The effects of using social data for personalized
advertising.
“The 2-‐sided coin that is called social targeting”
Master thesis
Author: Frank Hattink (10370579)
University of Amsterdam, Faculty of Economics and Business
MSc Business Studies – Marketing Track
November 11, 2013
Under supervision of: drs. J. Demmers Second assessor: Prof. dr. W.M. van Dolen
Table of contents
Abstract 2
Introduction 3
Theoretical framework 10
The online advertising industry 10 Behavioral targeting 12 Social targeting 15 Privacy concerns 18 Relational commitment 21 Consumer responses 22 Conceptual framework 27 Methodology 27 Experimental design 28 Stimuli and procedure 28
Measurements 29 Independent variables 29 Dependent variables 30 Mediating variables 31 Conditions 32 Survey 34 Results 35 Data 36 Manipulation checks 37 Experiment 37 Dependent variables 39 Mediating variables 42 Survey 47 General discussion 50 Findings 51 Implications 54
Limitations and future research 55
Conclusion 56
References 59
Appendix 64
Appendix 1: Questionnaire 64 Appendix 2: CRM Cycle 78 Appendix 3: Mediation Models 79
Abstract
Companies are nowadays more and more relying on specific targeting techniques to personalize Ads to increase effectiveness. Recent developments allow marketers to use social targeting techniques, in which Ads are personalized based on information posted on social networking sites. An online experiment and survey have been conducted to investigate consumer responses towards the use of such targeting techniques. Previous research mainly focused on the negative aspects of using such targeting techniques, whereas current research finds evidence for a positive side. Allowing consumers to give permission for the use of their social data by companies seriously affects their effectiveness. Social targeting seems to be a 2-‐sided coin, without permission privacy concerns increase, which leads to lower scores on Ad evaluation, click-‐through rates and firm evaluation. With permission, consumers have the feeling that the company is involved and committed, therefore perceived relational commitment scores increase, which lead in turn to higher scores on Ad evaluation, click-‐through rates and firm evaluation. Current research sheds a new light into the underexplored area of new advertising techniques. Welcome to the new world of advertising!
Introduction
MOTIVES
General introduction: Phenomenon
Recently, a man walked into the local Target store (an American retail chain) and demanded to have a word with the manager. His high school daughter was receiving coupons and promotions for items that indicated she was pregnant. The man asked if Target was encouraging his daughter to get pregnant. The manager couldn’t answer his question and the man went home. There he talked to his daughter and found out that she was indeed pregnant (Greengard, 2012), (Hill, 2012). Specialized software enables Target to make personalized offers based on previous buying patterns and behavior. These online and offline data indicated that she was expecting a baby (Greengard, 2012). Welcome to the new world of advertising!
Previously mentioned story is only one example of the use of online and offline data in advertising. Probably everyone can think of examples themselves in which these kinds of data are being used. Consider for example the story of Julie Matlin, which was featured in the New York Times. The shoes that she searched for on the internet subsequently followed her almost everywhere she went online and in the end she even had the feeling that the shoes were stalking her. “It is a pretty clever marketing tool. But it’s a little bit creepy, especially if you don’t know what’s going on” (Helft & Vega, 2010).
Technological advances increased the ability of firms to target their (digital) advertising to specific customers. These advertisements (Ads) are using information about customers to personalize the content (Tucker, 2011b). Targeting based on consumers online
and/or offline behavior is a technique used by online advertisers to enhance the effectiveness of their Ad campaigns (Yan et al, 2009). Ads can for example be delivered based on information about an individual’s previous web searches and browsing behavior. Based on visited pages and searches made, Ads are displayed to individuals that are most likely to be influenced by the content. For example, somebody that has recently visited several electronics sites would be more likely to receive Ads for TV sets than, say, somebody whose previously visited sites contained pictures of cats.
Research on the effects of this kind of advertising yielded mixed feelings among consumers. Firms are facing the risk that customers will find these tailored Ads intrusive and invasive of their privacy (White et al., 2008). Turow (2009) found that 86% of young adults do not want tailored Ads that are the result of their previous internet activities. These potential negative customer reactions reduced advertiser’s confidence in targeting techniques, fearing that customers would resist tailored Ads (Lohr, 2010). However, advertising rates for Ads that are behaviorally targeted are higher and more successful than standard online Ads (Beales, 2010). Therefore, on the one hand, using such targeting techniques may lead to privacy concerns (White et al., 2008). On the other hand, using such targeting techniques seems to be more valuable to consumers, as it is more likely to tell them about a product they probably want to buy (Beales, 2010).
