely loyalty programs, and personalization. Numerous studies investigate the impact of different loyalty program designs (Bombaij & Dekimpe, 2020; Haisley & Loewenstein, 2011; Dreze & Nunes, 2009), the advantages and disadvantages for firms (McCall & Voorhees, 2010; Sharp & Sharp, 1997), and the benefits for customers for example (Leenheer, Van Heerde, Bijmolt, & Smidts, 2007;
Mercadé-Melé, Molinillo, Fernández-Morales, & Porcu, 2018; Reinartz, 2006). Specifically, Bombaij and Dekimpe (2020) state that the efficiency of a loyalty program is determined by reward type because customers create the highest loyalty for tangible rewards. At the same time, customers are not sensitive to reward timing (Meyer-Waarden, 2015). Moreover, Rust and Verhoef (2005) argue that customers are heterogeneous and therefore respond differently to marketing expressions. Accordingly, since personalization has the potential to have
benefits for both customers and firms, marketers increasingly integrate personalization to make loyalty programs more valuable (Vesanen, 2007). Extant research on personalization distinguishes between three dimensions, which can lead to better matching rewards, products, or services (Kwon & Kim, 2012). In turn, this has a positive effect on the customer
experience (Fiore, Lee & Kunz, 2004; Ansari & Mela, 2003), loyalty, and satisfaction (Srinivasan, Anderson & Ponnavolu, 2002; Halimi, Chayosh & Choshalyc, 2011).
Contrastingly, Evans (2003) study the downsides of personalization which is customers’
privacy risk for example. Nevertheless, due to technological innovations, personalization has made huge improvements in speed and efficiency last decades (Fan & Poole, 2006).
Despite retailers implement personalized loyalty programs in many different ways, no study tests the efficiency of these different personalization types. Therefore, as a first
managerial contribution, we will study the differences between various personalization types in loyalty programs. In addition, not only the personalization type can be different, but also how retailers communicate the rewards. For managers, it is valuable to know what
communication channel is most effective for each personalization level. Accordingly, as a second contribution, this study investigates if the effectiveness of a personalized loyalty program is contingent on which channel is used for communicating the rewards. Finally, not every customer is the same, so it is important to look at customer characteristics that can change the effectiveness of different personalization types. Therefore, the final contribution of this study is to test if companies need to take consumer heterogeneity into account in
developing a personalized loyalty program.
To fill the gaps, this study looks at the effects of two different levels of personalization (individualized versus segmented) compared to a control group, executed in a fictional loyalty program. In addition, this study tests the moderating effect of both offline and online
communication channels. In terms of consumer characteristics, we will focus on customers’
promotion sensitivity. We adopt an experimental design with 317 respondents, where we manipulate personalization level and communication channel and measure promotion sensitivity. Extant research defines contrasting levels of personalization, but there is no empirical research confirming these differences. Therefore, the main academic implication of this study is that it provides insights on the differences between individualized, segmented, and no personalization within a loyalty program.
2. Literature review
In this chapter, we will discuss extant research on loyalty programs, personalization, communication channels, and consumer characteristics. Table 1 gives an overview of the existing literature on personalization, which can occur in a loyalty program. We are going to discuss this in more detail in the following sections.
2.1 Loyalty programs
Kim, Steinhoff, and Palmatier (2020, p.73) define loyalty programs as “any institutionalized incentive system that attempts to enhance consumers’ consumption behavior over time, which captures a broad span of types of programs”. We will discuss the advantages and
disadvantages of loyalty programs, the variety of programs that fall under the broad span, and the consequences of consumer heterogeneity next.
2.1.1 Reasons for adopting a loyalty program
Companies give a lot of attention to relationship management, since building customer loyalty is the only way of creating a sustainable competitive advantage (Parvatiyar & Sheth, 2001).
Companies should no longer just offer a product for a specific price but rather create a real experience for their customers (Pine, Pine & Gilmore, 1999). Therefore, firms increasingly try to create customer loyalty by running a loyalty program.
Customer loyalty is a two-dimension construct consisting of both attitudinal and behavioral loyalty. On the one hand, attitudinal loyalty is about the perception that customers have of a brand, product, or service (Furinto, Pawitra & Balqiah, 2009). This can be measured by commitment, satisfaction, and positive word-of-mouth (Dorotic, Bijmolt & Verhoef, 2012), and have a long-term orientation (Furinto et al., 2009). Behavioral loyalty, on the other hand, is short-term related (Bijmolt et al., 2010) and can be measured by the share of wallet, retention rate, and profit for example (Dorotic et al., 2012).
Firms’ advantages of loyalty programs are a steady customer base, higher profit, (McCall & Voorhees, 2010) a higher retention rate (Sharp & Sharp, 1997; McCall &
Voorhees, 2010), and an increase in the share of wallet (Leenheer et al., 2007). These effects are all related to behavioral loyalty. Brashear-Alejandro, Kang, and Groza (2016) look at attitudinal loyalty and conclude that firms who deliver non-financial rewards to their
customers (i.e., personal recognition, preferential treatment, exploration experience), end up in a deep relationship with the customer. Since behavioral loyalty has a positive outcome in the short-term only, it is important to give at least as much focus to attitudinal loyalty, to benefit in the long run as well (Yi & Jeon, 2003). A firm can boost customers’ satisfaction and attitudinal loyalty by increasing the attractiveness of the loyalty program and its rewards (Demoulin & Zidda, 2008; Steyn et al., 2010; Wirtz et al., 2007).
Customers’ advantages from loyalty programs include receiving price discounts and more efficient services (Leenheer et al., 2007). Moreover, loyal customers have the
opportunity to receive personalized communication (Mercadé-Melé et al., 2018), and timely information about discounts, special offers, and new products (Reinartz, 2006). Finally, participating in a loyalty program can also serve as a status symbol. This relates to the psychological feeling of satisfaction and reputation in an environment created by the membership of a loyalty program (Huberman, Loch, & ÖNçüler, 2004; Dreze & Nunes, 2009).
