1
Does more ad personalization work better?
Unlocking the value of personalized advertising for the consumers who are
in different stages of decision-making process.
Martyna Gorączka Student ID: 11427809
Master Thesis
Graduate School of Communication Master’s programme Communication Science
Supervisor: Sophie Boerman Date of completion: 31.01.2018
2 Abstract
This study aims to understand how different levels of personalization influence ad and brand attitude through ad relevance, for the consumers who are in different stages of decision-making process. Building on academic theories, such as Elaboration Likelihood Model and Construal-Level Theory, I compared the effect of high and low level of personalization on ad and brand attitude for the consumers who are in purchase decision and post-purchase stage of decision-making process by conducting an experiment (N=203). Moreover, I tested the mediation effect of ad relevance. Results of this study showed that in a post-purchase stage, high level of personalization (vs. low) increases ad relevance and it has a positive effect on ad and brand attitude. In addition, I found a positive effect of ad relevance on ad and brand attitude. Theoretically, this Master Thesis provides insights into how different levels of personalization influence ad relevance taking into account stages of decision-making process, and how this process affects ad and brand attitude. For marketers, this research gives
suggestions how to improve their online personalized advertisements in order to increase positive effects on attitudes toward their ads and brands.
3
Introduction
Nowadays, online advertising gives many opportunities for a brand, especially with the development of data-driven analytics, which helps marketers strengthen their marketing strategy. Most of the advertisers relay on data, which is an important source of information when designing promotion strategies for online campaigns (Wedel & Kannan, 2016).
Important information about consumers’ behavior online is relatively easy to obtain, and with tools such as Google Analytics, Advertising.com or ValueClick.com, equally simple to analyze (Malthouse & Li, 2017; Trusov, Ma & Jamal, 2016). Advertisers can use information about consumer browsing history, shopping behavior, interests and personal characteristics to present specific ads to the particular viewers, in the right time, and in the right context. This promotion strategy is called personalized advertising.
According to professionals, personalized advertising should help brands to reconnect with consumers by showing them relevant ads across different devices on different platforms (Dent, 2017). It should also provide a better experience by reducing cognitive overload for users who receive relevant content at the right time (Aquirre, Mahr, Grewal, de Ruyter & Wetzels, 2015). Indeed, the effectiveness of personalized advertising has been empirically proven by many studies that have shown that personalized advertising causes increases in brand’s revenues (Dias, Locher, Li, El-Deredy & Lisboa, 2008) and click-through rate (Yan,
Liu, Wang, Zhang, Jiang & Chen, 2009).
However, while personalized advertising gives many opportunities to advertisers and potential consumers, there is also empirical evidence that personalized advertising may backfire. In particular, it may have a negative effect on the ad’s effectiveness in terms of the consumer’s attitude toward an ad and advertiser. Personalized advertising may cause feelings
of vulnerability (Aquirre, Mahr, Grewal, de Ruyter & Wetzels, 2015) and intrusiveness (Doorn & Hoekstra, 2013) as consumers may consider personalization as a threat to their
4
privacy and freedom of choice (Aquirre, Mahr, Grewal, de Ruyter & Wetzels, 2015).
Consequently, it may cause ad skepticism and eventually lead to reactance of an ad (Baek & Morimoto, 2012).
Studies also have shown that various responses to ad personalization may depend on the levels of personalization of an ad (Boerman, Kruikemeier & Borgesius, 2017), and the multistage nature of the consumers’ decision-making process (Lambrecht & Tucker, 2013). Levels of ad personalization can be based on the consumer’s online behavior where high level of personalization refers to the ad, which includes a specific product that the consumer has had in their ‘basket’ on the brand’s website (Bleier & Eisenbeiss, 2015). Low level of personalization can be reflected in an ad that presents a random sample of the brand’s products from the website. When it comes to stages of decision-making process, people may react differently to the different levels of personalized content. It is because consumers who only decided which products they want to buy (purchase decision stage) or already purchased a product (post-purchase stage) have different needs, wants and preferences (Belch & Belch, 2015; Lambrecht & Tucker, 2013).
Taking all the above into account, the purpose of this study is to investigate how different levels of advertising personalization (low vs. high) affect both ad relevance and consequently ad and brand attitude. This study also aims to examine how such effect of personalization may vary depending from the consumer’s stages of decision-making process
(purchase decision vs post-purchase stage). Therefore, I formulated the following research question:
How do different levels of personalization (low vs. high) and ad relevance affect ad and brand attitude for consumers who are in different stages of the decision-making process (purchase decision vs. post-purchase stage)?
5
I study these variables because advertising personalization and ad relevance has been shown to influence ad and brand attitude, which are important outcomes of personalized advertising (Boerman, Kruikemeier & Borgesius, 2017; De Keyzer, Dens & De Pelsmacker, 2015). It can further influence purchase intention (MacKenzie & Lutz, 1989; MacKenzie, Lutz & Belch, 1986;), intention to receive other advertisements (Tsang, Ho & Liang, 2004), and other consumer’s behaviors such as click-through rate (Bleier & Eisenbeiss, 2015) or
purchase behavior (Doorn & Hoekstra, 2013). However, to my knowledge, only one study (Lambrecht & Tucker, 2013) investigated the effect of levels of ad personalization taking into account the multistage nature of consumer’s decision-making process. Therefore, from academic perspective, this Master Thesis contributes to the literature on personalized advertising and its effects on ad and brand attitude, by investigating moderation effect of different stages of the decision-making process. In particular, a unique contribution of my thesis is adding the post-purchase stage of the decision-making process to stages already investigated by prior research: purchase decision stage and information search stage (Lambrecht & Tucker, 2013). For practitioners, this research may be very helpful in more effective and profitable operationalization of online advertising campaigns and help in an effort to avoid situations in which personalized advertising may elicit negative attitude toward an ad and a brand.
Theoretical background
Effects of different levels of personalization on ad and brand attitude
Personalized advertising can be defined as “a communication strategy that involves incorporating elements in a message that refer to each individual recipient and are based on the recipient’s personal characteristics, such as name, gender, residence, occupation, and past
6
specific ads based on information collected about a particular consumer, so that the ad responds to the consumer’s personal characteristics and is in accordance with his or her interests or online behavior. Due to its many opportunities, personalized advertising is a very popular way for advertisers to reach consumers (Boerman, Kruikmeier & Borgesious, 2017) and increase a brand’s revenue (Dias, Locher, Li, El-Deredy & Lisboa, 2008).
However, results of empirical researches that have focused on the effectiveness of personalized advertising are mixed. For instance, the positive effect of personalized
advertising on ad attitude has been found by Lee, Kim and Sundar (2015), who investigated personalized advertising on mobile phones based on the consumer’s location. On the contrary, another study has shown that consumers have generally negative attitude towards
personalized mobile ads (Tsang, Ho & Liang, 2004). Consumers only had favorable attitudes toward personalization, when the ad was sent with their permission. Further, Maslowska, Smit and Putte (2013) found out that the positive effect of personalized advertising on ad and brand attitude only occurs in less savvy countries, such as Poland. In more marketing-savvy countries like the Netherlands, consumers had a less favorable attitude toward the brand and the personalized ad.