Specific introduction: Gap
Privacy concerns as mentioned above are especially present on social networking sites (SNS) like Facebook and Myspace (Gross & Acquisti, 2006). These websites are collecting huge amounts of personal data from their users. This data can be used by advertisers to tailor their Ads to customers (Stone, 2010). Research in this emerging field is very limited.
One example of using such techniques is social advertising, in which names and actions of individual’s online friends on SNS are used to personalize the Ads. Social advertising is found to be effective, which stems from the ability of advertisers to uncover customers that are likely to respond in the same way. However, when it is obvious that advertisers try to promote social influence, social advertising turns out to be less effective (Tucker, 2011a). This effectiveness of personalized Ads on SNS is also investigated in another study by Tucker (2011b). After improved privacy controls by SNS, users are found to be more likely to click on personalized Ads. This increase in effectiveness is larger for Ads that used more unique private information (Tucker, 2011b).
Both studies have some important limitations; especially the second study by Tucker (2011b) is limited in some important ways. First, a non-‐profit company with an appealing cause has been used in the experiment. Will these results be generalizable when using a for-‐ profit company? Second, the data used in the field experiment contained personalized Ads that matched with user’s profiles. For example matching an Ad with a celebrity that is liked or followed. How will people respond when they get personalized Ads based on self-‐ generated content (e.g. their own status updates or tweets)? And how about privacy concerns in this specific setting, since privacy concerns are shown to influence Ad effectiveness? (Goldfarb and Tucker, 2011c). Finally, Tucker (2011b) calls for an explicit “opt-‐in” approach to share information that explicitly addresses advertising. “Opt-‐in” means that customers have to give permission to companies for the use of their data. This is an important topic, as Internet user’s perception of control affects the likelihood to click on online Ads (Tucker, 2011b).
PROBLEM DEFINITION
Problem statement
The very limited literature regarding the effectiveness of Ads using SNS data opens up a rich area for further research. SNS is an important topic, because SNS are important media platforms that are growing rapidly, in importance and in reach (Tucker, 2011b).
In this study, the main research question will be:
“What is the effect of using data from social networking sites to personalize advertisements on consumer behavior”?
This research is commissioned by a large Dutch bank. For privacy issues, the name of this bank will not be mentioned and will be referred to as Bank X. Since a couple of years Bank X is interacting with its customers on SNS. Webcare activities aimed at customer service are of central concern. These webcare activities involve helping customers with their online questions, complaints and suggestions about the services of Bank X through social media channels. SNS offers Bank X excellent opportunities to approach their customers with relevant content to create brand preferences. For every SNS platform, Bank X developed a content strategy aimed at different types of users (e.g. entrepreneurs, youth, students). However, these different platforms offer insufficient tools to offer fully personalized and relevant content in their communication at times when customers experience certain relevant life events (e.g. graduation, first job, getting a child, changing jobs). Life events like these are relevant for companies in the banking sector because they give banks the opportunity to offer people more of their services. Due to the increasing use of SNS by people and business it is getting more and more difficult to catch the attention. Therefore
Bank X is interested in using SNS data to create relevant personalized content and brand preferences.
CONTRIBUTION
This study aims to contribute to the scientific literature by addressing an important gap. Simultaneously, Bank X may benefit from the findings, in a way that they gain interesting insights in (non)customer responses towards the use of SNS data in their marketing efforts.
Theoretical contributions
Research on advertising using personal SNS data to tailor Ads is very limited. This kind of advertising can be seen as a new type of behavioral targeting (BT). BT is defined as the tracking of consumers’ online activities in order to deliver tailored advertising (FTC Staff Report, 2009). BT delivers Ads to targeted users based on individual’s web search and browsing behavior (Yan et al., 2009). Ads using SNS data are not related to web search and browsing behavior. However, they are still a form of BT, since consumer’s online (SNS) activities are being tracked in order to tailor Ads. This form of targeting is still radically different than regular BT methods, since SNS are, in contrast to BT, all about voluntarily sharing and interaction (Kaplan & Haelein, 2009). High growth rates of SNS led companies to invest in advertising on these networks (Boyd & Ellison, 2008). One of the reasons that Ads are avoided in this environment is that the Ads are not relevant to the user (Kelly et al., 2010), which provides opportunities for a more behavioral based approach to create more relevant Ads.
could provide more specific personal information than search and browsing behavior. For example, a user that recently graduated could share this information on SNS, without searching or browsing on this specific topic. Ads using SNS data may therefore be more specific and relevant than standard BT Ads. Targeted Ads based on SNS data seems to be one step further in BT advertising and should therefore be treated as a distinct research topic, since consumer responses possibly differ to such Ads as opposed to regular BT Ads.
