DYNAMIC RETARGETING –
TO WHAT EXTEND DOES IT
AFFECT CONSUMERS’
PURCHASE INTENTION?
By Krisjan Oldekamp
2
DYNAMIC RETARGETING
TO WHAT EXTEND DOES IT AFFECT
CONSUMERS’ PURCHASE INTENTION?
Master Thesis / November 2012
University of Groningen
Faculty of Economics and Business
Msc Business Administration – Marketing Management
Author Krisjan Oldekamp Eerste Hunzestraat 21 9715 BJ Groningen (06) 108 12 592 krisjan@gmail.com Student number 1689452 Supervisors
3
ABSTRACT
A specific form of behavioral targeting is getting more and more popular and is known as dynamic retargeting. With this technique, companies are enabled to serve real-time personalized recommendations in a display banner advertisement outside of the company’s website. The recommendations represent the individual browsing history of consumers, as in items seen or searched for, gathered on the website of the advertising company. The main goal: seduce the consumer into finishing their transaction and convert the consumer into a buyer. This report tries to explain to what extend dynamically retargeted advertisements affect consumers’ purchase intention. Main predictors are the number of items in the banner and the presence of a discount. The number of items in a banner stimulates the ad’s informativeness and therefore relevance to the consumer and is in general perceived as useful. Discounts are commonly used and a powerful tool to manipulate the consumer and eventually finishing their transaction. The data from a field experiment among 154 respondents was used to test eight hypotheses. The field experiment incorporated a 3 (2, 4 or 6 items in the banner) x 2 (showing discount or not) between subjects factorial design. Further, privacy concerns and the intrusiveness of the ad were measured. Privacy concerns are a much discussed topic nowadays and influences the effectiveness of targeted banner advertisements. Due to the clutter of online banner advertisements, intrusiveness is also an important predictor in banner advertisement effectiveness. Number of products bought online and product involvement were used as control variables.
Main findings are that the number of items in a dynamically retargeted banner, discount, privacy concerns and intrusiveness did not directly influence purchase intention. In contrast with the results from previous research in the field of behavioral targeting, none of the hypotheses could be supported. On a 10% significance level, showing discounts increases purchase intention when 6 products are shown in the banner. Also privacy concerns are increasing purchase intention when 4 products are displayed.
4
discount and purchase intention and a moderating effect of discount between number of items and purchase intention could not be supported.
Most likely, the setup of the questionnaire could not interest the respondents enough, of which the greatest part consisted of students. The results also indicate that there are other aspects of a dynamically retargeted banner that predict purchase intention and therefore, future research is necessary.
5
PREFACE
This report is the final piece in completing my master Marketing Management at the University of Groningen. Although I did not found any ‘hard evidence’, I learned a lot about many aspects of online (and offline) marketing and enjoyed working on this document. I look forward to find an inspiring job where I can bring my knowledge into practice, but above all, learn a lot more.
Firstly, I would like to thank my first supervisor Prof. Dr. Janny Hoekstra for her dedicated time, useful and constructive feedback, support and pleasant collaboration during the process of writing my thesis. Secondly I would like to thank my second supervisor dr. Marielle Non for her useful feedback.
A final thanks goes out to my family, friends, classmates and colleagues who helped me collecting data and supporting me in the process of writing my thesis, but especially supporting me in having a great time in Groningen.
Krisjan Oldekamp
6
TABLE OF CONTENTS
1. INTRODUCTION………. 7
1.1 Background……… 7
1.2 Research question……….. 13
1.3 Academic & managerial relevance……… 14
1.4 Structure………. 14
2. THEORETICAL FRAMEWORK……….. 15
2.1 Conceptual model……….. 15
2.2 Purchase intention as a measure of advertisement effectiveness…….. 16
2.3 Number of items in banner……… 17
7 4. RESULTS……….. 33 4.1 Descriptives………33 4.2 Regression results……….. 35 4.3 Multicollinearity diagnostics……… 35 4.4 Direct effects ………. 35 4.5 Moderation effects………. 35 4.6 Model evaluation………... 36 4.7 Segments……… 36 4.8 Hypotheses overview……… 37 5. DISCUSSION……… 38 5.1 Conclusion………. 38
5.2 Theoretical & managerial implications………. 40
5.3 Limitations……… 40
6. REFERENCES……….… 42
8
1. INTRODUCTION
1.1 Background
In the last two decades the online advertising industry grew enormously. Since the first commercial web magazine HotWired began selling banner advertisements to advertisers in 1994 (Doubleclick, 2005), online advertising revenues in the US rose from $267 million in 1996 (IAB, 1996) to $31.7 billion in 2011 (IAB, 2011). In 2011, display banner advertisements, which are paid banner advertisements placed on websites available through internet access, account for 23% of online advertisement revenues, second after search advertising which holds for 45% (IAB, 2011). All this spending gives evidence for a huge increase in the amount online advertisements. Logically, this creates new problems. Click-through rates declined from 7% in 1996 to 0,5% in 2003 (Doubleclick, 2003) and consumers avoid or paying less attention looking at banner ads during their online activities, known as ‘banner blindness’ (Drèze and Hussherr, 2003; Cho & Cheon, 2004). One of the main reasons behind banner blindness is that on the internet, consumers are goal directed (Danaher & Mullarkey, 2003; Cho & Cheon, 2004) and advertisements interrupt this process, evoking negative feelings and irritation towards ads which lead to ad avoidance (Baek & Morimoto, 2012).
Phenomena such as banner blindness are forcing the advertising industry to continuously find new ways to rise above the clutter and improve advertisement effectivity. Targeting banner advertisements is one of these ways. Targeting banner advertisements increases the relevance of the advertisements for the consumer, because it is matched with the content of a website or behavior of the consumer. Consumers show a relatively high tolerance towards targeted ads, which decreases ad avoidance because the information is perceived as useful (Cho & Cheon, 2004; Edwards et al., 2002; Komiak & Benbasat, 2006; Baek & Morimoto, 2012). Therefore, personalized display banner ads are one of the key trends in the online marketing field (Yann et al., 2010).
9
targeting (Goldfarb & Tucker, 2011a). The match between content and advertisements improved when click logs, demographic and geographic data were made available to advertisers to pinpoint their advertisements more accurately to certain types of content.
A moreconsumer specific method of targeting emerged when companies started to base advertisements on past behavior, called behavioral targeting. Behavioral targeting uses historical user behavior to select the most relevant ads to display. (Chan, Pavlov and Canny 2009, p. 209). This user behavior is gathered by storing so called ‘cookies’ on the consumer’s computer (Alreck & Settle, 2007). Cookies enable advertising networks, such as Google Adsense, to track every consumer on their network, but also when coming and going. This data is referred to as external browsing data (Lambrecht & Tucker, 2011). Publishers can join these advertisement networks to display advertisements on their websites. With the data that the networks provide, advertisers can create segments for their ads. For example, an advertisement for shoes that is targeted to consumers that visited shoe related websites in the past. Besides creating segments, advertisers can specify time periods and number of views for their banners. Using more of these networks, advertisers can extend the reach of their advertisements. Generally, advertisers’ pay per click (CPC) or per number of impressions (CPM), of which the publisher receives a large and the network receives a small percentage.