2.1.2 Loyalty program designs
Distinctive ways of executing a loyalty program exist. First, a company can decide to reward a customer immediately after purchase (e.g., discount or a free sample) or only after obtaining a designated number of points (i.e., point-saving system) (Bombaij & Dekimpe, 2020).
According to Bombaij and Dekimpe (2020), the efficiency of a loyalty program is not contingent on reward timing. Second, a firm choose what reward type to use (Haisley &
Empirical studies on personalization
on individual Personalized
on segment LP Consumer
heterogeneity OV Methodology
Ansari & Mela (2003) X Website clicks Meta-analysis
Arora et al., (2008) X N/A Literature review
Bijmolt, Dorotic & Verhoef (2010) X Attitudinal loyalty and behavioral loyalty Literature review
Chellappa & Sin (2005) Privacy concerns N/A Survey
Evans (2003) X N/A Literature review
Fan & Poole (2006) X X N/A Literature review
Fiore, Lee & Kunz (2004) X Customers’ willingness to use co-design Lab experiment
Furinto, Pawitra & Balgiah (2009) X X Attitudinal loyalty Survey
Halimi, Chayosh & Choshalyc (2011) X Attitudinal loyalty and behavioral loyalty Survey
Han & Shavitt, (1994) X Culture orientation Individualism and collectivism Lab experiment
Kaneko, Kishita & Umeda (2018) X X N A Literature review
Kwon & Kim (2012) X X Customer retention Lab experiment
Murthi & Sarkar (2003) X N/A Meta-analysis
Rust & Verhoef (2005) X Behavioral loyalty Field experiment
Shen & Ball (2011) Customers’ preferences Preference stability belief Lab experiment
Simonson (2005) X Customers’ preferences N/A Literature review
Srinivasan, Anderson & Ponnavolu
(2002) Attitudinal loyalty and behavioral loyalty Survey
Venkatesan & Farris (2012) X Trip incidence and trip revenue Field experiment
Zhang & Wedel (2009) X X X Behavioral loyalty Field experiment
This study X X X Attitudinal loyalty Lab experiment
Note: LP = loyalty program, OV = outcome variable
Loewenstein, 2011), whereby, for example, the distinction can be made between tangible- and intangible rewards. Tangible rewards can be both cash and non-cash incentives but always have a monetary value. Examples are debit cards, merchandise, and gift cards (Presslee, Vance & Webb, 2013). Intangible rewards are less observable and measurable like social approval, verbal praise, and acknowledgment (Yoon, Sung, Choi, Lee & Kim, 2015). Meyer- Waarden (2015) states that customers’ preference for a loyalty program and store loyalty intentions are higher for tangible than intangible rewards. Third, firms decide if there needs to be a tiered structure in the loyalty program. Subsequently, within a tiered structure, firms choose the number of labels. Some will distinguish between two tiers (e.g., silver and gold), while others decide to distinguish between more status labels, such as bronze, silver, gold, and platinum (Dreze & Nunes, 2009). According to Dreze and Nunes (2009), a three-tier program results in higher customer satisfaction in comparison to a two-tiered program. Additionally, they conclude that the larger the second tier, the less special the top tier feels (Dreze & Nunes, 2009). Finally, companies can determine to set up a coalition with other retailers for their loyalty program. Bombaij and Dekimpe (2020) state that a multivendor program has a negative influence on the efficiency of a loyalty program in comparison to a sole-proprietary program.
2.1.3 Consumers heterogeneity
McCall and Voorhees (2010) argue that customers must see and identify the benefits of the program to become successful. Nevertheless, responses to marketing interventions are
heterogeneous (Rust & Verhoef, 2005). This makes it ambitious for firms to create a valuable program for the whole target audience. A loyalty program design that is based on raw
classifications such as loyal versus non-loyal is not appropriate enough (Krishnamurthi &
Papatla, 2003). Consequently, marketers try to integrate personalization into their loyalty
program designs. Extant research shows that personalized loyalty programs have a positive influence on customer behavior (Venkatesan & Farris, 2012; Zhang & Wedel, 2009).
Lopes, Cabral, and Bernardino (2016, p. 131) define personalization as “the customization of the outputs of a system based on the collected personal information of its users”. It is about selecting the right marketing effort for the right customer (Arora et al., 2008) and relies on user information like interest and preferences (Lopes et al., 2016). Moreover, personalization can be executed in many areas of marketing (e.g., promotion, products, service, price, and delivery) (Vesanen, 2007).
2.2.1 Advantages and disadvantages of personalization
Firms’ advantages of personalization can include an increase in customers’ satisfaction and customer loyalty (Srinivasan et al., 2002; Halimi et al., 2011). Moreover, companies might achieve a competitive advantage (Murthi & Sarkar, 2003), amount higher prices (Vesanen, 2007), and improve their profitability (Zhang & Wedel, 2009). Firms’ disadvantage of personalization is the costs since it is dependent on data collection, analyses, and expensive software (Arora et al., 2008).
Customers’ advantages of personalization include better matching products/services, reduced cognitive overload, convenience, and a better customer experience (Fiore et al., 2004;
Ansari & Mela, 2003). Nevertheless, there are a couple of disadvantages as well. Customers can experience privacy- and spam risks for example (Evans, 2003; Yu & Cude, 2009). As a consequence, it is the task of the firm to secure that customers’ advantages exceed the disadvantages of personalization (Simonson, 2005).
2.2.2 Dimensions of personalization
In-depth research on personalization investigates different dimensions of personalization. By integrating personalization in a marketing strategy, a firm should decide how to approach each dimension separately. The three personalization dimensions that receive the most attention in existing literature are differentiation, the object, and the subject.