In order to be successful with this form of targeting, companies should be active and take the lead on SNS to develop customer relationships (Kaplan & Haelein, 2009). Offering a proactive form of targeting which involves using SNS data to tailor the Ads may therefore lead to such customer relationships. Previous research mainly focused on the negative (privacy related) aspects of using BT techniques (e.g. Turow et al., 2009). However, consumers could perceive Ads using SNS data differently than standard BT Ads. Consumers may experience the company as involved and committed, since consumers are likely to be more loyal when they engage with a company on SNS and are therefore more willing to try new offerings (Culnan et al., 2010). This is in line with the research by Baird & Parasnis (2011) who found that almost 50% of the respondents are more likely to do future purchases when they engage with a company on SNS. Based on the above, there is reason to believe that consumer responses to targeted Ads based on SNS data are different than to BT Ads. Therefore, current study treats targeting based on SNS data as unique and distinctive with respect to previous BT research. However, Ads based on SNS data are probably only successful when companies actively engage with their customers on SNS, since this engagement may lead to relationships and more loyal customers.
Tucker (2011a) and (2011b) made a first start in researching targeting based on SNS data. Both studies are still in progress and are mainly focused on social influence on SNS. However, these studies provide some first interesting insights in targeting based on SNS data. Current research uses Tucker’s suggestion for an explicit “opt-‐in” approach in which consumers have to give permission to companies for the use of their SNS data, since permission is likely to seriously affect consumer responses towards (SNS) targeting techniques. Furthermore, current research extends these studies by using a for profit company in the experiment. No studies yet investigated the effectiveness of targeting based on SNS data in a for-‐profit setting. Current research will investigate how far companies can go in their advertising initiatives.
Managerial contributions
The increased popularity and use of SNS (Cheung et al., 2012) offers great potential for marketers. New technologies offer great opportunities to target individuals that are most likely to be influenced by specific Ads (Tucker, 2011b). In this way, advertising can be much more efficient, because less Ad impressions are wasted. But it also raises privacy concerns (White et al, 2008). It is important for managers to know how far they can go in their advertising initiatives. On the one hand, you don’t want to scare people off, with the risk of losing customers and/or negative publicity. On the other hand, you want to be as efficient as possible to increase profit. Therefore, it is important to find the perfect balance to increase company goals.
STRUCTURE/OUTLINE
The next chapter, chapter 2, introduces the literature review based on main theories and concepts. The methodology of this research will be outlined in chapter 3. In chapter 4 the data will be analyzed and outcomes will be presented. Finally, the discussion and conclusions will be presented in chapter 5.
Theoretical framework
THE ONLINE ADVERTISING INDUSTRY
Advertising networks (often referred to as third party Ad networks) are intermediaries that connect publishers with advertisers who are seeking to reach an online audience. These Ad networks serve a broad range of publishing partners and purchase available advertising space from publishers and resell it to the ultimate advertisers. Both involved parties benefit from these Ad networks (Beales, 2010). Ad networks are nowadays relying more and more on specific techniques to increase their efficiency. New techniques in online advertising “replace a sledgehammer with a scalpel”, because of greatly increased specificity due to finer and finer Ad targeting (Evans, 2009).
In the United States, Internet advertising revenues totaled $36.6 billion over the year 2012, 15.2% more than in 2011. The compound annual growth rate of 19.7% over the past 10 years outpaced US real GDP growth of 1.5% over the same period. In the online advertising market in 2012, (1) search and (2) display advertising lead the Ad formats with 46,3% and 33% respectively (PwC IAB report, 2012). (1) With search advertising, advertisers usually pay on a “CPC” basis, which means costs per click. Advertisers can bid on specific keywords that consumers enter in a search engine. Every time someone performs a search,
an auction is hold to determine which Ads are displayed on the results page and the position of the Ads on the page (Google Ad Words, Google Inc.). Bidders can specify specific keywords, a maximum price per click, and the text Ad it wants to display. With click-‐through rates, it is easy to measure the searcher’s response to these Ads. This combination of targeting and measuring makes search advertising extremely effective (Levin & Milgrom, 2010). (2) Display advertising includes display/banner Ads, rich media, digital video and sponsorship. Display Ads are typically sold on a “CPM” basis. CPM stands for "cost per 1000 impressions." Advertisers running CPM Ads set their desired price per 1000 Ads served and pay each time their Ad appears (Google Ad Words, Google Inc.). Matching an Ad’s content with the content of the website and increasing the obtrusiveness, for example a full screen Ad, is found to increase purchase intent, as long these do not occur in combination (Goldfarb & Tucker, 2011a). A final bidding option is “CPA”, which means costs per acquisition. Advertisers pay each time someone actually purchases the advertised product or service (Google Ad Words, Google Inc.).