These mixed results of empirical researches regarding the effectiveness of
personalized advertising can be explained by the so-called ‘personalized paradox’. According to Aquirre, Mahr, Grewal, de Ruyter and Wetzels (2015), it occurs when consumers on the one hand like advertisements, which are more personalized and hence more relevant for them, but on the other hand, they may have feeling of vulnerability and experience discomfort. Consequently, studies have shown that different responses to the personalized advertisement may depends from the consumer’s privacy concerns (Baek & Morimoto, 2012), levels of personalization (Doorn & Hoekstra, 2013) or trust in the advertiser (Bleier & Eisenbeiss, 2015). For instance, resistance to personalized ads differs among consumers who have
7
different beliefs and feelings regarding privacy (Baek & Morimoto, 2012). Consumers with high level (vs. low level) of privacy concerns, perceived risk and feelings of vulnerability may have a more skeptical attitude toward personalized ad, and in result, they can more often avoid personalized ads.
The response to personalized advertising may also vary depending from the levels of personalization, which are based on the type and amount of data used by the advertiser
(Boerman, Kruikmeier & Borgesius, 2017). The more information an advertiser is using about its consumer – such as demographics, interests or online behavior – the higher the level of personalization of an ad is. According to van Doorn and Hoekstra (2013), a higher level of personalization (vs. lower) has a negative effect on purchase behavior as it elicits feelings of intrusiveness. The relationship between different levels of personalization of an ad, and effectiveness of that ad can be also moderated by consumers’ trust in advertiser (Bleier &
Eisenbeiss, 2015). According to the study, a higher level of personalization of an ad of a less trusted retailer has a negative effect on the effectiveness of an ad. On the contrary, a more trusted advertiser might expect to have higher effectiveness metrics of online advertising such as click-through rate, when using a highly personalized banner.
In particular, the negative effects of a high level of personalization are often explained with Psychological Reactance Theory (Baek & Morimoto, 2012; Bleier & Eisenbeiss, 2015). This theory was developed by Brehm and Brehm (1981), and assumes that people may resist a persuasive message when they feel that it threats their personal freedom. Because
personalized ads are designed based on personal information about the consumer, it may cause feeling of intrusiveness because consumer may notice these information and
consequently cause negative feelings toward an ad and advertiser (Aquirre, Mahr, Grewal, de Ruyter & Wetzels, 2015). Eventually, the consumer may refuse or avoid personalized
8
To sum up, the effectiveness of personalized advertising may vary, depending from the consumer’s personal characteristics (Baek & Morimoto, 2012; Bleier & Eisenbeiss, 2015) but also levels of ad personalization (Boerman, Kruikmeier & Borgesius, 2017; Doorn & Hoekstra, 2013). Based on Psychological Reactance Theory (Brehm & Brehm, 1981) prior literature (Baek & Morimoto, 2012; Bleier & Eisenbeiss, 2015; Doorn & Hoekstra, 2013) suggests that more personalized advertisement have a negative effect on ad and brand
responses as it may cause feeling of intrusiveness and privacy concerns. Hence, I propose the following hypothesis:
H1: High level of personalization (vs. low) has a negative effect on consumer’s (a) ad and (b) brand attitude.
The mediating effect of ad relevance
While a number of studies suggest negative effect of ad personalization (Baek & Morimoto, 2012; Doorn & Hoekstra, 2013; Bleier & Eisenbeiss, 2015), another studies have shown that personalized advertising can also have a positive effect, as personalization for instance, increases ad relevance. This in turn may positively influence advertisement outcomes such as click-through rate (Yan, Liu, Wang, Zhang, Jiang & Chen, 2009), ad and brand attitude (Maslowska, Smit & Putte, 2013; De Keyzer, Dens & De Pelsmacker, 2015) among others.
Positive effect of personalized advertising
The positive effects of personalization of advertising has been proven in prior studies that have shown that personalization based on consumer’s behavior can increase click-through rate even by 670% (Yan, Liu, Wang, Zhang, Jiang & Chen, 2009). Improvements in this context can be done by using more information about consumer’s behavior online and
therefore present more personalized content to the receiver (Chen, Pavlov & Canny, 2009). The positive effect of using more data about consumers is also shown in another study done
9
for Amazon.com. According to the researchers, scaling data about the consumer’s purchased and rated items with similar items and further creating a recommendation list with those items is the most effective way for brands to use personalized ads (Linden, Smith & York, 2003). In addition, further research has shown that it may help not only to increase a company’s
revenue from the purchase of recommended items but also helps overcome shopper resistance towards for example fresh produce online (Dias, Lochar, Li, El-Deredy & Lisboa, 2008). These optimistic results regarding personalization are also supported by earlier research on the topic, which has shown that perceived personalization has a positive effect on trust in recommendation agents and their integrity (Komiak & Benbasat, 2006).
Previous studies on the topic of personalization, also has shown the effectiveness of personalized websites that match the consumer’s personal characteristics such as the
customers’ cognitive styles (Hauser, Urban, Liberali & Braun, 2009), and the effectiveness of
personalized content of e-mails on website traffic (Ansari & Mela, 2003). Two studies have found positive effects regarding the personalization of the content of the websites and e-mails, respectively. According to Hauser, Urban, Liberali and Braun (2009), in order to increase the firm’s sales and create a more preferable website, companies should segment its audience
according to the consumer’s cognitive style (e.g., analytics) and personalize the website accordingly. Personalization of content is also profitable in case of one-to-one marketing. Based on results of Ansari and Mela’s (2003) investigation, companies should personalize e-mails to potential consumers in order to increase click-through rate and website traffic. This strategy helps reduce information overload and therefore influence consumer’s reaction to the brand marketing actions in a positive way.
Positive effect of personalization on ad and brand attitude
Besides all these positive personalization outcomes, personalized advertising may also influence ad and brand attitude. To explain how personalized advertising may positively
10
influence ad and brand attitude, prior studies (De Keyzer, Dens & De Pelsmacker, 2015; Maslowska, Smit & Putte, 2013) build upon the Elaboration Likelihood Model (ELM) (Petty & Cacioppo, 1986) and self-referencing (Tam & Ho, 2006; De Keyzer, Dens & De
Pelsmacker, 2015).
According to the Elaboration Likelihood Model, people can process a given message in two ways. Under the central route to persuasion, an individual has a motivation to process the given message, which is influenced by personal relevance, involvement, and the
individual’s needs and arousal level (Belch & Belch, 2015). Under the peripheral route to
persuasion, the receiver of the message has no motivation and/or ability to process the information. It can be caused by the fact that he or she do not find the message relevant or compressible (Petty & Cacioppo, 1986). Since personalized advertisement is based on the consumer’s characteristics and/or behavior, the ad becomes more self-relevant for the receiver
(increases ad relevance) as it includes information/ products that the consumer may be interested in. This process in accordance with ELM should lead to processing via the central route and has, therefore, an influence on ad and brand attitude (De Keyzer, Dens & De
Pelsmacker, 2015; Maslowska, Smit & Putte, 2013). The positive effect of ad relevance on ad and brand attitude can be then caused by self-referencing (Tam & Ho, 2006) which in context of personalized ads, means, that the consumer would find personalized information more interesting.