First, differentiation is the personalization dimension by which the target of
personalization is meant (“to whom to personalize to”). The distinction can be made between the individual or a segment, which corresponds to one-to-one personalization and one-to-N personalization (Fan & Poole, 2006; Kaneko et al., 2018; Arora et al., 2008, Kwon & Kim, 2012). For personalization on a segment-level (one-to-N) the rewards are personalized to a group of customers (e.g., customers with the same preferences or past purchases) but are not tailored on an individual level. The most granular level, individualized personalization, refers to rewards that are personalized for each customer separately (Zhang & Wedel, 2009).
According to Kwon and Kim (2012), individualized interface personalization is more effective than segmented and one-to-all interface personalization. Regarding segmented personalization (one-to-N) both content and interface personalization is more efficient than no personalization (one-to-all).
Second, the ‘object’ has been studied as a personalization dimension whereby they refer to “what is personalized” (Kwon & Kim, 2012; Fan & Poole, 2006). According to Fan and Poole (2006), there are four aspects of information systems that can be personalized: the content, the user interface, the media channel, and the functionality. With this latter, they mean what users can do with the system (Fan & Poole, 2006). Kwon and Kim (2012) only test the effects of personalized content and personalized user interface. They conclude that both have a positive effect on customer satisfaction and loyalty.
Third, studies distinguish between different subject levels of personalization. With this dimension, they refer to “who does the personalization”. In other words, the subject level involves the degree to which personalization is automated. The distinction is made between user-initiated versus system-initiated personalization (Kwon & Kim, 2012) or implicit versus explicit personalization (Fan & Poole, 2006). On the one hand, a situation whereby customers actively provide information to the company about themselves is called explicit or user- initiated personalization. On the other hand, implicit or system-initiated personalization applies to situations whereby the personalization happens automatically by the system (Fan &
Poole, 2006). An example of system-based (or implicit) personalization would be personalized loyalty programs, where customers' information typically consists of past purchases. Based on past purchases a company can form segments consisting of people with similar preferences or determine one’s individual needs. Kwon and Kim (2012) state that explicit interface personalization is more effective than implicit interface personalization whereas explicit content personalization is not more effective than implicit content personalization (Kwon & Kim, 2012).
2.2.3 Consumers heterogeneity
There are a couple of customer characteristics that could influence the effectiveness of personalization. First, customers’ privacy concerns determine the effectiveness of
personalization. Customers need to provide personal information to tailor products, services, or promotions to their tastes and preferences. Therefore, personalization is impossible without some loss of privacy (Chellappa & Sin, 2005). Before the disclosure of information, the decision processes involve an exhaustive assessment of the cost and benefits related to
information disclosure (Dinev & Hart, 2006). This means that firms can improve performance by addressing and reducing customers’ privacy concerns (Lee, Ahn & Bang, 2011).
Second, customers’ cultural orientation influences the response to personalization.
Regarding cultures, prior research distinguishes between individualism and collectivism.
Individualistic cultures consist of people who prioritize their individual preferences (Kramer, Spolter-Weisfeld & Thakkar, 2007), whereas collectivistic cultures consist of people who put their collective preferences in the first place (Kramer et al., 2007). For individualistic cultures, individualized rewards are more effective, and rewards that focus on groups are more
effective in collectivistic cultures (Han & Shavitt, 1994).
Finally, the stability and clarity of customers’ preferences determine the effectiveness of personalization (Simonson, 2005; Shen & Ball, 2011). Personalization assumes that customers have stable and well-defined preferences which ensures that marketers can learn about and build a relationship with the customer (Shenn & Ball, 2011), while in reality, this is not the case. As a consequence, personalized appeals do not fit customers' actual preferences well and customers fail to recognize them (Simonson, 2005).
2.3 Marketing communication channels
According to Danaher and Rossiter (2011), with marketing communication channels we refer to all media channels through which firms can send marketing communication to customers.
Table 2 gives an overview of existing literature on online marketing communication channels.
Interestingly, the methodology literature review is quite dominant. For each study, we specify if it contains offline communication channels as well. We are going to discuss this in more detail in the following sections.
2.3.1 Advantages and disadvantages of online versus offline communication channels Offline communication channels can benefit from customers’ trust since it has a solid
historical foundation and positive image (Tong, 2018). Moreover, the credibility and authority of traditional (offline) media are higher compared to new (online) media (Tong, 2018; Hao &
Prior empirical research on online marketing communication channels
Includes offline channels
heterogeneity OV Methodology
Ailawadi, Beauchamp, Donthu,
Gauri & Shankar (2009) X Literature review
Bezjian-Avery, Calder & Iacobucci
(1998) X Attitude toward the
brand, attitude toward the ad, purchase intent
Danaher & Rossiter (2011) X X Customer
engagement and persuasion
Gopal, Pathak, Tripathi & Yin
(2006) Abnormal stock
Hao & Jin (2013) X Literature review
Hearn, Foth & Gray (2009) Literature review
Hun & Yazdanifard (2014) X X Literature review
Lee, Lee, Kim, Kim, Cho, Jang,
Jang (2018) Literature review
Lynn, Maltz, Jurkat & Hammer
(1999) X Use and
effectiveness Survey Peltier, Schibrowsky, Schultz
(2003) Literature review
Sahni, Wheeler & Chintagunta
(2018) Number of leads,
link clicks, number of emails opened
Souiden, Chtourou & Korai (2017) X Attitude toward the ad, attitude toward online advertising
Stewart & Pavlou (2002) X Literature review
Todor (2016) X Literature review
Tong (2018) X Literature review
Vesanen (2007) Literature review
Yoon & Steege (2013) X Internet banking use Survey
This study X X Attitudinal loyalty Lab experiment
Note: OV = outcome variable
Jin, 2013; Todor, 2016). The disadvantages of traditional media are high prices (Lynn, Maltz, Jurkat & Hammer, 1999), customers’ lack of control, and the fact that it is passively mass communication which makes it less attractive (Bezjian-Avery, Calder & Iacobucci, 1998).