Ad networks use three different strategies for matching advertisers with users of Internet content and services, (1) contextual, (2) vertical and (3) behavioral strategies (Beales, 2010). First, contextual networks are based on the content of the page. Search engine platforms sell Ads on the pages of publishers that belong to their networks. Advertisers bid on keywords just as they do for search advertising. Ads are shown based on whether those keywords appear on a page (Evans, 2008). This form of content match relies heavily on selecting Ads relevant to the page content with little focus on the user (Joshi et al., 2011). Matching an Ad’s content to the content of the website is found to increase purchase intent among exposed consumers (Goldfarb & Tucker, 2011a). Second, vertical
networks consist entirely of sites within a specific industry and can create significant scale of a highly desirable audience across strong publishers. For example, clothing companies are likely to want to advertise in publications geared toward viewers interested in clothes, so vertical networks will group together these clothing websites (Lowe, 2006). Third, behavioral networks use behavioral targeting (BT), a technique used by online advertisers to increase Ad effectiveness. BT delivers Ads to targeted users based on individual’s web search and browsing behavior (Yan et al., 2009). The Federal Trade Commission (FTC) defines BT as the tracking of consumers’ online activities in order to deliver tailored advertising (FTC Staff Report, 2009).
One view of internet advertising is that it will move increasingly toward finer and finer Ad targeting, in which every impression is treated as distinct and unique (Levin & Milgrom, 2010). This is in line with the view of Evans (2009), who states that internet advertising is transforming by providing more and more efficient methods to match advertisers and consumers.
BEHAVIORAL TARGETING
In the introduction we have already seen two examples of using BT techniques, the Target example and the “stalking shoes” example. Why would advertisers use such techniques? Many advertisers believe that consumer’s browsing and shopping behavior indicate what products they like and which Ads will catch their attention (Turow et al., 2009). The Internet is a form of mass media with targeted Ads relying on massive data collection on an incredible scale (McDonald & Cranor, 2010), in which BT has been applauded as the new “Holy Grail” in online advertising (Chen & Stallaert, 2010).
Advertisers have struggled for years to better understand the whims of the marketplace and target consumers more effectively, but there has been a revolution over the last few years (Greengard, 2012). One of the new realities of advertising is that personal information can be used to ensure that consumers are only seeing Ads that are relevant to them. In theory this means that advertising can be more informative to consumers than it was before (Tucker, 2012). Ads can be targeted to consumers who value the information the most and are most likely to act on it (Evans, 2009), (McDonald & Cranor, 2009). These targeted Ads have obvious benefits to advertisers because fewer Ad impressions are wasted (Goldfarb & Tucker, 2011d), (McDonald & Cranor, 2010), (Kim et al., 2001), the time it takes to find products is reduced (McDonald & Cranor, 2010) and the likelihood of sales is higher (Evans, 2009). BT tries to serve more relevant Ads to consumers using information about behavior, including sites visited and interest in particular types of content (Beales, 2010). Generic user profiles can be created with this information (McDonald & Cranor, 2010), in which historical user activity is key (Ahmed et al., 2011) and there is a clear relationship with click performance (Joshi et al., 2011). With statistical models of BT, click-‐through rates of Ads can be predicted from user behavior (Chen et al., 2009). These models determine whether a specific individual that is browsing on a website has performed browsing behaviors and personal characteristics that make that individual a good target for an Ad (Evans, 2009). When an Ad is personalized, consumers are more likely to assume that there is a match between them and the advertised product (Anand & Shachar, 2009).