Therefore, based on Elaboration Likelihood Model (Petty & Cacioppo, 1986) and prior research (De Keyzer, Dens & De Pelsmacker, 2015; Maslowska, Smit & Putte, 2013), it’s assumed, that more personalized advertisements lead to increases in ad relevance and this,
in line with assumptions of self-referencing causes a more positive ad and brand attitude. I propose second hypothesis, which includes the mediation effect of ad relevance:
11
H2: High level of personalization (vs. low) increases ad relevance, and it has a positive effect on (a) ad and (b) brand attitude.
The moderating effect of stages of decision-making process
Next to the mediation effect of ad relevance, the response to levels of ad
personalization may vary, depending on the stages of the decision-making process in which the potential consumers are (Tucker & Lambrecht, 2013). Prior research explains this relationship with the Construal-Level Theory (Trope & Liberman, 2010; Tucker &
Lambrecht, 2013). In the following section, I will discuss the possible moderation effects of various stages in the decision-making process and therefore propose a moderated mediation model presented with Figure 1.
Stages of decision-making process
For the purpose of this research, I have chosen two stages in the decision-making process, namely purchase decision stage and post-purchase stage. These two stages represent two significantly different phases in the consumer’s shopping journey, and clearly relate to different consumer needs and wants. The purchase decision stage occurs, when the consumer developed a purchase intention or predisposition to purchase a certain brand (Belch & Belch, 2015). In particular, it takes place when the attributes or characteristics of a brand under consideration match the purchase motives (Belch & Belch, 2015). The second stage chosen in this study is the post-purchase stage, which simply occurs when the consumer has bought the product of a particular brand.
Construal-Level Theory
According to Construal-Level Theory (CLT) (Trope & Liberman, 2010), people can think about objects in two ways, depending on how abstract a particular representation becomes for the individual. Trope and Liberman (2010) have distinguished two construal levels – high vs. low level – that represent, more broadly or more narrowly constructed
12
thoughts about particular object respectively. In context of shopping behavior, consumer may have very broad idea what he wants to buy, and thus have constructed preferences at high level. The consumer may also know precisely which product, of which brand he wants to get, and therefore has the preferences constructed at low level.
Built on the Construal-Level Theory, Tucker and Lambrecht (2013) have shown that there is a relationship between different stages of the decision-making process – which represents two construal levels – and the reaction to the different levels of personalized advertising. In particular, the research holds that consumers who have broadly constructed preferences (high-level construal), so the ones who are in, for example information stage – they are only searching for potential brand and products – respond more positively to ads which are less personalized. It is because these advertisements match consumer’s needs by
depicting a more general and broad idea of the brand’s products. On the contrary, consumers who have narrowly constructed preferences regarding a particular product (low-level
construal), so the ones who are in, for instance purchase decision stage of decision-making process, respond positively to more personalized ads (Tucker & Lambrecht, 2013). This effect occurs because a more personalized ad presents products that the potential consumer was browsing or had in his ‘basket’ on the brand’s website. The ad therefore matches the
preferences of the consumer, who already know what he or she wants.
Therefore, based on the Construal-Level Theory (Tucker & Lambrecht, 2013) and previous research (Tucker & Lambrecht, 2013), I expect that consumers in decision stage already have narrowly constructed preferences regarding products and brands they want to buy and hence a more personalized ad (high level of personalization) will be more accurate and more relevant for them. This in turn, based on Elaboration Likelihood Model and self-referencing, will have a positive effect on ad and brand attitude. I therefore propose:
13
H3a: In the purchase decision stage, a high level of personalization (vs. low) increases ad relevance, and it has a positive effect on (a) ad and (b) brand attitude.
On the contrary, consumers who are in the post-purchase stage and have already bought particular brand’s products will react differently to a high level of personalization.
Since an ad presents products which they already purchased, the ad is not relevant for them anymore. This, based on self-referencing should cause a negative ad and brand attitude. Hence, I also hypothesize:
H3b: In the post-purchase stage, a high level of personalization (vs. low) decreases ad relevance and it has a negative effect on (a) ad and (b) brand attitude.
a) b)
c)
Figure 1.Proposed moderated mediation model
Methods
In order to answer the main research question and test related hypotheses, I choose a 2x2 between-subjects design with two groups for levels of personalization (high vs. low) and two groups for stages of decision-making process (purchase decision vs. post-purchase stage).
Level of personalization: low vs. high Ad attitude Brand attitude Ad relevance Stages in decision-making process
14 Participants
The population of this research is defined as males and females who shop online. Only consumers who have experience with online shopping may experience personalized
advertising based on their online shopping behavior. I formulated the inclusion and exclusion criteria for the selection of participants. I excluded from the analysis respondents who had no experience with online shopping. Moreover, since my experiment was conducted fully in English and not all participants were native language speakers (e.g. Polish or German), I expected from them to have good knowledge of the language. Other respondents were excluded from the further analysis.
In this study, I recruited in total 208 international participants, with the mix of convenience and snowball sampling. I used social media sites such as Facebook and Instagram to approach potential respondents. Each individual was asked to take part in the following online experiment by clicking on a hyperlink and then he or she was randomly assigned to one of four conditions of the experiment. Based on the exclusion criteria, I excluded from the analysis three respondents who indicated that they “never” shop online, and two respondents who indicated having not good English proficiency on “2” level (in scale 1-7 where 1=not well at all and 7=extremely well). In total, 203 participants were selected for the final analysis, including 128 females (63.1%) and 75 males (36.9%) with the average of age 25 years, with the youngest 18 years old, and the oldest 45 years old respondent
(M=25.15, SD=3.93). Procedure
After clicking on a hyperlink, participants of the online experiment were directed to the page, where, all the respondents had to give consent to take part in the online experiment. Each individual was exposed to the one of the four scenarios which reflect one of four
15
level of personalized ad and post-purchase stage’, ‘low level of personalized ad and purchase decision stage’ and ‘low level of personalized ad and post-purchase stage’. All given
scenarios were based on the Lambrecht and Tucker (2013) experiment, with modified examples of advertisements and scenarios for the purpose of the research. After that, respondents were asked to answer for three questions, related to the given scenarios and advertisements which were aimed to measure ad relevance, ad attitude and brand attitude. Further, I conducted manipulation check, and displayed four control variables (gender, age, installed ad blocker, likelihood of buying products from given scenario in real life), and variables which could help select proper sample of the experiment (experience with online shopping, English proficiency). The procedure was closed with debriefing about real purpose of the study (full questionnaire can be found in the Appendix I). Participation in the
experiment took approximately 5 minutes. Materials
Pre-test
The effectiveness of the initial stimulus material (two scenarios and two ads) was pre-tested. Respondents were asked to evaluate ads in terms of observed level of personalization, by using scale created by Maslowska, Smit and Putte (2016) (further also used as a manipulation check). Scenarios were evaluated in terms of the stages of the decision-making process in which is consumer. Respondents were asked to answer the following question: “How would
you describe this stage in your shopping process?” with four possible options: (1) “I’m still searching for product/products that I want to buy and consider different brands”, (2)”I made the decision to buy a particular product/products”, (3) I already have bought a particular product/products, (4) “I just realized I need to buy a particular product/products. Moreover, scenarios were evaluated in terms of realism. Respondents were asked: “How realistic,
16
according to you, was the given scenario?” with answer coded in 7-point Likert-type scale (1=not realistic at all, and 7=very realistic).