10
from the advertising networks to track the specific consumers across the web and display the banner. Figure 1.1 visually represents dynamic retargeting.
Figure 1.1
Visual representation of dynamic retargeting
The difference with existing techniques is that ‘dynamic retargeting’, does not fit banner ads to the content of a website, nor to consumer segmentsbased on browsing history. It improves advertisements on of external websites by creating personalized, consumer specific banners. Figure 1.2 shows two examples of dynamically retargeted banners.
Figure 1.2
Examples of retargeted banners
11
In short, dynamic retargeting extends the consumer-specific communication of a firm, because it is no longer limited to consumers who decide to return to the company’s website (Lambrecht & Tucker, 2011). It ‘retargets’ the consumer to the products he or she viewed earlier. The main goal: seduce the consumer into finishing their transaction and convert the consumer into a buyer. Table 1.1 summarizes the banner advertisement methods that are discussed.
Table 1.1
Summary of relevant online banner advertisement methods.
Method Type Example
Contextual targeting Ads are targeted to the content of a website
Advertisement of a shoe firm is shown on a website that writes about shoes but does not sell shoes.
Behavioral targeting Ads are targeted to consumer segments based on past user behavior
Shoe firm advertises on websites in the advertisement network to consumers who visited a shoe related website in the past, for example a fashion blog. Dynamic retargeting Ads are targeted to
specific consumers that visited a specific website
Shoe firm advertises on all websites in the advertisement network to only the consumers who previously visited the same shoe firm’s website. The
advertisement reflects behavior of the individual consumer.
12
concerns, it has large influence on the effectiveness of behaviorally targeted ads. Alreck & Settle (2007), Wathieu & Friedman (2009), Goldfarb & Tucker (2011b) and Baek & Morimoto (2012) all conclude that privacy concerns negatively influence the effectiveness of targeted ads because when privacy concerns are present, a prevention focus may increase sensitivity to manipulative intent of targeted ads (Kirmani & Zhu, 2007). The perceived level of privacy on-site has a positive effect on purchase intention (Ranganathan & Ganapathy, 2002), however when confronted with personal browsing history outside the company’s website, the level of perceived privacy is violated and negatively influences purchase intention. Besides, when consumers have the feeling that they are chased by, for example, shoes on every website they visit, it could interfere with cognitive processing and increase the level of perceived intrusiveness. Intrusiveness predicts irritation, and irritation predicts behavioral attitudes such as ignoring the ad (McCoy et al., 2008).
Most studies regarding to privacy concerns and intrusiveness are focusing on contextually targeted ads or general attitude towards these ads. Little research on dynamic retargeting has been done but the results of previous studies indicate that a greater specificity between consumers and advertisements makes sense; informativeness and personal relevance is perceived as useful by consumers (Komiak & Benbasat, 2006; Dias et al., 2008; Yann et al, 2010; Goldfarb & Tucker, 2011a; Lambrecht & Tucker, 2011). The number of items in a banner represents informativeness, and thus increases the relevance for consumers, especially when targeted to the consumer (Thongpapanl & Ashraf, 2011). Leigh (1984) also concludes that there is a positive linear effect in the number of items in print advertisements and the attitude towards the ad. Discounts are commonly used and a powerful tool to manipulate the consumer and eventually finishing their transaction (Compeau & Grewal, 1998), especially when the advertised items reflect their interests.
13
increased ad specificity (dynamic retargeting) is less effective (in terms of purchasing a product) than generic information when retargeting ads, due to a mismatch between the specificity of a dynamic retargeted ad and whether a consumer has well defined product preferences. Generic ads may be more effective than specific ads; it does not indicate that specific ads are ineffective. Dynamic retargeting can avoid bombarding consumers with irrelevant messages, which led to phenomena such as ‘banner blindness’ and it enables the marketer to send messages that are most likely to generate purchases (Pavlou & Stewart, 2000). Besides, mass advertising is usually costly and unnecessary for many products that do not apply to all people (Pavlou & Stewart, 2000) and returns from targeted banner advertisements are the highest for the number of advertising exposures, compared to non-targeted banner advertisements (Manchanda, 2006). This report tries to contribute by conducting a field experiment to analyze how characteristics of this fairly new method of behavioral targeting and the relation with the highly debated issue of privacy concerns and intrusiveness influence purchase intention.
1.2 Research question
The following research question and sub questions are addressed;
To what extend do dynamically retargeted advertisement banners affect consumers purchase intention?
Sub questions:
1. To what extend do banner characteristics in terms of the number of items in the banner affect purchase intention?
2. To what extend do banner characteristics in terms of showing price discounts affect purchase intention?
3. To what extend do privacy concerns and the level of perceived intrusiveness affect purchase intention?
14
5. To what extend do discounts moderate the relation between the number of items in the banner and purchase intention?
1.3 Academic and managerial relevance
This report tries to extend the knowledge in the behavioral targeting field by researching dynamically retargeted banner advertisements. Lambrecht & Tucker (2011) made a start and this report continues where they left off. Relevant is that dynamically retargeted advertisements are based on individual behavior, and automatically personalized for every consumer. Therefore, the number of items in a banner makes sense, since consumers value the amount of informativeness, especially when it has personal relevance. Discounts are taken in to account since they may influence the behavior towards the ad and are a powerful tool to manipulate the consumer. Privacy concerns are a much discussed topic nowadays and influences the effectiveness of targeted banner advertisements, even as intrusiveness. All these factors are not studied before in combination with dynamically retargeted advertisements. And while other studies used measures such as brand awareness, banner recall or most commonly click-through rate, purchase intention is used as the dependent variable. Although click-through rate practically is behavior, it does not predict purchase behavior (Maddox, 2003) unlike purchase intention that does predict actual behavior (Morrison, 1979; Pavlou 2003). Purchase intention is closer to the final goal, selling a product, and therefore it has more managerial relevance since the outcomes can help to understand the functioning of dynamic retargeting and used to optimize the setup of the banners to increase sales.
1.4 Structure
15
2. THEORETICAL FRAMEWORK
2.1 Conceptual model
Several factors related to dynamic retargeting are identified in the literaturethat might influence purchase intention. The conceptual model (figure 2.1) visually represents the variables and their hypothesized relations.
Figure 2.1 Conceptual model
16
reactance (interfering with cognitive processing) when there is no justified fit between the offer in the message and the consumers’ personal characteristics (White et al., 2008). In the case of dynamic retargeting, there is an explicit fit; personalized recommendations based on browsing history. Furthermore, a higher level of privacy concerns negatively influences the relation between price discounts and purchase intention (Dinev & Hart, 2004) as well as the relation between number of items and purchase intention. The number of times that a respondent has bought a product on the internet in the last year and product involvement are used as control variables. The variables will be explained and discussed further on in this chapter.