In the Target example from the introduction, previous buying patterns and behaviors (offline behavioral data) were being used to personalize Ads. Customized and personalized Ads based on past purchasing behavior are shown to be a critical success factor for internet
stores and web service providers (Kim et al., 2001). In the other example about the “stalking shoes”, earlier browsing history (online behavioral data) was being used to personalize Ads. The latter is known as retargeting in the literature. With retargeting, information from internal browsing data is used to improve internet advertising content on external websites. Consumers who previously visited a firm’s website when surfing online, are shown Ads that contain images of products they have looked at before on the firm’s website (Lambrecht & Tucker, 2012). In short: retargeting tries to get consumers back to the firm’s previously visited website. Lambrecht and Tucker (2012) found that these retargeted Ads are less effective than generic Ads. However, retargeted Ads are more effective than generic Ads when consumer browsing behavior suggests stable product preferences, which is indicated by the search for product reviews. BT Ads that have a high fit with consumer preferences may increase purchase intention (Franke et al., 2009), while low fit BT Ads is found to cause irritation (Thota & Biswas, 2009). Retargeting is only one variant of BT. Another variation is clustering, or grouping users into categories based on their web behavior (Beales, 2010).
The whole idea behind BT is to increase Ad effectiveness. Beales (2010) found that behaviorally targeted Ads are more effective than standard Ads, creating greater utility for consumers and clear appeal for advertisers. This is in line with the findings of Yan et al. (2009), who found that the click-‐through rate can be improved by as much as 670% when using BT techniques. Joshi et al. (2011) also found a clear relationship between BT and click performance. Short term user behavior is shown to be more effective than long term user behavior for BT and customers that click on the same Ad, have similar behaviors on the web (Yan et al., 2009). According to Baek & Morimoto (2012), increased personalization of Ads with BT techniques directly leads to decreased Ad avoidance. Because of the potential of
higher effectiveness of BT Ads, advertisers typically pay a higher price than standard online Ads (Evans, 2009), (Beales, 2010), (McDonald & Cranor, 2009).
SOCIAL TARGETING
A person’s web browsing patterns, credit history, what magazines they read, and conversations they have had on social networking sites (SNS), offer deep insight into life events and changes and can nowadays be used by marketers to customize Ads (Greengard, 2012). Advertisers can make use of information posted on SNS such as Facebook to identify for example new mothers (Tucker, 2012). This allows companies to create highly targeted and segmented advertising profiles and the delivery of the most customized product offerings based upon consumer’s individual interests (Shelton, 2012). To identify pregnant women for example, mixing and matching a variety of targets, such as women that have interest in baby products in combination with the like of children’s music, produce a high likelihood of hitting the market (Delo, 2012). Both men and women aged between 18-‐24 are found to reject Ads targeted based on SNS data, with stronger results for women (Hoy & Milne, 2010). However, acceptance of this form of targeting remains questionable due to the very limited generalizability of the research by Hoy and Milne (2010). In this research, it is expected that Ads targeted based on social data are better evaluated than standard online Ads due to the higher informativeness of ST Ads.
On Facebook, Ads can be aimed toward finely segmented groups of users, based on gender, age, location and preferences (such as favorite music and activities) (Stone, 2010). For example, concert promoters can show Ads for a band’s concert to a select group of Facebook users who live in the area and that have mentioned the band’s name on their profile page or in their status updates. Or a wedding photographer can show Ads only to
people who live in a certain city and that have switched the status of their relationship to “engaged”. There now is the possibility that Facebook users get tailored Ads based on their changed or updated status (Stone, 2010). In the previously described paragraphs we have seen the industry moving from “a sledgehammer to a scalpel”. But now with the new possibilities of using data from SNS (also referred to as social data) in advertising, the industry is moving towards an almost microscopic precision of targeting.
This new form of targeting is a highly underexplored area in the emerging field of online advertising techniques. Tucker (2011b) investigated the relative effectiveness of personalizing Ad copy with posted personal information on SNS and found that, after improved privacy controls, users were twice as likely to react positively to personalized Ad content and click on personalized Ads. This increase in effectiveness was larger for Ads that used more private information to personalize the Ad. Tucker (2011b) found no comparable change in Ad effectiveness that did not explicitly mention that private information was being used when targeting. One important limitation of the study by Tucker (2011b) is that the experiment was conducted by a non-‐profit company with an appealing cause. Consumers may respond differently to personalized Ads from for-‐profit companies. Tucker (2011b) calls for an explicit “opt-‐in” approach to share information that explicitly addresses advertising. “Opt-‐in” means that customers have to give permission to companies for the use of their data.