In total, 21 respondents took part in the pre-test. The average age was 24 (M=24.98,
SD=1.94), 57.1% were male. The realism of the scenarios was rated by participants on
average with 6,29 (SD=0.96), which means that overall the scenario appeared to be perceived as very realistic. Further, the results of the pre-test has shown that in the purchase decision stage, 7 out of 11 respondents (63.6%), indicated correct stage in decision-making process with the answer: (2)”I made the decision to buy a particular product/products”. Four
respondents (36.4%) answered incorrectly for the given question, with the answer (1) “I’m still searching for product/products that I want to buy and consider different brands”. In order
to improve stimulus material, I decided to add the following sentence in the final experiment: “Remember that you are not considering any other brands or products, as you already
decided which headphones and extension cords of the Sweetsound brand you want to buy”. When it comes to post-purchase stage, 90% of the respondents (9 out of 10) indicated correct stage in decision-making process, choosing answer (3)” I already have bought particular product/products”. This outcome of the pre-test was satisfying hence, I decided not to change
anything in this part of stimulus material.
The results of independent sample t-test which was used to pre-test the levels of personalization of given ads indicated that there is no statistically significant difference, between high (M=4.86, SD=0.84) and low level of personalization (M=4.32, SD=1.38) perceived by respondents t(19)=1.093, p=.288, CI[-.49, 1.57). Levene’s test for equality of variance was not significant (F=3.739, p=.068) which means that equal variances were assumed. Since the initial ad with a low level of personalization presented different type of headphones or earphones that appeared to be similar to the high level of personalization, I decided to improve my stimulus material by changing the ad with low level of
17
personalization. I designed a new ad, with less related products to the headphones and extra extension cords such as microphones and speakers, removing therefore other examples of headphones and headphones accessories. The ad with high level of personalization remained the same.
Final stimulus material
To manipulate the independent variable (low vs. high level of personalization), I created two ads based on Bleier and Eisenbeiss (2015) definition of levels of personalization. An ad with high level of personalization presented products, which the consumer decided to buy, – so put in the ‘shopping bag’ – or already have bought (see Figure 2). An ad with low level of
personalization presented the sample of products from the brand’s website (see Figure 3). These two types of ads presented a non-existing brand called Sweetsound with electronics products (headphones, earphones) and related electronic products (microphones, speakers). Both ads were created especially for the purpose of this experiment and have never been used in real life.
In order to manipulate moderation variable (purchase decision stage vs. post-purchase stage), I designed two scenarios, based on Lambrecht and Tucker’s (2013) experiment. For the decision stage in the decision-making process, I asked respondents: “Imagine that you would like to buy headphones and extra extension cords to that product. You were browsing the Internet, and decided to choose the brand Sweetsound and a specific model of the
headphones and extension cords. You put both products to your ‘shopping bag’ (‘basket’) on the brand’s website. You can see them below (after 5 sec. you can skip to the next question).”
I also showed them described products (see Figure 4). For the post-purchase stage in decision-making process, I asked respondents: “Imagine that you recently have bought online
18
see them below (after 5 sec. you can skip to the next question)”. I also showed them the described products (see Figure 4).
Figure 2. Ad with high level of personalization Figure 3. Ad with low level of personalization
Figure 4. Described products
Measures Ad relevance
Ad relevance was measured by using 7-point Likert-type scale (1 – strongly disagree, 7 –
19
the following statements? This ad is…” followed by six attributes based on a scale measuring
ad relevance (Zaichkowsky, 1994): “important”, “relevant”, “means a lot to me”, “valuable”, “interesting”, “exciting”, “appealing”, “fascination” , “needed” and “involving”. Factor
analysis revealed the items load on one factor (Eigenvalue=6.65; explained variance=66.53%; Cronbach’s alpha=.94). High scores of ad relevance correspond to more ad relevance,
whereas low scores correspond to less ad relevance (M=3.23, SD=1.39). The mean score of the six items has been used as a measurement of ad relevance in the further analysis.
Ad attitude
In order to measure ad attitude, participants were asked to answer the question: “How would you rate the ad above? This ad is…” followed by a six-item, seven-point semantic differential
scale (Spears & Singh, 2004): unpleasant/pleasant, unlikable/likable, boring/interesting (reversed), tasteless/tasteful, negative/positive and bad/good. Since I used reverse scale for one item (boring/interesting), I first reversed the coding of this variable before I have run factor analysis. Factor analysis revealed the items load on one factor (Eigenvalue=3.86; explained variance=64.48%; Cronbach’s alpha=.87). High score of ad attitude correspond to more positive attitude toward an ad, whereas low scores correspond to more negative attitude toward an ad (M=5.16, SD=1.41). The mean score of the six items has been used as a
measurement of ad attitude. Brand attitude
In order to measure brand attitude, respondents were asked: “How would you evaluate the brand? Brand Sweetsound is…” followed by an eleven -items, seven-point semantic differential scale (Spears & Singh, 2004): unappealing/appealing, bad/good,
unfavorable/favorable, undesirable/desirable, worthless/valuable, unlikable/likable, not nice/ nice, useless/useful (reversed), unattractive/attractive, not advisable to use/advisable to use, unenjoyable/enjoyable. Factor analysis revealed the items load on one factor
20
(Eigenvalue=7.54; explained variance=68.5%; Cronbach’s alpha=.95). High scores of brand attitude correspond to positive brand attitude, whereas low scores correspond to more
negative brand attitude (M=4.11, SD=1.13). The mean score of eleven items has been used as a measurement of brand attitude.
Manipulation check
To check whether the stimulus material was perceived as intended, I added four questions considered as a manipulation check. Using 7-point Likert-type scale (1=strongly disagree, 7=strongly agree) all respondents had to answer to what extend they agree with the following statements (Maslowska, Smit & Putte, 2016): “I had a feeling that I was addressed personally in the advertisement”, “The advertisement was targeted at me”, “I could recognize myself in the group the advertisement was targeted at”, “I noticed personal information in the
advertisement”. Factor analysis has shown that the items load on one factor
(Eigenvalue=2.81; explained variance=70.4%; Cronbach’s alpha=.85). High scores of manipulation check indicate higher level of personalization, whereas low scores indicate lower level of personalization. The mean of score of the four items is used as a measurement of ad personalization level (M=4.18, SD=1.63).
Control variables
In total, four control variables were measured in order to exclude possible effect of them on the dependent variables: gender, age, likelihood of buying products from the scenarios, installed ad-block. To measure gender and age, I asked participants to answer the questions: “What is your gender” (1=male, 2=female, 99=other) and “How old are you? Please write
only a number (e.g.20)” respectively. Most participants were female (63.1%) with the minimum age of 18 and maximum age of 45. On average age for the sample was 25
(M=25.15, SD=3.93). To measure likelihood of buying products from the scenarios, I asked them “How likely would you buy online products from the given scenario (headphones and
21
extra extension cords) in real life?”. The possible answer was measured using 7-point
Likert-type scale (1-extremely unlikely, 7-extremely likely). On average, respondents indicated that they would “slightly likely” buy products from the given scenario (M=3.57, SD=1.72).