2.2 Purchase intention as a measure for advertising effectiveness
Over the years, there has been much debate about measuring online advertisement effectiveness (Chandon et al., 2003). The internet is providing the technology to make online advertisements more measurable than traditional advertising such as television and printed ads. The percentage of the total number of ad exposures that induce a surfer to actually click on a banner, the click-through rate (Novak & Hoffman, 1997) is the most widely used measure of online advertising (Chandon et al., 2003). Pavlou & Stewart (2000) and Chandon et al. (2003) state that the click-through rate is the right measure if the only objective of the ad is to generate direct response. However, Wen & Maddox (2003) found that click-through rate is significant in predicting banner recall, but gathered no evidence that the click-through rate is affecting brand recall, attitude recall and more interesting seen the subject of this research, purchase intention.
The main objective of dynamic retargeting is goal conversion, e.g. buy a product and not creating a direct response in terms of only returning to the firm’s website. Click-through rate could be a step in between in seeing the ad and purchasing a retargeted product, but it only measures the direct, short-term effect (of returning) and is not predicting purchase intention (Maddox, 2003). Consumers could also return to the company’s website at a later point in time to finish their transaction. Therefore purchase intention is an interesting measure.
17
(2003) defines purchase intention as consumers’ intent to use a retailer's website to obtain information and then complete a transaction by purchasing a product or service. Purchase intention, along with the intention to return, are one of the most used behavioral intention measures online (Hausman & Siekpe, 2009). The direct effect (directly returning to the company’s website) and the indirect effect (return to the company’s website at a later point in time) are both captured in this measure. Despite actual purchases are not part of this research, Pavlou (2003) found that consumers who have positive purchase intentions are more likely to make an actual purchase, which is in line with more traditional studies that also found that purchase intention is a predictor of the actual purchase (Morrison, 1979). Altogether, purchase intention is used as a dependent variable.
2.3 Number of items in banner
Dynamically retargeted ads contain items which represent the products a consumer viewed earlier on a specific website. The number of items in the banner can differ. The items consist of a product image and some brief product information like product name, price and promotional incentives (discounts). Several universal sizes are used for online banner advertisements, so that the implementation of advertisements becomes easier for publishers and advertisers (IAB, 2012). Publishers can standardize their space for advertisements and advertisers do not have to adjust their banner size for all the different publishers. That means that also dynamic retargeted ads are restricted to predefined sizes. Only the number of items in a banner varies. This can have several reasons: the advertiser predefined the number of items, the size of the banner could limit the number of items and/or the available internal browsing data of the consumer determines the amount of items. The selection and number of products is very company specific, because every company is able to use a different algorithm based on the behavioral tracking data to select the items shown. It could be that when a consumer only viewed two products, two complementary products are added and in the end four products are displayed.
18
shown. Figure 2.2 shows two possible formats of dynamically retargeted banners. The first format includes two items, the second six items.
Figure 2.2
Formats of dynamically retargeted banners
When looking for the effects of product images in online banner advertisements, research is limited. Research on outdoor and print advertisements could give some clues. Leigh (1984) tested several umbrella advertisements. In these advertisements, the promoted products are linked by a common theme. He found that consumers evaluated an ad with eleven products more positively than with five or eight products, where spacing and format was kept the same. This might indicate that there is a linear effect in number of items. The author notes that it is an important research topic to determine when the number of items becomes dysfunctional, however most dynamically retargeted ads do not contain more than six items.
click-19
through rate (Chandon et al., 2003), but for example returning to website on a later moment in time to finish the transaction. Image presence therefore, could positively affect purchase intention.
When looking at the size of product images, which in this case is related to number of items, Park & Stoel (2005) concluded that image size in the product catalog of a web store did not significantly affect purchase intention. However, the amount of information presented on a firm’s website has a positive impact on purchase intention (Lohse & Bellman, 2000; Ranganathan & Ganapathy, 2002). Information in this regard can be seen as visual elements and product information. Especially content that is targeted to the consumer positively influences purchase intention (Thongpapanl & Ashraf, 2011). The number of items in a banner can be seen as amount of information and therefore could positively influence purchase intention. Although there is not much literature that specifically focuses on number of product items in a advertisement banner, it is expected that the number of items positively affects purchase intention.
H1: The number of items in a display banner positively influences purchase intention
2.4 Price discounts
20
lower than internal reference price, it could increase purchase intention of the specific product.
Chtourou et al. (2001) studied a database and found that mentioning promotions in contextually targeted advertisement banners decreased the click-through rate. They suggest that involved consumers rely more on their internal price reference and are less sensitive for the pricing argument. It this case, it is likely that consumers are more involved, since retargeted banners reflect their behavior. Baltas (2003) also found that promotional incentives, including price discounts, were insignificant in predicting click-through rate. Chtourou et al. (2001) and Baltas (2003) used click-click-through rate as a dependent variable and therefore the indirect effects (returning to the website on a later point in time) were not measured. Besides, none of the two studies used behaviorally targeted banners. It is expected that when consumers are confronted with a price discount for one or more products that they previously showed interest in, increases their purchase intention because the discount can be the decisive factor to continue with the purchase.
H2a: Displaying price discounts positively influences purchase intention
21
up by the difference between internal and external price reference and therefore purchase intention. Especially when consumers are repeatedly confronted with the discounted products in the banner (Grewal et al., 1998) as is the case for dynamic retargeting. Therefore a moderating effect of price discounts is expected.
H2b: Displaying price discounts negatively influences the relation between number of items and purchase intention
2.4 Privacy concerns
As stated in the introduction, the developments regarding online behavioral targeting heightened consumers' privacy concerns (Goldfarb & Tucker, 2011b). Several studies indicate that privacy concerns are of importance when predicting advertisement effectiveness such as the findings of Goldfarb & Tucker (2011a). They conducted a large-scale field experiment and found that contextually targeted ads and highly visible ads independently increase purchase intention. However, in combination they are ineffective. The reason they give for the ineffective combination between targeted ads and highly visible (obtrusive) ads are privacy concerns. In 1999, Sheehan & Hoy found that when privacy concerns increased, respondents of personalized e-mail campaigns were more likely to request removal form mailing lists. In the experiment of Goldfarb & Tucker (2011a), the negative effect became higher for consumers who guard their privacy more closely and for product categories that are likely to be more private, like healthcare and financial products. The explanation for the positive effect of targeted ads on purchase intention, are that consumers may be willing to tolerate contextually targeted ads more because they potentially provide information. However, when the ad is also obtrusive, it increases the perception of manipulation (Campbell, 1995). Kirmani & Zhu (2007) found a similar result. When privacy concerns are present, a prevention focus may increase sensitivity to manipulative intent of targeted ads. And in some cases, dynamically retargeted ads have the characteristics of an obtrusive ad such as pop-up, pop-under, take-over, floating, full page banner, audio or video elements (Goldfarb & Tucker, 2011a).