Finally, with communicating through offline media, a firm is unable to measure the direct effect of their communication efforts (Stewart & Pavlou, 2002; Todor, 2016).
Online channels can benefit from lower costs. In comparison to traditional media, the costs of new media technologies are decreased enormously. While offline channels belong mostly to large firms because of the expensiveness, small firms are now benefiting from the
online channels in marketing (Lynn et al., 1999). Moreover, online communication channels can create interactivity with the customer (Peltier, Schibrowsky, Schultz, 2003; Todor, 2016).
Furthermore, because online information is permanently available and offered on an ongoing basis, the duration and active users’ approach are very rich benefits as well. Finally, for online channels, it is easy to measure the effects (Todor, 2016). Disadvantages of online channels include the required disciplinary skill sets marketers must contain as well as the rapid changes in terms of content, message, and technology (Hearn, Foth & Gray, 2009). Moreover, as opposed to offline media, online channels are not embraced by all people (Todor, 2016). To summarize, Table 3 gives an overview of all the advantages and disadvantages of offline versus online communication channels.
Advantages and disadvantages of communication channels
Offline channels Online channels
Advantages Trustworthiness Relatively cheap
Credibility Offered on an ongoing basis
Easiness to measure the effects
Disadvantages Expensive Required skills set
Customers’ lack of control Rapid changes
No interactivity possible Not fully accepted by everyone No possibility to measure direct effects
2.3.2 Different types of communication channels
Communication channels are subdivided into offline and online channels (Danaher &
Rossiter, 2011). Traditional communication channels, like direct paper mail, radio, TV advertisements, sales flyers, telemarketing, billboards, and brochures, are always offline (Todor, 2016). Although, due to new technologies, a new online media landscape (e.g., email, blog, video, social media, and apps) has emerged and firms do not focus on only offline channels anymore. Customers can access online channels with any internet-enabled device (e.g., smartphone or tablet) by browsing the web (Maghsoudi, Shapka & Wisniewski, 2020).
This new media landscape consequently opens up new promotional communication channels
for retailers (Ailawadi, Beauchamp, Donthu, Gauri & Shankar, 2009). For example, customized electronic coupons, offering price-promotions on a deal-forum (Gopal, Pathak, Tripathi & Yin, 2006), targeted email promotions (Sahni, Wheeler & Chintagunta, 2018;
Singh, Singh & Shriwastav, 2019), and mobile-app promotions (Lee et al., 2018).
2.3.3 Consumer heterogeneity
The way customers react to a specific communication channel depends on various aspects, such as generation, customers’ needs, and personality characteristics. First, generations distinguish from one another because of their different experiences, values, beliefs, attitudes, preferences, feelings, and ideas (Hun & Yazdanifard, 2014). The use of the internet is highest among young people. That is why Danaher and Rossiter (2011) expect that younger people are more likely to act on an offer sent through an online communication channel compared to a traditional offline media channel.
Second, customers’ needs can have an influence on the way a customer reacts to a communication channel. When a customer wants to gather information, for example, he or she prefers print since this is an information-rich medium. On the other hand, television is a more entertaining communication channel (Danaher & Rossiter, 2011).
Third, customers’ persistent personality characteristics influence one’s attitude towards communication channels. The idea is that customers’ personality characteristics influence attitudes and behavior (Souiden, Chtourou & Korai, 2017). According to the Five- Factor Model, everyone’s personality is characterized by five traits, namely extraversion, openness, conscientiousness, agreeableness, and neuroticism (Digman, 1990). For example, when a customer has an open character, online communication channels are more preferable compared to when someone has a closer character (Yoon & Steege, 2013).
2.4 Sales promotions and promotion sensitivity
Sales promotions are special offers to the customer, with the potential to attract new
customers or induce existing customers to buy more (Yang, Cheung, Henry, Guthrie & Fam, 2010). Some customers are more sensitive to sales promotions, who we refer to as high promotion sensitive customers. Oh and Kwon (2009, p. 869) define promotion sensitivity as
“an inherent consumer characteristic, that is, self-perception of how purchases are actually influenced by price promotions”. Table 4 gives an overview of the existing literature on sales promotions, which we discuss in more detail next.
Prior empirical research on sales promotions
heterogeneity OV Methodology
Buil, De Chernatony and Martínez (2013)
Brand loyalty Survey Chandon, Wansink &
Laurent (2000) Brand sales Lab experiment
DelVecchio, Henard &
Freling (2006) X Promotion sensitivity Brand preference Meta-analysis
Gupta (1988) Brand sales Field
experiment Ndubisi & Moi (2005) Fear of losing face Brands repurchase Survey Oh & Kwon (2009)
sensitivity, promotion proneness, promotion knowledge, shopping enjoyment
Brand sales Survey
Santini, Vieira, Sampaio &
Short-term results, Long-term results
Meta-analysis Yang, Cheung, Henry,
Guthrie & Fam (2010) Customer’ attitude Survey
Yi & Yoo (2011) X Promotion sensitivity Purchase behavior Lab experiment Zhang & Wedel (2009)
X Promotion sensitivity Purchase incidence, Purchase choice, Purchase quantity
Field experiment This study X Promotion sensitivity Attitudinal loyalty Lab experiment Note: OV = outcome variable
2.4.1 Reasons for adopting sales promotions
Firms’ advantages of sales promotions diverge among extant research. Except for the agreement on the short-term effects, like increase of sales volume and purchase intentions (Santini, Vieira, Sampaio & Perin, 2016; Gupta, 1988; Ndubisi & Moi, 2005), for the long-
term effects, most research is not aligned. Some studies conclude that sales promotions do not affect post-promotion brand preference (DelVecchio, Henard & Freling, 2006), while others say it strengthens brand loyalty and attitudes towards the brand and product (Santini et al., 2016). Yet another study concludes that sales promotions decrease brand loyalty and quality perceptions, whereas it increases brand switching and price promotion sensitivity (Delvecchio et al., 2006). Finally, Buil et al. (2013) argue that the effectiveness depends on the promotion type whereby monetary promotions harm perceived quality and nonmonetary promotions have a positive effect on perceived quality and brand associations.