Further, I asked participants “Do you have installed an ad blocker (such as AdBlock) on your computer/laptop?” (1=no, 2=yes). Out of 203 respondents, 146 declared to have installed an ad blocker (71.9%).
In addition, in order to select the sample, I measured experience with online shopping by asking “How often do you shop online?” (1=Never, 2=Yearly, 3=Monthly, 4=Weekly, 5=Daily).On average, respondents declared monthly shop online (M=3.00, SD=0.69). I measured English proficiency by asking “How well you can speak English?” using 7-point Likert-type scale (1-not well at all, 7-Extremely well). On average participants had knowledge of English equivalent to the 6 point on the 7-point Likert-type scale (M=6.02, SD=0.94).
Results
Manipulation checkTo check whether the stimulus material was perceived as intended, thus whether there is difference in perception of low vs. high level of personalization I conducted an independent sample t-test to compare means with levels of personalization as the independent variable and combined variables for the measurement of perceived personalization as the dependent variable. Results of the t-test have shown, that there is no statistically significant difference between high (M=4.37, SD=1.65) and low level of personalization (M=3.98, SD=1.59) perceived by respondents t (201)=1.731, p=.085, CI[-.05, .84]. Levene’s test for equality of variances was not significant (F=0.110, p=.740) which means that equal variances were assumed. Overall, it means that stimulus material has failed. Nonetheless, the material was still used for the purpose of the further analysis.
22 Randomization
The four experimental groups do not differ with respect to age, F(3.199)=0.871, p=.457, likelihood of buying products from the given scenarios, F(3.199)=0.230, p=.875, online shopping experience, F(3.199)=0.600, p=.616, and having installed ad blocker x2(3)=4.234,
p=.237. Despite random assignment, there is statistically significant difference between
groups when it comes to gender, x2(3)=9.446, p=.024, hence I will be controlling for gender in all the further analyses to be sure that it does not confound the effects.
Effect of levels of personalization
To test H1, thus, whether high (vs low) level of personalization has a negative effect on the consumer’s ad and brand attitude, I conducted Multivariate Analysis of Covariance
(MANCOVA) with levels of personalization as the independent variable, ad and brand attitude as dependent variables and gender as a control variable. The result showed statistically significant difference between high vs. low level of personalization on the combined dependent variables after controlling for gender, F(2.199)=3.942, p=.021, Pillai’s Trace=.04, partial η2=.04. Further MANCOVA showed marginally significant main effect of levels of personalization on ad attitude. It means that ad attitude do not differ in low (M=3.56,
SD=1.30) and high (M=3.86, SD=1.14) level of personalization F(1.200)=2.837, p=.094, η2
=.014 . Similar, brand attitude do not differ in low (M=4.18, SD=1.15) and high (M=4.05,
SD=1.11) level of personalization F(1.200)=0.897, p=.345, η2=.004. There is also no main effect of gender on ad F(1.200)=0.232, p=.630, η2=.001 and brand attitude F(1. 200)=0.604,
p=.438, η2=.003. Thus, H1 is rejected. Mediation effect of ad relevance
To test H2, thus the indirect effect of levels of personalization on ad and brand attitude through ad relevance, with gender as covariate, I used Model 4 of Andrew Hayes’ PROCESS v3.0 macro in SPSS. This method uses 5,000 bootstrapped samples to estimate the bias
23
corrected and accelerated confidence intervals (BCACI) (Preacher & Hayes, 2008). I ran separate analyses for each dependent variable (ad and brand attitude), with levels of
personalization as independent variable (high vs. low), ad relevance as mediator, and gender as covariate.
The results showed no significant main effect of levels of personalization on ad relevance (b=0.22, p=.254). However, ad relevance has a significant positive effect on ad attitude (b=0.45, p<.001), which means that more relevant advertisement for the consumers, leads to more positive ad attitude. Further, the analysis showed no significant main effect of levels of personalization on ad attitude when controlling for gender (b=0.19, p=.205), and no significant mediation or indirect effect of levels of personalization on ad attitude through ad relevance (indirect effect=.10, SE=.08, 95% BCACI [-.07, .27]).
When it comes to brand attitude, similar results are noted. Ad relevance has a
significant positive effect on brand attitude (b=0.30, p<.001), which means that more relevant advertisement for the consumers leads to more positive brand attitude. Further, the analysis showed no significant main effect of levels of personalization on brand attitude when controlling for gender (b=-0.22, p=.142), and no significant mediation or indirect effect of levels of personalization on brand attitude through ad relevance (indirect effect=.06, SE=.06, 95% BCACI [-.05, .18]).
This means that a higher level of personalization (vs. low) does not positively affect ad and brand attitude as result of increased ad relevance. However, more relevant ad leads to a more positive ad and brand attitude. Thus, H2 is partly supported as ad relevance has a positive effect on ad and brand attitude, however, not as result of higher level of personalization.
Moderated mediation effect of stages of decision-making process and ad relevance To test H3a and H3b, I used Model 7 of Andrew Hayes’ PROCESS v3.0 macro in SPSS. I ran
24
separate analyses for each dependent variable (ad and brand attitude) with the levels of personalization (high vs. low) as the independent variable, stages of the decision-making process (purchase decision vs. post-purchase stage) as moderator, ad relevance as mediator and gender as control variable. The proposed moderated mediation model is presented in Figure 5.
The result showed a significant positive interaction effect of levels of personalization and stages of the decision-making process on ad relevance (b=0.81, p=.035) when controlling for gender. In addition, moderated mediation is significant (index of moderated mediation:.37,
SE=.19, 95% BCACI [.02, .77]) meaning that higher level of personalization leads to more
positive ad attitude through increases in ad relevance, however, this positive effect is
significant only for post-purchase stage (indirect effect: .26, SE=.12, 95% BCACI [.03, .52]), and not for purchase decision stage (indirect effect: -.10, SE=.13, 95% BCACI [-.38, .13]) (see Table 1).
When it comes to brand attitude, results are similar. Moderated mediation is
significant (index of moderated mediation:.24, SE=.14, 95% BCACI [.00, .57]), meaning that higher level of personalization leads to more positive ad attitude through increases in ad relevance, however, this positive effect is significant only for post-purchase stage (indirect effect:.17, SE=.09, 95% BCACI [.02, .38]) and not for purchase decision stage (indirect effect:-.06, SE=.09, 95% BCACI [-.27, .09] (see Table 1). In other words, higher level of personalization leads to more positive ad and brand attitude, through increases in ad relevance but only for post-purchase stage. This means that H3a must be rejected, and H3b is partially supported as reverse effect occurs.
25
s
tages of decision-making processd a
ad relevanceb
c
levels of personalization
ad attitude
c’
brand attitudeControl variable
gender
Figure 5. Model of moderated mediation effect of stages of decision-making process and ad relevance
Table 1. Coefficients represent standardized beta weights (with boot SE between parentheses);
significant indirect effects are bold; N=203; *p<.05, **p<.001
Variable Indirect effect a b c (direct) c’ (total) d
Ad attitude .37(.19)[.02,.77] .22 (.19) .45(.05)** .19(.15) .10 (.08) .81(.38)* Post-purchase .26(.12)[.03,.52] Purchase decision -.10(.13)[-.38,.13] Brand attitude .24(.14)[.00,.57] .22(.19) .30(.05)** .22(.14) .06 (.06) .81(.38)* Post-purchase .17 (.09)[.02,.38] Purchase decision -.06(.09)[-.27,.09]
Discussion
Recently, in order to increase a brand’s revenue, marketers are using increasing amount of data to design more personalized online advertisements and present them to consumers who have visited the brand’s website (Wedel & Kannan, 2016). However, the effects of
personalized advertising are still debatable. The main goal of this Master Thesis was to unlock the value of personalized advertising for the consumers who are in different stages of
26
low) level of personalization on ad and brand attitude, taking into account moderated mediation role of stages of decision-making process and ad relevance.