22
(2007) studied consumer reactions to online behavioral tracking and found that consumers are much more aware of this than other studies suggested at that time. Reactions to it are largely negative. Also Wathieu & Friedman (2009) are supporting this claim;customer appreciation of the informativeness of targeted ads is tempered by privacy concerns. Baek & Morimoto (2012) found that privacy concerns directly positively influence ad avoidance of personalized advertisements. Personalized advertising could make consumers perceive that their privacy is threatened and evokes a resistance towards advertisements that involve storing personal data or tracking (Simonson, 2005). Since literature gives some strong clues about the negative effect of privacy concerns on different measures of advertisement effectiveness, the next hypothesis is stated;
H3a: Privacy concerns negatively influences purchase intention
On a company’s website, the perceived level of privacy by the consumer has a positive effect on purchase intention (Ranganathan & Ganapathy, 2002). However, when consumers that value their privacy are confronted with their personal on-site browsing history outside of the company’s website, the level of perceived privacy on the firm’s website is violated and negatively influences purchase intent. The number of items in a banner represents their browsing history, and therefore increasing the number of items may insinuate more privacy intrusion. It is therefore expected that privacy concerns negatively moderate the relation between number of items and purchase intention.
H3b: Privacy concerns negatively influences the relation between the number of items in a display banner and purchase intention
23
price discounts could arouse this feeling. Therefore a moderating effect of privacy concerns is expected.
H3c: Privacy concerns negatively influences the relation between price discounts and purchase intention
2.6 Intrusiveness
Li et al. (2002) define intrusiveness as a ‘psychological reaction to ads that interfere with a consumer's ongoing cognitive processes’. They state that the definition is context free; it applies regardless of when or where the interference takes place, for example viewing a dynamically retargeted ad when reading a news article online. As long as the ad interferes with cognitive processing, the perception of the ad by consumers as intrusive is possible. Since dynamic retargeting makes use of external browsing data to track down the consumer online (Lambrecht & Tucker, 2011), the banners can appear on almost every website they visit. Consumers are goal directed on the internet (Danaher & Mullarkey, 2003; Cho & Cheon, 2004) and when ads interrupt a consumer’s goal, it may result in negative attitudes and ad avoidance (Baek & Morimoto, 2012). The perceived intrusiveness of dynamically retargeted ads therefore could have consequences for the ad’s effectiveness.
When consumers have the feeling that they are chased by, for example, shoes on every website they visit, it could interfere with cognitive processing. This can result in consumers developing negative attitudes towards the dynamically retargeted ads and could inflict a decrease in purchase intention (MacKenzie & Lutz, 1989). McCoy et al. (2008) underpin these results. They found that intrusiveness predicts irritation, and irritation predicts behavioral attitudes such as ignoring the ad. Therefore it is expected that intrusiveness has a negative effect on purchase intention.
H4a: Intrusiveness negatively influences purchase intention
24
perceived level of privacy could distract the consumer from the content of a website. Besides, when confronted with more items that represent personal browsing history and thus violation of perceived privacy, the perceived invasiveness or distraction from the content could increase. In other words, it could interfere with the consumers’ ongoing cognitive process (Li et al., 2002) and makes the ad more intrusive in nature. Therefore it is expected that intrusiveness has a moderating influence between the relation of number of items and purchase intention.
25
3. RESEARCH DESIGN
3.1 Research Method
A field experiment was conducted among 227 respondents through a 3 (2, 4 or 6 products) x 2 (showing discounts or not) between subjects factorial design. A convenience approach was used to collect responses. Family, friends, colleagues and fellow students were asked to fill out the questionnaire. Also snowballing was used to gather respondents. Every respondent was asked to forward the questionnaire to their network. The questionnaire can be found in Appendix I.
The procedure of the experiment was as follows. Firstly the respondents were asked to fill out several demographic variables namely gender, age, education and how many times the respondent bought a product on the internet last year. Secondly, the privacy concerns of the respondents were measured. Thirdly, the respondent had to imagine that he was looking for a specific product, in this case a backpack for holiday purposes. Backpacks were chosen, because of the gender neutral nature of the product. Six backpacks, of which the consumer had to imagine he/she had found these in a webshop (Perry Sport), were displayed. The store logo and products were highly visible, to stimulate recognition in next part of the survey. The introduction of this part of the questionnaire is shown below;
Imagine that you have planned a vacation in a couple of months. However, you still need a backpack for making small trips while you are on holiday. For example to carry your towel when heading to the beach or a coat when you go out walking. Besides, a backpack always comes in handy. After some browsing on the internet and making comparisons, you end up on the website of Perry Sport. Perry Sport has an online webshop, which you evaluated positively in the past and had some good experiences with. In this webshop you can order many products, including backpacks. In the webshop, you saw some backpacks that fitted your needs and were affordable. However, you were distracted and left the website of Perry Sport. You did not decide yet. Assume that there is enough time left to order the backpack. The backpacks that you viewed in the webshop of Perry Sport are shown below. {Six examples of backpacks}
26
The next part consisted of six different banner scenarios (2, 4 of 6 items x discount or no discount) and was randomly assigned to the respondents (see figure 3.1). The order of the products in the banners was kept the same as in the introduction of the products. The banner size used in each case was a medium sized rectangle (300 x 200 pixels), which is a size commonly used for dynamically retargeted banners and an industry standard (IAB, 2012).
The different banners were placed in a (faked) popular Dutch news website (NOS.nl). The date and time of the news article in the dummy website is automatically changed to the current date and one hour before the current time. A somewhat noticeable and timeless news article was used to create a natural situation. Everything else was kept the same for every respondent. Table 3.1 shows the scenarios. Figure 3.1 shows the different banners used in the scenarios. In Appendix I, the dummy website of NOS.nl can be found.
Table 3.1
Scenarios (2, 4 and 6 products x discount or no discount)
Scenario Number of items Discount Respondents Successfully
27
The respondent was asked to fill in the questions regarding the intrusiveness of the banner and intention to purchase. To check if the manipulation was successful, the respondent also had to fill in how many items he/she thought were displayed and if discount was included. The last page included questions about the involvement with backpacks (control variable). If a respondent did not fill out a question, a warning was given. This led to minimal missing values when a respondent completed a survey. 3.2 Pretest
28 Figure 3.1
Banners used in the six different scenarios
29 3.3 Measurement instruments
For measuring purchase intention, privacy concerns, intrusiveness and product involvement, existing scales are used to measure their value. In table 3.1 an overview is given of the different measurement scales.
Table 3.2 Measurement scales
Concept Items (all measured on a 7-point Likert scale) C. α
Purchase intention
Hausman & Siekpe (2009)
(1). I will definitely buy products from the website of Perry
Sport in the near future.
(2). I intend to purchase through the website of Perry Sport in the near future.
(3). It is likely that I will purchase through the website of Perry
Sport in the near future.
(4). I expect to purchase through the website of Perry Sport in the near future.
0,950
Privacy concerns
Chellappa & Sin (2005)
(1). I am sensitive about giving out information regarding my preferences.
(2). I am concerned about anonymous information (information collected automatically but cannot be used to identify me, such as, network information, browsing history, search history etc) that is collected about me.