Customers’ benefits of sales promotions are classified by utilitarian and hedonic benefits. On the one hand, utilitarian benefits are functional, cognitive, and instrumental.
Examples of these benefits are product quality and shopping convenience. On the other hand, hedonic benefits are experiential, affective, and non-instrumental, such as value expression, entertainment, and exploration (Chandon, Wansink & Laurent, 2000). According to Chandon et al. (2000), non-monetary promotions mostly deliver hedonic benefits for the customer whereas monetary promotions provide utilitarian benefits.
2.4.2 Different types of sales promotions
In general, there are two types of sales promotions: monetary promotions and nonmonetary promotions (Yi & Yoo, 2011; Chandon et al., 2000; Buil, De Chernatony & Martínez, 2013).
Monetary promotions are price-oriented promotions, like price reductions, coupons, and rebates, whereas nonmonetary promotions are non-price-oriented and include gifts, premiums, contests, bonus packs, and BOGOF (buy one get one free) (Chandon et al., 2000). Yi and Yoo (2011) argue that nonmonetary promotions have a more positive effect on brand attitude in the long run as compared to monetary promotions.
In comparison to loyalty programs, sales promotions are mostly short-term oriented whereas loyalty programs are running for a longer period (Sharp & Sharp, 1997).
Nevertheless, the design of a loyalty program is partly comparable to sales promotions. For example, companies decide what reward type to use for their loyalty program, whereby a firm distinguishes between monetary and nonmonetary promotions as well (Haisley &
2.4.3 Promotion sensitiveness
According to Delveccio et al. (2006), one of the consequences of sales promotions is customers’ promotion sensitivity. This customer characteristic refers to the probability of committing to price discounts (Zhang & Wedel, 2009). High promotion-sensitive customers behave differently compared to low promotion-sensitive customers because they are more eager to buy price discounts than others. Customers’ promotion sensitiveness can influence loyal customers as well (Dominique-Ferreira, Vasconcelos & Proença, 2016) because these customers buy promotions on brands they are used to buy (Gazquez-Abad & Sanchez-Perez, 2009). Customers who believe to receive a good deal will have a better customer experience (Oh, 2003). Customer experience, in turn, enhances attitudinal loyalty (Srivastava & Kaul, 2016), which makes the investigation on the influence of promotion sensitivity very essential.
Not only sales promotions enhance customers’ promotion sensitivity. According to Oh and Kwon (2009), three different customer characteristics can influence promotion sensitivity.
First, they argue that customers' promotion sensitivity might differ based on proneness. With shopper promotion proneness we mean the degree that customers use price promotion
information as a basis for making a purchase. Second, the characteristic promotion knowledge influences promotion sensitiveness. This refers to customers’ degree of deal knowledge.
Third, shopping enjoyment influences promotion sensitivity. With shopping enjoyment, we mean the emotional value and pleasure realized from a purchase (Oh & Kwon, 2009). Oh and Kwon (2009) conclude that promotion proneness and shopping enjoyment have a positive effect on promotion sensitivity for both online and offline shopping, whereas promotion
knowledge mostly enhances promotion sensitivity in an online shopping environment (Oh &
3. Conceptual framework
In this chapter, we visualize the hypotheses and their relationships in a conceptual framework.
This study aims to examine the effects of the contrasting personalization types, namely individualized (one-to-one), and segmented (one-to-N) personalization, in contrast to no personalization in a loyalty program. The personalization dimensions object and subject will not be taken into account because within a personalized loyalty program the object is always the reward that the customer receives. Moreover, the assumption is that for personalized loyalty programs firms conduct the most valuable knowledge of a customer through past purchases, causing that in personalized loyalty programs the subject is mostly system-initiated (implicit personalization).
Regarding the dependent variable, it is important to give at least as much focus to attitudinal loyalty as to behavioral loyalty since the latter has a positive outcome in the short term only (Yi & Jeon, 2003). Extant research on loyalty programs paid little attention to this first type of loyalty (Yi & Jeon, 2003; Venkatesan & Farris, 2012; Zhang & Wedel, 2009;
Leenheer et al., 2007; Sharp & Sharp, 1997). Therefore, in this study, we measure the effects on attitudinal loyalty.
Moreover, as a first moderator, we focus on communication channels because according to Breugelmans et al. (2015) the effectiveness of a loyalty program is potentially contingent on the communication channel which is used. In this study, we will distinguish between two different types namely offline and online. Finally, multiple customer
characteristics can influence the efficiency of personalized loyalty programs. As a second moderator, we study the effect of customers’ promotion sensitivity. According to Delveccio et
al. (2006) sales promotions enhance promotion sensitiveness. Despite that loyalty programs and sales promotion are not completely comparable, there are some similarities, which makes it interesting to study if this customer characteristic influences the effectiveness of a loyalty program setting as well. We summarize all proposed variables and relationships in a
conceptual framework (Figure 1).
3.1 Personalization versus no personalization
We first want to know whether personalization improves a customer’s loyal attitude towards a firm. According to Petty and Cacioppo (1986), the likelihood that an attitude change will occur partly depends on whether a message is processed cognitively. The model underlying this is called The Elaboration Likelihood Model (ELM) which describes two routes to attitude formation as a response to persuasive messages. First, the central route occurs when
customers are motivated and able to process the message. Second, peripheral processing takes place when motivation and/or ability to process the message are low (Petty and Cacioppo, 1986; De Keyzer, Dens & De Pelsmacker, 2015). According to ELM, for both high and low
Attitudinal Loyalty Communication channel
elaboration, personalization should benefit attitudes and behavior. For high elaboration, personalization messages could be perceived as stronger compared to messages which are not personalized. Therefore, perceived personalization can lead to biased message processing. For low elaboration, perceived personalization can serve as a heuristic cue which results in a positive attitude change (De Keyzer et al., 2015).