The results demonstrated that in post-purchase stage, a high level of personalization (vs. low) increases ad relevance, and it has a positive effect on ad and brand attitude. In other words, consumers who purchased a brand’s products, consider more personalized ads as more relevant in comparison to less personalized ads, and hence evaluate the ad and brand more positively. This reverse effect to what I expected perhaps can be explained with the cognitive dissonance which occurs when the consumer experiences ‘discomfort’ after purchasing a product (Belch & Belch, 2015). The consumer may seek out reassurance, and be more attentive to advertisements of the products and brand that he or she purchased. Due to cognitive dissonance, highly personalized advertisements may become more relevant for the consumers who are in post-purchase stage and consequently, in line with ELM and self-referencing, elicit more positive attitudes toward the ad and brand. Although these findings are still in accordance with prior studies that showed a positive effect of personalization on ad relevance (Yan, Liu, Wang, Zhang, Jiand & Chen, 2009; Chen, Pavlov & Canny, 2009), and consequently on ad and brand attitude (Maslowska, Smit & den Putte, 2013), the assumption of cognitive dissonance as a reason for increase in ad relevance should be investigated in further research.
Further, the same analysis has shown that the mediation effect is not significant for consumers who are in the purchase decision stage. It means that for the consumers who decided what to buy, more personalized advertisements do not become more relevant and this in turn do not elicit more positive ad and brand attitudes. It means that they either do not have narrowly constructed preferences regarding what to buy, or that these preferences do not positively correlate with high levels of ad personalization. These findings therefore do not support results of study done by Lambrecht and Tucker (2013), who based on the
Construal-27
Level Theory (Trope & Liberman, 2010) have shown the positive effect of a high level ad personalization for consumers in purchase decision stage.
The following significant finding of my research demonstrated the positive effect of ad relevance on ad and brand attitude. In other words, consumers for whom an ad was relevant had formed a more positive ad and brand attitude. This result is in line with self-referencing which holds that more relevant advertisement allows a consumer to relate the message to oneself that in turn causes more positive attitudes toward an ad and advertiser (Tam & Ho, 2006). These results also support prior researches which have shown that more relevant content for the consumer has a positive effect on, for instance, click-through-rate, (Yan, Liu, Wang, Zhang, Jiand & Chen, 2009), attitude toward the advertisement (Maslowska, Smit & den Putte, 2013) or website traffic (Ansari & Mela, 2003).
When it comes to the Psychological Reactance Theory (Brehm & Brehm, 1981) results of my Master Thesis did not find a support for the theory. I only found a positive effect of high level of personalization on ad and brand attitude through ad relevance for consumers in post-purchase stage, undermining the assumption of the Psychological Reactance Theory which holds that more personalized ads may elicit a more negative attitude towards an ad and brand.
Summarizing all findings, this Master Thesis has shown that ad relevance has a positive effect on ad and brand attitude. Moreover, there is no resistance towards personalized
advertisement. More specifically, higher level of ad personalization increases ad relevance that consequently leads to more positive ad and brand attitude for the consumers who are in the post-purchase stage of the decision-making process. The results also showed that for the consumers who are in the purchase decision stage, levels of ad personalization (high vs. low) do not affect ad and brand attitude through ad relevance, yet there is also no resistance toward high level of ad personalization. Therefore, This Master Thesis shows new insights into how
28
levels of personalization and ad relevance can influence ad and brand attitude for the consumers who are in different stages of the decision-making process.
Limitations
Notwithstanding, it is important to keep in mind, that this research has some limitations. First of all, manipulation check has shown that my stimulus material has failed. It means that people do not perceive ads with high and low levels of personalization differently. This result undermines a previous study, which has shown differences between levels of personalization (Bleier & Eisenbeiss, 2015) based on the consumer’s shopping behavior. However, despite the failure of my stimulus material, interestingly I still have found significant results in my experiment. Second, the sample of study has been selected with convenience and snowball method, that means that my findings cannot be representative of the full population. I suggest replicating this research by using a different sample of respondents to see whether a different population reacts differently to ad personalization. Thirdly, respondents of the experiment were exposed to made-up scenarios and a non-existing brand (Sweetsound) which were created especially for the purpose of this research. This setting provides only a snapshot of reality (Bleier & Eisenbeiss, 2015) that could bias positively my findings. In real life, people react to ads more spontaneously and subconsciously. They might not think about how they evaluate an ad and brand as they can be distracted by other activities. This could cause more positive response to the ad and brand which were presented to the participants of the
experiment. To reduce this limitation, in the future research I suggest perhaps replace the non-existing brand (Sweetsound) with an non-existing brand in order to increase the reality of the circumstances.
Besides tackling these methodological limitations, future research may extend this study conceptually. First, my stimulus material was focused on electronic music equipment, so that ads and scenarios were about buying headphones or other related products. I suggest
29
extending future research to other product categories or services. People may react differently to different type of products, for instance, taking into account low and high involvement type of products. Secondly, I used one specific definition of levels of ad personalization (Bleier & Eisenbeiss, 2015) which was based on consumer’s shopping behavior. However, although this
is one of the most popular ways of personalizing advertising online by the advertisers, there are also other ways for designing ads by using different data about potential consumers. More research is needed to see whether different type of levels of ad personalization, for instance based on consumer’s individual characteristics such as age, gender or geographic, influence
ad and brand attitude. The reaction of consumers to different levels of personalization may vary for ads that are designed based on different information collected about consumers (Bleier & Eisenbeiss, 2015).
Implications for theory and practice
Even though this study has several limitations, it also has some useful practical and societal implications. Firstly, from an academic perspective, my research gives evidence for the assumption of self-referencing that holds that more relevant content causes more positive responses to that content. My findings show that ad relevance always increases positive attitude toward an ad and brand. Concerning Construal-Level Theory (Trope & Liberman, 2010) my research also demonstrated that narrowly constructed preferences of the consumers are not positively correlated with high level of ad personalization. Hence, I suggest replicating the study (Lambrecht & Tucker, 2013) that confirmed the assumption that consumers in the purchase decision stage will react more positively to a high level of ad personalization. For marketers, this finding says that if they want to elicit positive attitudes toward their ad and brand, the content of the ad must be relevant for the receiver, however not necessary as a result of high level of personalization of an ad. They should optimize their ad in a way that it become relevant for the consumer.