(3). I am concerned about how my personally un-identifiable information (information that I have voluntarily given out but cannot be used to identify me, e.g., Zip Code, age-range, sex, etc.) will be used by the firm.
(4). I am concerned about how my personally identifiable information (information that I have voluntarily given out AND can be used to identify me as an individual, e.g., name, shipping address, credit card or bank account information, social security number, etc.) will be used by the firm
0,763
Intrusiveness
Li et al. (2002)
I think the ad is distracting I think the ad is disturbing I think the ad is forced I think the ad is interfering I think the ad is invasive I think the ad is obtrusive
0,884
Product involvement
Laurent & Kapferer (1985)
I attach great importance to backpacks
Backpacks interest me a lot.
Backpacks leave me totally indifferent.
30
All the items were measured on a 7-point Likert scale. The four item purchase intention scale of Hausman & Siekpe (2009) is developed for online use. The scale was slightly modified in terms of adding the webshop name in the questions. The 4 item privacy concerns scale of Chellappa & Sin (2005) is chosen, because it measures the concerns of anonymously gathered and voluntarily provided data. Intrusiveness is measured by a broadly used scale of Li et al (2002). Product involvement is measured by a three item scale developed by Laurent & Kapferer (1985). The direction of the last item in this scale was recoded. The Cronbach’s Alpha is used to determine the reliability of the concepts. When Cronbach’s alpha is higher than 0,6 (Maholtra, 2010), the scale is internal consistent. In the last column table 3.2 can be found that all Cronbach’s Alpha’s are higher than 0,6 and therefore reliable.
3.4 Control variables
Product involvement and number of times that the consumer bought a product online were used as control variables.
3.5 Manipulation
To check if the manipulation in number of items and discounts were successful, a manipulation check was included in the questionnaire. The respondents were asked to fill out the number of products in the banner he/she had seen and if discounts were included. The option to visit the previous page was disabled.
Cross tabulation (table 3.3) reveals that 90,7% of the respondents remembered the number of products correctly. For the manipulation of discount, a cross tabulation reveals that 87,7% of the respondents that were shown a banner that did not include discounts, correctly remembered that the banner did not show a discount. Of the respondents that were shown a banner with a discount, 60,2% correctly remembered that the banner was including a discount. In total, 74% of the respondents correctly remembered discount.
31 Table 3.3
Cross tabulation manipulation number of products & discount
Products Actual # Respondents correct Percentage correct
2 77 70 90,9%
4 67 63 94%
6 83 73 88%
227 206 90,7%
Discount Actual # Respondents correct Percentage correct
Yes 113 100 87,7%
No 114 68 60,2%
227 168 74%
3.6 Plan of analysis
A multiple linear regression analysis was executed with purchase intention as dependent variable. First of all, a multicollinearity check was conducted by analyzing the Variance Inflations Factors (VIFs). When entering the moderators into the model,
32 Table 3.4 Regression models Model (1) PI = α + b1NI1 + b2NI2 + b3DI + ε (2) PI = α + b1NI1 + b2NI2 + b3DI + b4INV + b5OP + ε
(3) PI = α + b1NI1 + b2NI2 + b3DI + b4INV + b5OP + b8DI*NI1 + b9DI*NI2+ ε
(4) PI = α + b1NI1 + b2NI2 + b3DI + b4INV + b5OP + b6INT + b13INT*NI1 + b14INT*NI2 + ε
(5) PI = α + b1NI1 + b2NI2 + b3DI + b4INV + b5OP + b7PC + b10PC*NI1 + b11PC*NI2 + b12PC*DI + ε
(6) PI = α + b1NI1 + b2NI2 + b3DI + b4INV + b5OP + b6INT + b7PC + b8DI*NI1 + b9DI*NI2 +
b10PC*NI1 + b11PC*NI2 + b12PC*DI + b13INT*NI1 + b14INT*NI2+ ε
where:
PI = Purchase Intention α = Constant/Intercept
NI1 = Number of Items (1 = 4 items, 0 = no items)
NI2 = Number of Items (1 = 6 items, 0 = no items)
DI = Discount (1 = Discount, 0 = No discount) INV = Involvement
33
4. RESULTS
4.1 Descriptives
240 respondents participated in the research. After deleting incomplete answers and extreme outliers, 227 respondents remained. Out of these 227, the non-successfully manipulated respondents were deleted. This resulted in 154 respondents, as seen in table 4.1. 96 (62,3%) of the participants were male, 38 (37,7%) were female. The average age was 24,56 and most of the respondents finished HBO/HEAO (31,2%) or WO /
University (57,8%). On average, the respondent bought 10,03 products online last year. Table 4.1 Descriptives Variable Gender Male Female 96 58 62,3% 37,7% 154 100% Education Basischool LBO/MAVO/VMBO MBO/MEAO HAVO VWO HBO/HEAO WO/Universitair 0 1 10 1 5 48 89 0% 0,6% 6,5% 0,6% 3,2% 31,2% 57,8% 154 100%
Variable Mean Min Max Std. Dev
Age 24,56 16 56 6,47
Number of online purchases 10,03 0 80 10,40
Purchase intention 3,45 1 6 1,48
Involvement 3,60 1 6,33 1,38
Intrusiveness 3,89 1 6,33 1,28
34 Table 4.2
Multiple linear regression results (standardized coefficients) for Purchase Intention
1 2 3 4 5 6 Main effects 4 products -,089 -,095 -,033 -,110 -,107 -,043 6 products -,055 -,051 -,187 -,049 -,058 -,180 Discount ,013 ,023 -,062 ,023 ,033 -,035 Control variables Involvement ,020 ,048 ,018 ,028 ,060 Online purchases ,075 ,056 ,059 ,104 ,062
Direct effect moderators
Intrusiveness -,094 -,084 Privacy concerns -,156 -,165 Moderators Discount * 4 products -,088 -,119 Discount * 6 products ,247 c ,232 Intrusiveness * 4 products ,060 -,097 Intrusiveness * 6 products ,067 ,058
Privacy concerns * 4 products ,205c ,220c
Privacy concerns * 6 products ,083 ,063
Privacy concerns * Price discount ,000 ,045
R2 ,006 ,012 ,050 ,026 ,032 ,090
Adjusted R2 -,014 -,021 ,005 -,027 -,029 -,002
R2 change (model 2) ,026 -,032 -,002 ,027
F-value ,319 ,361 1,104 ,489 ,521 ,980
Note: a p-value < .01; b p-value < .05; cp-value < .10
35 4.2 Regression results
Six multiple linear regression models were estimated, the results are shown in table 4.2. The results will be discussed in the next subsection.
4.3 Multicollinearity diagnostics
A multicollinearity check was done by analyzing Variance Inflation Factors (VIFs). A VIF higher than 10 indicates highest multicollinearity (Leeflang, 2010; Hair et al, 2006). When adding the moderators to the model, the VIFs increased above 10. As a solution, the independent variables were mean centered (Aiken & West, 1991). After centering, the VIFs had acceptable values between 1,023 and 4,296 in models 3 to 6. 4.4 Direct effects
4 and 6 products did not significantly influence purchase intention, compared to 2 items. Both 4 and 6 products show a negative direction compared to 2 products. Also discount (H2a), intrusiveness (H4a) and privacy concerns (H3a) turned out to be insignificant. Intrusiveness shows a negative relation towards purchase intention. The control variables involvement and number of online purchase are also insignificant, but both indicate a positive relation towards purchase intention.