Another theory, called the need for uniqueness construct, describes that individuals are seeking for material goods to differentiate themselves from others and reinforce one’s self- image (Knight & Kim, 2007). According to this theory, individuals pursue brands and products to express uniqueness to enhance social identity (Chan, To & Chu, 2015). In a loyalty program, one can distinguish between personalized and non-personalized rewards.
Customers who receive a personalized reward feel potentially more unique than customers who receive a reward that is offered to everyone. Personalized rewards are based on their past purchase behavior, which enhances their personal image and social identity. Therefore, according to the need for uniqueness construct, customers will infer that personalized rewards have a higher value as compared to non-personalized rewards.
Moreover, the idiosyncratic fit theory explains that customers evaluate marketing promotions based on their fit with the offer. If customers have the feeling that there is a relatively good fit with the offer, customers may conclude that this alternative is particularly attractive for them. Customers’ idiosyncratic fit indicates that customers have a relative advantage for that alternative (Kivetz & Simonson, 2003). Since personalized rewards are based on customers’ preferences and past purchases, the fit with the customer is better compared to rewards that are not personalized. Therefore, according to the idiosyncratic fit theory, customers will have an advantage for personalized rewards.
To summarize, three theories are substantiating that personalized rewards have more value than rewards that are not personalized. This is in line with extant research on
personalization. Particularly, Noar, Benac and Harris (2007) state that personalized messages are more effective compared to non-personalized messages since they are more memorable, more likable, and sparking behavioral change compared to mass communication.
Because personalized rewards are more unique and have a better fit with the customer, we expect higher reciprocity for personalized rewards. Reciprocity is an example of a social norm, which entails the ongoing process of exchange between parties, to establish and maintain equality (Maiter, Simich, Jacobson & Wise, 2008). According to the concept of reciprocity, the better the reward, the better the response (Haisley & Loewenstein, 2011). We assume that personalized rewards will lead to higher reciprocity, which thus, in turn, will lead to higher attitudinal loyalty. Therefore, based on the theories we explain above, we formulate the following hypothesis:
H1: Any type of personalized loyalty program (both individualized and segmented) leads to higher attitudinal loyalty than a loyalty program without personalization (control group).
3.2 Individualized versus segmented personalization
Second, we want to know what personalization type has the strongest effect on customer’s loyal attitude towards a firm. Based on the four theories we explain in favor of personalization in general (both individualized and segmented), we assume that these positive effects will be stronger for individualized than for segmented personalization. According to the ELM
construct, the need for uniqueness construct, and the idiosyncratic fit theory the expectation is that customers will value the individualized rewards higher as compared to segmented
rewards. Subsequently, we expect that members view individualized rewards as ‘better’
rewards than rewards that are based on segments. When people receive an individualized reward that is based on past purchases, customers know it represents their social identity (Chan et al., 2015). This, in turn, according to the social norm reciprocity, will lead to better
results (e.g., higher attitudinal loyalty) (Haisley & Loewenstein, 2011). This is different for segmented rewards since the whole segment receives this reward, which makes it less unique.
One construct that stands in sharp contrast to the arguments above is the privacy concerns customers might have. Lee et al. (2011) argue that privacy concerns have a negative influence on customers’ willingness to uncover personal information in transactions.
Individualized personalization goes hand in hand with losing privacy (Chellappa & Sin, 2005) since firms offer rewards based on personal information. Firms offering personalized rewards on a segment level, base them on personal information to a lesser extent. Nevertheless, these privacy concerns do not outweigh all four arguments in favor of individualized
personalization. Therefore, we formulate the following hypothesis:
H2: Individualized (one-to-one) personalization has the strongest effect on attitudinal loyalty compared to segmented (one-to-N) personalization.
3.3 Communication channels
Third, we want to know whether the communication channel moderates the relationship between personalization and attitudinal loyalty. Aguirre, Roggeveen, Grewal, and Wetzels (2016) state that the benefits of personalization depending on the communication channel.
According to the cue consistency theory, multiple extrinsic cues are more useful when they provide corroborating information than when they provide contradictory information because customers evaluate cues in an integrated and cooperative manner (Miyazaki, Grewal &
Goodstein, 2005). In this case, with the conjunction of communication channel and
personalization type, it is not only about the information within the message itself, but also about the way how this message is conveyed. We assume that according to the cue
consistency theory, the better the ‘fit’ between the communication channel and the message
itself, the higher the effectiveness will be. More specifically, Šerić, Ozretić-Došen, and Škare (2020) state that consistency enhances trust, customer loyalty, and commitment.
One characteristic of offline communication channels is that there is no possibility to interact, as opposed to online communication channels where this is possible (Peltier et al., 2003; Todor, 2016). To be able to offer individualized personalized rewards, firms must have the availability of much more individual customers’ data (Kwon & Kim, 2012) as compared to segmented personalized rewards, in which the necessary customer data is a bit more general and based on group preferences (Zhang & Wedel, 2009). Therefore, according to the consistency theory, there seems to be a lesser ‘fit’ between offline communication channels and individualized personalization, as compared to offline channels and segmented
personalization. Particularly because offline communication channels have no possibility to gain interactivity with the customer which results in limited data, whereas individualized personalization requires a lot of customer data to be as accurate as possible.
One characteristic of online communication channels is that the technology, content, and messages rapidly change (Hearn et al., 2009). Shen and Ball (2011) state that individuals do not have stable preferences, which makes constantly anticipating to changes in customers’
preferences necessary for firms, to be effective. As such, online communication channels seem to have a good ‘fit’ with individualized personalization because they offer the
possibility to anticipate to changes. Therefore, for individualized personalized rewards, the expectation is that there will be a synergetic interaction when the rewards are communicated through an online media channel.