30
Second, this Master Thesis has shown an important role of stages in decision-making process in perception of levels of personalization. More specifically, it showed that consumers who are in different stages of decision-making process (purchase decision vs. post-purchase stage) react differently to the personalized content. Consumers who already purchased the brand’s products perceive a more personalized ad as more relevant, that consequently
positively affects ad and brand attitude. Based on this finding, marketers should design highly personalized ads for the consumers who are in post-purchase stage in order to elicit positive ad and brand attitudes. When it comes to the purchase decision stage, according to my Master Thesis, regardless of levels of ad personalization, consumers react the same way to the given ad. This may encourage marketers to test different type of personalization for that target group to see whether other ways of personalization are more effective.
Nowadays, marketers tend to use increasing amount data regarding their consumers. They believe it can increase ad relevance and consequently positively influence ad and brand attitude (Wedel & Kannan, 2016; Dent, 2017). The effectiveness of personalized advertising can also be supported in academia, based on the Elaboration Likelihood Model (Petty & Cacioppo, 1986), self-referencing (Tam & Ho, 2006), and the Construal-Level Theory (Trope & Liberman, 2010). This research also supports the common belief that more relevant content has a positive effect on ad and brand attitude. Nonetheless, the findings of my Master Thesis also show that high level of personalization has a positive effect only in very specific
circumstances, namely, consumers who are exposed to an ad must have already purchased the brand’s products. Only then, the high (vs. low) level of ad personalization may increase ad
31
References
Aquirre, E., Mahr, D., Grewal, D., de Ruyter, K. & Wetzels, M. (2015). Unraveling the personalization paradox: The effect of information collection and trust-building strategies on online advertisement effectiveness. Journal of Retailing, 91(1), 34-49. doi: 10.1016/j.jretai.2014.09.005
Ansari, A., & Mela, C. F. (2003). E-Customization. Journal of Marketing Research, 40 (5), 131–45. doi: 10.1509/jmkr.40.2.131.19224
Baek, T. H., & Morimoto, M. (2012). Stay away from me: Examining the determinants of consumer avoidance of personalized advertising. Journal of Advertising, 41(1), 59-76. doi: 10.2753/JOA0091-3367410105
Belch, G. E. & Belch M. A. (2015). Chapter 2: The role of IMC in the marketing process, pp. 41-65.
Bleier, A. & Eisenbeiss, M. (2015). The Importance of Trust for Personalized Online Advertising. Journal of Retailing, 91, 390-409. doi: 10.1016/j.jretai.2015.04.001 Boerman, S. C., Kruikemeier, S. & Zuiderveen Borgesius, F. J. (2017). Online Behavioral
Advertising: A literature Review and Research Agenda. Journal of Advertising,
46(3), 363-376. doi: 10.1080/00913367.2017.1339368
Brehm, S. S., & Brehm, J. W. (1981). Psychological reactance: A theory of freedom and control. New York: Academic Press.
Chen, Ye, Dmitri Pavlov, and John F. Canny (2009). Large-Scale Behavioral Targeting.
Proceedings of the 15th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, KDD ‘09. New York: Association for Computing
32
De Keyzer, F., Dens, N., & De Pelsmacker, P. (2015). Is this for me? How Consumers
Respond to Personalized Advertising on Social Network Sites. Journal of Interactive
Advertising, 00. doi: 10.1080/15252019.2015.1082450
Dent, J. 2017, September 20). The Secret to Creating Personalized Content That’s Relevant
Not Creepy. Retrieved on 29 January 2017, from
https://www.campaignmonitor.com/blog/email-marketing/2017/09/secret-to-creating-personalized-content-relevant-not-creepy/
Dias, M. B., Locher, D., Li, M., El-Deredy, W., & Lisboa, P. J. G. (2008), The Value of Personalised Recommender Systems to E-Business: A Case Study. Proceedings of
the 2008 ACM Conference on Recommender Systems, RecSys ’08. New York:
Association for Computing Machinery, 291–94. doi: 10.1145/1454008.1454054 Doorn, J. & Hoekstra, J. C. (2013). Customization of online advertising: The role of
intrusiveness. Marketing Letters, 24(4), 339-351. doi: 10.1007/s11002-012-9222-1 Hauser, J. R., Glen L. U., Liberali, G., & Braun, M. (2009). Website Morphing. Marketing
Science, 28 (2), 202–223. doi: 10.1287/mksc.1080.0459
Komiak, S. Y.X., & Benbasat, I. (2006). The Effects of Personalization and Familiarity on Trust and Adoption of Recommendation Agents. Management Information Systems
Quarterly, 30 (4), 941–60. doi: 10.2307/25148760
Lambrecht, A. & Tucker, C. (2013). When Does Retargeting Work? Information Specificity in Online Advertising. Journal of Marketing Research, 50(5), 561-576. doi:
10.1509/jmr.11.0503
Lee, S., Kim, K. J., & Sundar, S. S. (2015). Customization in location-based advertising: Effects of tailoring source, locational congruity, and product involvement on ad attitudes. Computers in Human Behavior, 51, 336-343. doi:
33
Linden, G., Smith, B., & York, J. (2003). Amazon.com recommendations: Item-to-item collaborative filtering. Internet Computing, IEEE, 7(1), 76-80. doi:
10.1109/MIC.2003.1167344
Mackenzie, S., & Lutz, R. (1989). An Empirical Examination of the Structural Antecedents of Attitude toward the Ad in an Advertising Pretesting Context. Journal of
Marketing, 53(2), 48-65. doi: 10.2307/1251413
MacKenzie, S., Lutz, R. J, & Belch, G. E. (1986). The role of attitude toward the ad as a mediator of advertising effectiveness a test of competing explanations. Journal of
Marketing Research, 23(2), 130-143. doi: 10.2307/3151660
Malthouse, E. C., & Li, H. (2017). Opportunities for and Pitfalls of Using Big Data in Advertising Research. Journal of Advertising, 46(2), 227–235. doi:
10.1080/00913367.2017.1299653
Maslowska, E., Smit, E., & Van Den Putte, B. (2013). Assessing the cross-cultural
applicability of tailored advertising. International Journal of Advertising, 32(4), 487-511. doi: 10.2501/IJA-32-4-487-511
Maslowska, E., Smit, E., & Van Den Putte, B. (2016). It Is All in the Name: A Study of Consumers' Responses to Personalized Communication. Journal of Interactive
Advertising, 16(1), 74-85. doi: 10.1080/15252019.2016.1161568
Petty, R. E., & Cacioppo, J. T. (1986). The elaboration likelihood model of
persuasion. Advances in experimental social psychology, 19, 123-205. doi: 10.1016/S0065-2601(08)60214-2
Preacher, K. J., & Hayes, A. F. (2008). Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behavior Research
34
Spears, N., & Singh, S. (2004). Measuring Attitude toward the Brand and Purchase
Intentions. Journal of Current Issues & Research in Advertising, 26(2), 53-66. doi: 10.1080/10641734.2004.10505164
Tam, K. Y., & Ho, S. Y. (2006). Understanding the impact of web personalization on user information processing and decision outcomes. MIS Quarterly,30(4), 865-890. doi: 10.2307/25148757
Trope, Y., & Liberman, N. (2010). Construal-Level Theory of Psychological Distance. Psychological Review, 117(2), 440-463. doi: 10.1037/a0018963
Trusov, M., Ma, L., & Jamal, Z. (2016). Crumbs of the Cookie: User Profiling in Customer-Base Analysis and Behavioral Targeting. Marketing Science, 35(3), 405–426. doi: 10.1287/mksc.2015.0956
Tsang, M., Ho, S., & Liang, T. (2004). Consumer Attitudes Toward Mobile Advertising: An Empirical Study. International Journal of Electronic Commerce, 8(3), 65-78. doi: 10.1080/10864415.2004.11044301
Wedel, M., & Kannan, P. K. (2016). Marketing Analytics for Data-Rich Environments.