4.5 Moderating effects
None of the moderating effects were significant on the 5% significance level. On the 10% significance level however, in model 3 discount * 6 products (B = ,205 / p = ,97) and in model 5 (B = ,247 / p = ,079) and 6 (B = ,220 / p = ,077) privacy concerns * 4 products became significant. Discount is positively influencing purchase intention when 6 products are shown, and privacy concerns are positively influencing purchase intention when 4 products are shown.
Adding discount as a moderator between number of products, the explanatory power of the model increased with 2,6% compared to the model with only the main and control variables. Al the other additions to the model decreased the explanatory power.
36
showing a negative direction between the relation of number of items and purchase intention (H3a)
4.6 Model evaluation
The F-value’s indicate that all six models are highly insignificant (p > 0,10), there could not be fitted a regression line. The explanatory power of the models is low. The r-square values range from 0,6% (model 1) to 9% (model 6). With the exception of model 3 (0,5%), all the adjusted r-square values are negative. As mentioned in the footnote of table 4.2, the models were also estimated with the data of all the respondents (successfully and non-successfully manipulated respondents), which resulted in slightly worse outcomes. Therefore the models based on the successfully manipulated respondents were chosen. Taking the number of products as a continuous variable did also result in insignificance. Additionally, when assuming that the number of products directly influences intrusiveness in a separate model, both 4 products (B = -,058 / p = ,529) and 6 products (B = ,049 / p = ,599) did not significantly influence intrusiveness. 4.7 Segments
To determine if there are segments regarding involvement in the data, two segments were created by performing a median split; high and low involvement groups. Values above the median are classified as high involvement (n = 74) and values lower than the median as low involvement (n = 80). A Mann-Whitney U test, since not all variables are normally distributed, indicates that the means of purchase intent (p = ,424),
37 4.9 Hypothesis overview
Table 4.3 shows the formulated hypotheses and if support was found. Table 4.3
Overview of hypotheses
Supported Hypotheses
H1 No The number of items in a display banner positively influences purchase intention
H2a No Displaying price discounts positively influences purchase intention H2b No Displaying price discounts negatively influences the relation between
number of items and purchase intention
H3a No Privacy concerns negatively influences purchase intention
H3b No Privacy concerns negatively influences the relation between the number of items in a display banner and purchase intention
H3c No Privacy concerns negatively influences the relation between price discounts and purchase intention
H4a No Intrusiveness negatively influences purchase intention
38
5. DISCUSSION
5.1 Conclusion
The aim of this report is to find out to what extend dynamically retargeted advertisement banners affect consumers purchase intention. Unfortunately, given the results, none of the hypotheses could be supported because none of the models turned out to be significant. As a result, also none of the relations turned out to be significant on a 5% significance level.
With 10% significance, two moderating effects became significant. When 6 products are displayed, discounts increase purchase intention and when 4 products are shown in the banner, privacy concerns increase purchase intention. Both effects show a positive direction. However, it was hypothesized (H2b) that displaying price discounts negatively influenced the relation between number of items and purchase intention. Also the second effect is strange and inconsistent with hypothesis H3b, since literature suggested that a higher level of privacy concerns decreased purchase intention (Ranganathan & Ganapathy, 2002). Therefore, both hypotheses must be rejected. Since the models are insignificant and the two relations became significant at the 10% percent level, it provides not enough statistical power to generalize these findings. Besides, the moderating effects are compared to the effect of 2 products, and should therefore also become significant when 6 products are considered or show the same direction, which is not the case. Unless the relation is nonlinear.
Especially for privacy concerns and intrusiveness, literature gave some strong clues about the relation of these variables towards purchase intention, and therefore noteworthy that this cannot be confirmed.
39
errors and outliers. However, after examining and preparing the data, these possible explanations cannot be the cause.
Although only the respondents that were correctly manipulated were taken into account, it is rather strange that only 60,3% of the respondents could recall correctly that they did not saw a discount, against 87,7% that correctly could recall they did saw a discount. The pretest however, did not indicate this behavior. Therefore it can be concluded that the manipulation of discount, for some reason, was not strong enough. As a result, the sample size decreased with 32% due to deletion of the non-successfully manipulated respondents.
Comparing the models estimated with the data of all the respondents and the models estimated with data of the successfully manipulated respondents resulted in a minimal difference. This could indicate that, even when successfully manipulated, the banner advertisements could not interest the majority of the respondents, which were mostly students, apart from the involvement with backpacks. This can be proven by the fact that there is no significant difference in purchase intent, intrusiveness and privacy concerns between low and high involvement segments.
As the negative adjusted r-square values indicate, the chosen models are ‘worse’ than a horizontal line. In other words, the models fit the data really poorly. As mentioned earlier, literature gave some strong clues about the expected relations so one of the possible solutions is more data. Due to the deletion of unsuccessfully manipulated respondents, which decreased the sample size with 32%, more respondents (equally divided over the 6 different groups) would be a logical choice, since it will increase the statistical power (Maholtra, 2010, p. 374). Probably the effects are so small, that a larger sample size is needed.
40
using a product that could interest more respondents. Literature and/or conducting prior research on this topic could determine which product is more appealing to respondents. Adjusting the model could also serve as a solution. The influence of the much debated privacy policies, which affect dynamic retargeting, could be included. Since the banner characteristics tested in this research turned out insignificant, other banner characteristics of dynamically retargeted ads, such as animations or aspects as trust, product types, brand familiarity or number of times seen could be explored and included in the model.
5.2 Theoretical & managerial implications
The influence of dynamically retargeted banner characteristics in combination with consumers’ attitudes towards the banners on purchase intention was not explored before. However the banner characteristics, number of products and showing discounts or not, did not predict purchase intention. Therefore it can be concluded that there are other banner characteristics of dynamically retargeted banners such as animations, product types, brand familiarity or trust that might influence purchase intention. Or the research has to be set up differently. Most remarkable are the outcomes of privacy concerns and intrusiveness. Previous research found strong proof of privacy concerns and intrusiveness negatively influencing purchase intention (Goldfarb & Tucker, 2011a; Goldfarb & Tucker, 2011b; Cho & Cheon, 2004). This research could not confirm these findings, but did not provide the statistical power to disprove the results.
5.3 Limitations
Firstly, the bias towards young, male and highly educated students which can be related to the convenience approach of gathering the sample is a limitation. Besides the sample was smaller than originally intended, due to the deletion of the unsuccessfully manipulated respondents.
41
niche product, and therefore the results of this report cannot be generalized for the whole online retail industry.
42
6. REFERENCES
Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting interactions. Newbury Park, CA: Sage Publications.