Moreover, the privacy concern statement of Lee et al. (2011) which says that privacy concerns have a negative influence on customers’ willingness to share information, is also applicable here. Individualized personalization requires more personal information than segmented personalization (Zhang & Wedel, 2009). When a customer receives a reward by
email or via an app notification, it might feel more privacy-invasive than when he receives it via an offline channel. Therefore, we assume that people who have many privacy concerns would prefer receiving segmented rewards which are communicated through an offline channel because both the personalization type and communication channel are the least privacy invasive. Customers who prefer individualized personalization, are probably less privacy concerned and might want fast (online) access to their rewards. Based on all arguments, we formulate the following two hypotheses:
H3a: The relationship between individualized personalization and attitudinal loyalty is stronger when online communication channels are used as compared to offline
H3b: The relationship between segmented personalization and attitudinal loyalty is stronger when offline communication channels are used as compared to online communication channels.
3.4 Promotion sensitiveness
Finally, we want to know whether promotion sensitivity moderates the relationship between personalization and attitudinal loyalty. Customers who have a high promotion sensitivity, are more eager to buy price discounts than others (Zhang & Wedel, 2009). In other words, the motivation of people with a high sensitivity to promotions is to score the best deal. Oh (2003) argue that customers who believe to receive a good deal, in general, will have a better
customer experience, which in turn enhances attitudinal loyalty (Srivastava & Kaul, 2016).
According to the functional theory of attitudes, to be a good deal, you need to identify the underlying customers’ attitudes (Schlosser, 1998). The identification of underlying attitudes is better for individualized personalization than segmented personalization.
Individualized rewards are customized based on individual customers' information and preferences (Vesanen & Raulas, 2006). As such, according to the idiosyncratic fit theory, it is reasonable to expect that individualized rewards are seen as better deals since they have a better fit with customers' personal preferences (Kivetz & Simonson, 2003). Accordingly, individualized personalization is most influential for high-promotion-sensitive customers since it matches with their motives. This will, in turn, enhance customer loyalty. For
segmented personalization, the opposite is true because this type of personalized reward has a lesser fit with the customer. High-sensitive customers will estimate segmented rewards as a less good deal, which results in lower attitudinal loyalty. For low-sensitive customers, the expectation is that there is no difference in effect between individualized and segmented personalization since they do not care so much about having the ‘best’ deal.
Moreover, the privacy concern statement is also applicable for promotion sensitivity.
According to Lee et al. (2011), customers with high privacy concerns are less willing to share personal data. Because high promotion sensitive people give priority to having the best deal, we assume that these customers have lesser privacy concerns compared to low promotion sensitive people. Highly sensitive people want to share personal information because they know that when they provide this data, they will receive individualized personalized rewards which have a better ‘fit’. As such, based above theories we formulate the following two hypotheses:
H4a: Individualized rewards (one-to-one) offered to high-promotion sensitive customers, will result in higher attitudinal loyalty as compared to segmented rewards (one-to-N).
H4b: There is no difference in the effect on attitudinal loyalty for low-promotion sensitive customers between individualized (one-to-one) and segmented (one-to-N) rewards.
In this chapter, we describe the methodology, whereby we outline the research design, the pretest, the variables, the experiment, and the sample.
4.1 Research design
Managing an explanatory, deductive research approach for this study, we choose an online laboratory experiment to answer the research questions. To test the hypotheses, we conduct an experiment with a 3 (personalization type: individualized/segmented/no personalization) x 2 (communication channel: online/offline) factorial between-subjects manipulation, with a measurement of promotion sensitivity. Thus, in total there are six different treatment groups.
The dependent variable is attitudinal loyalty. The independent variable, personalization type, will be manipulated in a loyalty program design. At last, the two moderators are promotion sensitivity and communication channel, where we measure the former and manipulate the latter. We use Qualtrics to create the experiment, MTurk to recruit the respondents, and SPSS to analyze the data.
To determine whether participants perceive the manipulations of personalization and
communication channels correctly, we conduct a pretest. First, to check the manipulation of personalization type, we create two scenarios of all three types (individualized/segmented/no personalization): a short and an extended description. We expose participants randomly to one of these personalization descriptions. Consequently, the pretest contains a question to measure respondents’ perceived level of personalization, whereby participants have the option to
choose for personalization on an individual level, on a segment level, no personalization, or they could choose for any type of personalization. Second, we also test the perceived communication channel. For both offline and online communication channels, we visualize two expressions. As offline communication channel, we create a flyer and a direct mail (by post) whereas for online we create an email and a mobile app notification. We expose
participants randomly to one of the expressions. Consequently, the pretest contains a question to measure respondents’ perceived communication channel type, whereby participants could choose for an offline communication channel, an online communication channel, or they could choose for any type of communication channel. We visualize the manipulations of the pretest in Figure 2, and we include the questionnaire in Appendix A.
In total, we randomly allocate 81 participants to one of the personalization scenarios (short individualized, n = 15; extensive individualized, n = 13; short segmented, n = 15;
extensive segmented, n = 18; short no personalization, n = 10; extensive no personalization, n
= 10), and one of the communication channels (flyer, n = 22, direct mail, n = 21, e-mail, n = 22, app notification, n = 16) in the pretest. Regarding personalization, we conduct a chi- square test which looks at the proportion of the correct short descriptions versus the correct extensive descriptions. Results show that there is no significant difference between the two proportions, X2(1, 81) = 2.988, p = 0.084. Nonetheless, since the shorter description of each personalization type scores higher percentages as compared to the extensive description (Table 5), we choose to select this one for the experiment. We visualize the results in a graph which we add in Appendix B.