Journal of Marketing, 80(6), 97–121. doi: 10.1509/jm.15.0413
Yan, J., Liu, N., Wang, G., Zhang, W., Jiang, Y. & Chen, Z. (2009). How Much Can Behavioral Targeting Help Online Advertising? Association for Computing
Machinery, 261-270. doi: 10.1145/1526709.1526745
Zaichkowsky, J. L. (1994). The Personal Involvement Inventory: Reduction, Revision, and Application to Advertising. Journal of Advertising, 23(4), 59-70. doi:
35
Appendix I
Questionnaire (4 conditions)
Factsheet Dear participant,
I would like to invite you to participate in a research study to be conducted under the auspices of the Graduate School of Communication, a part of the University of Amsterdam. The title of the study for which I am requesting your cooperation is ‘Personalized Online Advertising and Online Shopping Behavior’. In the online survey, one scenario and advertisements will be presented. Each time, you will be asked to answer few questions regarding given scenario and advertisements.
As this research is being carried out under the responsibility of the ASCoR, University of Amsterdam, we can guarantee that:
1) Your anonymity will be safeguarded, and that your personal information will not be passed on to third parties under any conditions, unless you first give your express permission for this.
2) You can refuse to participate in the research or cut short your participation without having to give a reason for doing so. You also have up to 24 hours after participating to withdraw your permission to allow your answers or data to be used in the research.
3) Participating in the research will not entail your being subjected to any appreciable risk or discomfort, the researchers will not deliberately mislead you, and you will not be exposed to any explicitly offensive material.
4) No later than five months after the conclusion of the research, we will be able to provide you with a research report that explains the general results of the research.
For more information about the research and the invitation to participate, you are welcome to contact the project leader Martyna Gorączka at any time. Should you have any complaints or comments about the course of the research and the procedures it involves as a
consequence of your participation in this research, you can contact the designated member of the Ethics Committee representing ASCoR, at the following address: ASCoR Secretariat, Ethics Committee, University of Amsterdam, Postbus 15793, 1001 NG Amsterdam; 020‐525 3680; ascor‐secr‐fmg@uva.nl. Any complaints or comments will be treated in the strictest confidence. We hope that we have provided you with sufficient information. We would like to take this opportunity to thank you in advance for your assistance with this research, which we greatly appreciate.
Kind regards, Martyna
Q1 I hereby declare that I have been informed in a clear manner about the nature and method of the research, as described in the invitation for this study.
I agree, fully and voluntarily, to participate in this research study. With this, I retain the right to withdraw my consent, without having to give a reason for doing so. I am aware that I may halt my participation in the experiment at any time.
If my research results are used in scientific publications or are made public in another way, this will be done such a way that my anonymity is completely safeguarded. My personal data will not be passed on to third parties without my express permission.
If I wish to receive more information about the research, either now or in future, I can contact Martyna Gorączka (magoraczka@gmail.com). Should I have any complaints about this
36
research, I can contact the designated member of the Ethics Committee representing the ASCoR, at the following address: ASCoR secretariat, Ethics Committee, University of Amsterdam, Postbus 15793, 1001 NG Amsterdam; 020‐525 3680; ascor‐secr‐fmg@uva.nl.
o
I understand the text presented above, and I agree to participate in the research study (1)Q156 Now, please read carefully given scenario and answer honestly for a number of questions.
High level of personalization + the purchase decision stage
Q13 Imagine that you would like to buy headphones and extra extension cords to that product. You were browsing the Internet, and decided to choose brand Sweetsound and specific model of the headphones and extension cords. You put both products to your ‘shopping bag’ (‘basket’) on the brand’s website. You can see them below (after 5 sec. you can skip to the next question).
Q157
Q15 In the meantime, you are also browsing the Internet to read some news platforms and you see Sweetsound’s ad next to the news articles. The situation is presented below (after 5 sec. you can skip to the next page).
37 Q97
High level of personalization + the post-purchase stage
Q33 Imagine that you recently have bought online headphones and extra extension cords of the brand Sweetsound on the brand website. You can see them below (after 5 sec. you can skip to the next question).
Q89
Q35 Further, you are browsing the Internet to read some news platforms and you see
Sweetsound’s ad next to the news articles. The situation is presented below (after 5 sec. you can skip to the next question).
38 Q98
Low level of personalization + the purchase decision stage
Q49 Imagine that you would like to buy headphones and extra extension cords to that product. You were browsing the Internet, and decided to choose brand Sweetsound and specific model of the headphones and extension cords. You put both products to your ‘shopping bag’ (‘basket’) on the brand’s website. You can see them below (after 5 sec. you can skip to the next question).
Q91
Q51 In the meantime, you are also browsing the Internet to read some news platforms and you see Sweetsound’s ad next to the news articles. The situation is presented below (after 5 sec. you can skip to the next page).
39 Q171
Low level of personalization + the post-purchase stage
Q65 Imagine, that you recently have bought headphones and extra extension cords of the brand Sweetsound, on the brand website. You can see them below (after 5 sec. you can skip to the next question).
Q93
Q67 Further, you are browsing the Internet to read some news platforms and you see
Sweetsound’s ad next to the news articles. The situation is presented below (after 5 sec. you can skip to the next page).
40 Q172
Questionnaire
Q161 Remember that you are not considering any other brands or products, as you already decided which headphones and extension cords of Sweetsound brand you want to buy.
Q159 Remember that you already have bought headphones and extension cords of Sweetsound brand.
Ad relevance
To what extent do you agree or disagree with the following statements? For me, this ad is...
Strongly disagree (1) Disagree (2) Somewhat disagree (3) Neither agree nor disagree (4) Somewhat agree (5) Agree (6) Strongly agree (7) Important (3)
o
o
o
o
o
o
o
Relevant (1)o
o
o
o
o
o
o
Valuable (8)o
o
o
o
o
o
o
Exciting (9)o
o
o
o
o
o
o
Interesting (4)o
o
o
o
o
o
o
Appealing (5)o
o
o
o
o
o
o
41 Fascinating (10)
o
o
o
o
o
o
o
Needed (11)o
o
o
o
o
o
o
Involving (6)o
o
o
o
o
o
o
Means a lot to me (7)o
o
o
o
o
o
o
Q132 Remember that you have previously browsed the company’s website and later the company showed you this ad:
42 Q176
Ad attitude
How would you rate the ad above? This ad is...
1 (0) 2 (1) (2) 4 (3) (4) 6 (5) Unlikable (1)
o
o
o
o
o
o
o
Likable Unpleasant (2)o
o
o
o
o
o
o
Pleasant Negative (3)o
o
o
o
o
o
o
Positive Interesting (4)o
o
o
o
o
o
o
Boring Bad (5)o
o
o
o
o
o
o
Good Tasteless (6)o
o
o
o
o
o
o
TastefulQ139 Now, I would like to know what you think about the brand which shows you the ad below.