Alba, J. & Lynch, J. (1997). Interactive Home Shopping: Consumer, Retailer, and Manufacturer Incentives to Participate in Electronic Marketplaces. Journal Of Marketing, 61(3), 38-53.
Alreck, P. L. & Settle, R. B. (2007). Consumer reactions to online behavioral tracking and targeting. Journal Of Database Marketing & Customer Strategy Management, 15(1), 11-23.
Baek, T., & Morimoto, M. (2012). Stay Away From Me. Journal Of Advertising, 41(1), 59-76.
Baltas, G. (2003). Determinants of internet advertising effectiveness: an empirical study. International Journal Of Market Research, 45(4), 505-513.
Biswas, A., & Blair, E.A. (1991). Contextual effects of reference prices in retail advertisements. Journal Of Marketing, 55(3), 1-12.
Brown, M., Pope, N., Voges, K., (2003). “Buying or browsing?: An exploration of shopping orientations and online purchase intention”, European Journal of Marketing, Vol. 37 Issue 11/12, p1666-1684
Campbell, M. C. (1995). When attention-getting tactics elicit consumer inferences of manipulative intent: The importance of balancing benefits and investments. Journal of Consumer Psychology. 4(3) 225–254.
Carroll, A., Barnes, S.J., Scornavacca, E. & Fletcher, K. (2007) Consumer perceptions and attitudes toward SMS advertising: recent evidence from New Zealand. International Journal of Advertising, 26(1), 79–98.
43
Chellappa, R. K., Sin, R. (2004). Personalization versus Privacy: An Empirical Examination of the Online Consumer's Dilemma. Information Technology and Management, 6(2/3), 181-202.
Cho, C.-H. & H. J. Cheon. (2004). Why do people avoid advertising on the internet? Journal of Advertising 33(4) 89–97.
Chtourou, M., Chandon, J., & Zollinger, M. (2001). Effect of Price Information and Promotion on Click-Through Rates for Internet Banners. Journal Of Euromarketing, 11(2), 23.
Compeau, L. D., & Grewal, D. (1998). Comparative Price Advertising: An Integrative Review. Journal Of Public Policy & Marketing, 17(2), 257-273.
Danaher, P. J., & Mullarkey, G. W. (2003). Factors Affecting Online Advertising Recall: A Study of Students. Journal Of Advertising Research, 43(3), 252-267. Dias, M. B., Locher, M., Li, W., El-Deredy, W. & Lisboa, P.J. (2008). The value of personalized recommender systems to e-business: a case study. In Proceedings of the 2008 ACM conference on Recommender systems, RecSys '08, New York, NY, USA, p. 291-294.ACM.
Dinev, T., & Hart, P. (2004). Internet privacy concerns and their antecedents - measurement validity and a regression model. Behaviour & Information Technology, 23(6), 413-422
Doubleclick (2003), Ad Serving Trend Report Doubleclick Q3 2003, Retrieved October 16 2012 from http://banners.noticiasdot.com/ termometro/boletines/docs/
consultoras/doubleclick/2003/doubleclick_1q2003_ingles.pdf
44
Edwards, S.M., Li, H. & Lee, J.-H. (2002). Forced exposure and psychological reactance: antecedents and consequences of the perceived intrusiveness pop-up ads. Journal of Advertising, 31(3), pp. 83–95.
Finn, A. (1988). Print Ad Recognition Readership Scores: An Information Processing Perspective. Journal Of Marketing Research (JMR), 25(2), 168-177.
Gallagher, K., Foster, K.D. & Parsons, J. (2001) The medium is not the message: advertising effectiveness and content evaluation in print and on the web, Journal of Advertising Research, 41, 4, pp. 57–70.
Goldfarb, A., & Tucker, C.E. (2011a). Online Display Advertising: Targeting and Obtrusiveness. Marketing Science, 30(3), 389-404.
Goldfarb, A., & Tucker, C. E. (2011b). Privacy Regulation and Online Advertising. Management Science, 57(1), 57-71.
Goldsmith, R. E., & Emmert, J. (1991). Measuring Product Category Involvement: A Multitrait-Multimethod Study. Journal Of Business Research, 23(4), 363-371.
Gotlieb, J. B., & Sarel, D. (1991). Effects of Price Advertisements on Perceived Quality and Purchase Intentions. Journal Of Business Research, 22(3), 195-210.
Grewal, D., Krishnan, R.R., Baker, J., & Borin, N. (1998). The Effects of Store Name, Brand Name and Price Discounts on Consumers' Evaluations and Purchase Intentions. Journal Of Retailing, 74(3), 331-352.
Hausman, A. V., & Siekpe, J. (2009). The effect of web interface features on consumer online purchase intentions. Journal Of Business Research, 62(1), 5-13.
Helft, M., Vega, T. (2010). Retargeting ads follow surfers to othersites, The New York Times, August 29, 2010.
45
IAB (2011), Internet Advertising Revenue Report conducted by PWC, Retrieved May 21 2012 from http://www.iab.net/media/file/IAB_Internet_Advertising_Revenue _Report_FY_2011.pdf
IAB (2012), Display Advertising Guidelines. Retrieved June 28 2012 from http://www.iab.net/guidelines/508676/508767/displayguidelines
Kirmani, A., & Zhu, R. (2007). Vigilant Against Manipulation: The Effect of
Regulatory Focus on the Use of Persuasion Knowledge. Journal Of Marketing Research (JMR), 44(4), 688-701.
Komiak, S. Y. X., & I. Benbasat (2006). The effects of personalization and familiarity on trust and adoption of recommendation agents. MIS Quarterly 30(4), 941-960. Lambrecht, A., Tucker, C., (2011). When does retargeting work? Timing information specificity. MSI Working Paper 11-105.
Laurent, G., & Kapferer, J. (1985). Measuring Consumer Involvement Profiles. Journal Of Marketing Research (JMR), 22(1), 41-53.
Leigh, J.H. (1984). Recall And Recognition Performance for Umbrella Print Advertisements. Journal Of Advertising, 13(4), 5-30.
Li, H., Edwards, S.M. & Lee, J.-H. (2002). Measuring the intrusiveness of
advertisements: scale development and validation. Journal of Advertising, 31(2), pp. 37–47.
Lohse, G. L., & Bellman, S. (2000). Consumer buying behavior on the Internet: Findings from panel data. Journal Of Interactive Marketing (John Wiley & Sons), 14(1), 15-29.
MacKenzie, S. B., & Lutz, R. J. (1989). Am Empirical Examination of the Structural Antecedents of Attitude Toward the Ad in an Advertising Pretesting Context. Journal Of Marketing, 53(2), 48-65.
46
Malhotra, N. K., Sung S., K., & Agarwal, J. (2004). Internet Users' Information Privacy Concerns (IUIPC): The Construct, the Scale, and a Causal Model. Information Systems Research, 15(4), 336-355.
Manchanda, P., Dube, K.Y., Goh, & Chintagunta, P.K. (2006). The effect of banner advertising on internet purchasing. Journal of Marketing Research 43(1), 98- 108. McCoy, S., Everard, A., Polak, P., & Galletta, D. F. (2008). An Experimental Study of Antecedents and Consequences of Online Ad Intrusiveness. International Journal Of Human-Computer Interaction, 24(7), 672-699.