Regarding the communication channels, the results of the chi-square test show that there is a significant difference between the two offline channels, X2(1, 43) = 6.981, p = 0.008 as well as between the two online channels, X2(1, 38) = 5.182, p = 0.023. Moreover, the
Direct mail Mobile app notification
Note: see Appendix A for full images
Valid: 03/03/2021 until 03/10/2021
Online code: HIVU882
Valid: 03/03/2021 until 03/10/2021
Online code: HIVU881
An offer from The Food Market!
Results pretest personalization type
Perceived personalization type Personalization type Individualized
Could be any type of personalization 1. Short individualized
description 40% 20% 20% 20%
2. Extensive individualized description
31% 31% 7% 31%
3. Short segmented description
27% 47% 13% 13%
4. Extensive segmented
description 22% 22% 28% 28%
5. Short no personalization description
0% 30% 40% 30%
6. Extensive no personalization description
60% 0% 20% 20%
Note: descriptions in bold are correct categorizations
results in Table 6 indicate that for the offline communication channel the direct mail (by post) scores the highest, and for the online communication channel, the mobile app notification scores the best. Therefore, regarding the communication channel, we will use direct mail (by post) as an offline communication channel and mobile app notification as an online channel in the experiment. We visualize the outcomes in a graph which we add in Appendix B.
Results pretest communication channel
Perceived communication channel
Communication channel Offline Online Could be any type of
1. Flyer 18% 32% 50%
2. Direct mail 57% 29% 14%
3. E-mail 4% 73% 23%
4. Mobile app notification 0% 100% 0%
Note: descriptions in bold are correct categorizations
4.3 Variable operationalization
In this chapter, we explain how the stimuli are designed, what measurements we use, and we explain the control variables.
Personalization type. Inspired by Kwon and Kim (2012), we manipulate personalization type
by using the two differentiation levels of personalization and a control group: individualized personalization, segmented personalization, and no personalization. We use vignettes to set up the manipulation, which are three different scenarios of a loyal customer of a fictional
American supermarket concern. We choose to use the grocery sector in this experiment because food & beverage is one of the industries where loyalty programs are extremely effective (Davis, 2019). Moreover, we decide to use a fictional brand as a supermarket, since supermarket brands are typically highly dispersed. However, for the rewards, we use real brands that are well-known everywhere (Oreo cookies and Kellogg’s Froot Loops). Based on the outcomes of the pretest, we use the shorter version for all three personalization types in the experiment.
Communication channel. We manipulate communication channels by distinguishing two
types, namely: offline and online communication channels. Because both communication channel types need to have the ability to personalize on the most granular level in this experiment, many communication channels are eliminated like magazines, billboards,
television commercials, social media, and blogs. Based on the outcomes of the pretest, we use direct mail (by post) as offline communication channel and mobile app notification as online communication channel in the experiment.
Promotion sensitivity. We measure customers’ promotion sensitivity by using an existing scale. Wakefield, Kirk, and Inman (2003) set up a three-item scale based on Lichtenstein,
Burton, and Netemeyer (1997), to assess price sensitivity. We adapt this scale to the current context of promotion sensitivity. For each statement participants rate how much they agree or disagree. A seven-point Likert scale is set up anchored by “strongly disagree” and “strongly agree”. In this study, we compare the three statements using factor analysis to form an overall promotion sensitivity score for each participant.
Attitudinal loyalty. We measure attitudinal loyalty using an existing scale from Yi and Jeon
(2003), who measure brand loyalty by four items that ask the relative attitude toward a focal shop. For this study, we adapt the statements to the current context. Participants rate how much they agree or disagree with each of the four statements on a seven-point Likert scale which we anchor by “strongly disagree” and “strongly agree”. To form an overall attitudinal loyalty score for each participant, we compare the four statements using factor analysis.
4.3.3 Control variables
We integrate the variables age, experience with loyalty programs, and gender into the experiment as control variables. As the minimum age for most loyalty programs is eighteen, we do not consider participants below that age. Moreover, we can test if there is a difference in loyalty between different age groups. Furthermore, we ask participants about their
experience with loyalty programs, to test if this affects their loyalty. Finally, we ask
participants to identify their gender because past research shows that women respond more positively to personalized loyalty programs as compared to men (Melnyk & van Osselaer, 2012).
For the main experiment, which is presented in English, we randomly assign the participants to one of the six conditions. First, we expose participants randomly to the loyalty program reward containing either individualized personalization, segmented personalization, or no
personalization at all. Second, for all of these conditions, we display participants either to a reward that is offered through an online channel (i.e., mobile app notification) or through an offline channel (i.e., direct mail by post). Then, we measure customers' promotion
sensitiveness and attitudinal loyalty. Moreover, we add a random question for verification.
Subsequently, participants rate their perceived personalization and perceived communication channel in the same way as in the pretest. Finally, we ask three control questions regarding age, loyalty program experience, and gender. We add the questionnaire of the experiment in Appendix C.
4.5 Sample size and the sample
We calculate the sample size with GPower analysis, whereby we assume a Type-I error protection of 0.05, a power of 0.8, and use a ‘small’ effect size which is according to
Sawilowsky (2009) 0.2. GPower indicates a sample size of 327 participants, which means that each treatment group needs to have at least 55 participants.
The online experiment took place between the 3rd and 4th of May 2021. We recruited the participants via MTurk. The raw data consists of 384 participants who agreed to
participate in the experiment, whereby 67 participants were removed from the dataset because they did not answer the control question right (n = 66) or had an age of 478 (n = 1).
Furthermore, the variables are Likert scales or manipulated. Therefore, we do not have to check for outliers since there are no extreme values. Moreover, no participants were below 18 years old. Thus, the final dataset consists of 317 participants, who are evenly distributed among the three personalization types (individualized personalization, n = 107; segmented personalization, n = 107; no personalization, n = 103) and the two communication channels (online communication channel, n = 162; offline communication channel, n = 155). The sample consists of 63.4% male, 36.3% female, and 0.3% non-binary, with an average age of 33 years old.