Mittal, B. (1989). Measuring Purchase-Decision Involvement. Psychology & Marketing, 6(2), 147-162.
Morrison, D. G. (1979). Purchase Intentions and Purchasing Behavior. Journal Of Marketing, 43(2), 65-74.
Novak, T.P., & Hoffman, D.L. (1997). New Metrics for New Media. Toward the Development of Web Measurement Standards. World Wide Web Journal, 2 (1), 213– 246.
Palazon, M., & Delgado-Ballester, E. (2009). Effectiveness of price discounts and premium promotions. Psychology & Marketing, 26(12), 1108-1129.
Park, J., Lennon, S. J., & Stoel, L. (2005). On-line product presentation: Effects on mood, perceived risk, and purchase intention. Psychology & Marketing, 22(9), 695-719. Pavlou, P.A., & Stewart, D.W. (2000). Measuring the Effects and Effectiveness of Interactive Advertising: A Research Agenda. Journal of Interactive Advertising, 1 (1), 62-78
47
Pfeiffer, M., & Zinnbauer, M. (2010). Can Old Media Enhance New Media? How Traditional Advertising Pays off for an Online Social Network. Journal Of Advertising Research, 50(1), 42-49.
Ranganathan, C. C., & Ganapathy, S. (2002). Key dimensions of business-to-consumer web sites. Information & Management, 39(6), 457.
Sheehan, K., & Hoy, M. (1999). Flaming, Complaining, Abstaining: How Online Users Respond to Privacy Concerns. Journal Of Advertising, 28(3), 37-51.
Simonson, I. (2005). Determinants of Customers' Responses to Customized Offers: Conceptual Framework and Research Propositions. Journal Of Marketing, 69(1), 32-45. Thongpapanl, N., & Ashraf, A. (2011). Enhancing Online Performance Through
Website Content and Personalization. Journal Of Computer Information Systems, 52(1), 3-13.
Tucker, C. E. (2012). The economics of advertising and privacy. International Journal Of Industrial Organization, 30(3), 326-329.
Unnava, H., & Burnkrant, R.E. (1991). An Imagery-Processing View of the Role of Pictures in Print Advertisements. Journal Of Marketing Research (JMR), 28(2), 226-231.
Wathieu, L., Friedman, A. (2009). An empirical approach to understanding privacy concerns. ESMT Working Paper 09-001, ESMT European School of Management and Technology, Berlin.
Wen, G., & Maddox, L. M. (2003). Measuring Web Advertising Effectiveness in China. Journal Of Advertising Research, 43(1), 34-49.
48
APPENDIX I – QUESTIONNAIRE
INTRODUCTION Beste deelnemer,
Om mijn studie aan de Rijksuniversiteit Groningen succesvol af te ronden, heb ik uw hulp nodig bij het invullen van deze enquête. Het onderzoek gaat over online adverteren en neemt maximaal 5 minuten van uw tijd in beslag.
De resultaten worden anoniem verwerkt. Alvast hartelijk dank voor uw deelname! Krisjan Oldekamp
Student MSc Business Administration / Marketing Management
k.m.oldekamp@student.rug.nl
PAGE 1
Het eerste deel van deze enquête bevat enkele algemene vragen over uzelf. Wat is uw leeftijd? __________
Geslacht man / vrouw
Hoe vaak heeft u in het afgelopen jaar online een aankoop gedaan? (schatting) ________
Wat is uw hoogst genoten opleiding?
49 PAGE 2
De volgende vragen gaan over online privacy. Geeft u alstublieft aan in hoeverre u het eens bent met de onderstaande stellingen.
[7 point Likert scale ranging from strongly disagree till strongly agree]
1. Ik ben bezorgd over het verstrekken van informatie met betrekking tot mijn voorkeuren.
2. Ik ben bezorgd over anonieme informatie die over mij verzameld wordt. (Informatie die automatisch verzameld wordt, maar NIET gebruikt kan worden om mij te identificeren, zoals netwerk informatie, surfgeschiedenis en zoekgeschiedenis)
3. Ik ben bezorgd over hoe persoonlijke informatie, die ik vrijwillig heb verstrekt, maar NIET gebruikt kan worden om mij persoonlijk te identificeren, gebruikt wordt door bedrijven (leeftijdscategorie, geslacht, cijfers van je postcode).
4. Ik ben bezorgd over hoe persoonlijke informatie die ik vrijwillig heb verstrekt en WEL gebruikt kan worden om mij persoonlijk te identificeren gebruikt wordt door bedrijven ( informatie zoals naam, adres, credit card of bank gegevens en burgerservicenummer).
PAGE 3
Zou u voordat u verder gaat naar de volgende pagina, eerst de onderstaande tekst zorgvuldig door willen lezen? Het is belangrijk dat u zich zoveel mogelijk inleeft in de situatie die beschreven wordt, aangezien hierna vragen worden gesteld met de beschreven situatie in gedachte.
50
Hieronder staan enkele voorbeelden van rugzakken die u bekeken heeft in de webshop van Perry Sport. Bekijk de producten goed. De volgende pagina zal u enkele vragen stellen met de beschreven situatie in gedachten.
51 PAGE 4
52 PAGE 5
De volgende vragen gaan over de advertentie banner op de website van de NOS. Houdt u daarbij de eerder geschetste situatie in gedachten.
Hoeveel verschillende producten zag u in de banner? ______
Werden de producten aangeboden met een korting? o Ja
o Nee
Ik vond dat de adverteniebanner.. Helemaal mee oneens Helemaal mee eens afleidend was 1 2 3 4 5 6 7 verstorend was 1 2 3 4 5 6 7 geforceerd was 1 2 3 4 5 6 7 in de weg stond 1 2 3 4 5 6 7 binnendringend was 1 2 3 4 5 6 7 opdringerig was 1 2 3 4 5 6 7
Geeft u alstublieft aan in hoeverre u het eens bent met de onderstaande stellingen. [7 point Likert scale ranging from strongly disagree till strongly agree]
1. Ik zal in de nabije toekomst zeker een rugzak kopen op de website van Perry Sport
2. Ik ben van plan om in de nabije toekomst een rugzak te kopen op de website van Perry Sport
3. Het is waarschijnlijk dat ik in de nabije toekomst een rugzak koop op de website van Perry Sport
53 PAGE 6
De laatste vraag gaat over uw betrokkenheid bij rugzakken. Geeft u alstublieft aan in hoeverre u het eens bent met de onderstaande stellingen.
[7 point Likert scale ranging from strongly disagree till strongly agree] Rugzakken..
1. Zijn erg belangrijk voor mij 2. Interesseren me erg
3. Laten me totaal onverschillig
PAGE 7
Nogmaals hartelijk dank voor uw deelname aan dit onderzoek! De resultaten worden anoniem verwerkt.
Met vriendelijke groet, Krisjan Oldekamp