Another study by Tucker (2011a) investigated social advertising, in which Ads are targeted based on underlying social networks. The content of these Ads is tailored with information relating to the social relationship. A social Ad is an online Ad that incorporates user interactions that the consumer has agreed to display and be shared. The resulting Ad
displays these interactions along with the user’s persona (picture and/or name) within the Ad content (Interactive Advertising Bureau [IAB] 2009). Tucker (2011a) found that social Ads are effective and this effectiveness stems mainly from the ability to uncover consumers that are likely to respond the same. But social Ads are less effective when it is explicitly stated in the Ad that the advertiser is trying to promote social influence (Tucker, 2011a). Like the other study by Tucker (2011b), this research by Tucker (2011a) has the same important limitation with using a non-‐profit company, which may bias the results.
Tucker (2011b) and (2011a) investigated personalized Ads on SNS, with both different approaches. Things the user liked on SNS (Tucker, 2011b) and names of user’s friends (Tucker, 2011a) were used to personalize the Ad. In other words: social data is used to personalize the Ad. In the literature, there is no name yet to this new form of targeting. In this study, this kind of targeting will be referred to as “social targeting” (ST): a form of BT in which targeting is based on social data. Current research will take a different approach with respect to previous research in three major ways. First, a clear distinction will be made between ST with permission and ST without permission. Second, a for-‐profit company will be used in the experiment to see if Tucker’s findings (2011b) and (2011a) are generalizable to a for-‐profit setting. It is expected that ST Ads are more effective than standard online Ads. Third, Ads will be personalized based on the relevant content of status updates that users are posting on SNS. Specifically, when users are posting status updates about certain life events, they will be approached by a company with a relevant Ad. Also this study will address other important issues, such as user’s privacy concerns.
PRIVACY CONCERNS
Firms often make use of personal information to customize their communications (Ansari & Mela, 2003). One would think this is exactly what people want: Ads that are as relevant to their lives as possible (Turow et al., 2009). However, for the collection of such data, advertisers require some degree of privacy intrusion, which sets up a tradeoff between the informativeness of advertising and the degree of privacy intrusion (Tucker, 2012). This tradeoff is also referred to as the privacy calculus in the literature, which suggests that anticipated benefits and perceived risks influences a user’s decision to share information on SNS (Dinev & Hart, 2006). Consumers only have some degree of control over their privacy (Evans, 2009). Consumer responses may be negative because such targeted Ads may be perceived as too personal (White et al., 2007), leading to criticism on search engine providers for capturing and storing customer data (Dye, 2009). Recent plans by Visa and MasterCard to use information about consumer’s credit-‐card purchases for targeting online Ads caused worried reactions among consumers (Steel, 2011).
According to a report by TRUSTe (2008), 57% of the respondents are not feeling comfortable that advertisers use online behavioral data to serve relevant Ads, even when this information cannot be tied to their names or any other personal information. At that moment BT was an uncharted territory without clear laws and regulations. In February 2009, the FTC published guidelines for companies collecting data of web users specifically for Ad targeting (FTC Staff report, 2009). One of the principles is to encourage customer control and transparency. The top online privacy concerns have been studied in 2002 and again in 2008 by Antón et al. (2009). Information transfer, notice/awareness, and information storage were the top online privacy concerns of Internet users in 2002 and this
top 3 has not changed in 2008. Only the level of concern of information used for BT has increased. In a report by Gomez et al. (2009), the biggest privacy concern among Internet users is the lack of control. Turow et al. (2009) found that most American adults (66%) do not want BT Ads tailored to their interests. When Americans are informed about the ways in which such data is collected by advertisers, even higher percentages (73% -‐ 86%) say no to BT Ads. But it is not exactly clear why Americans do not want these BT Ads; however, there are some explanations in the literature.
Consumers may experience a reaction similar to psychological reactance, a motivational state arising in a person whose freedom is perceived to be threatened (Brehm, 1966). White et al. (2007) build upon this research to suggest that Ads that are too personal may result in “reactance”, which means that consumers resist intimidating Ads in behaving the opposite way to the one intended. These BT-‐related privacy issues resulted in a 75% reduction in BT advertising (Ponemon institute, 2010). However, reactance does not directly feature the cause of inconvenience associated with privacy intrusion (Tucker, 2012).
McDonald & Cranor (2009) studied how people perceive BT. In their study people raised privacy concerns spontaneously, without knowing that the study had to do with BT-‐ related privacy concerns. Consumers are found to have a very poor understanding of how Internet advertising techniques works (McDonald & Cranor, 2009), which is in line with the findings by Ur et al. (2012), who also found a lack of knowledge in online behavioral advertising techniques. In another study, McDonald and Cranor (2010) investigated consumers’ view on advertising and the ability to make decisions about privacy tradeoffs. McDonald and Cranor (2010) found a gap in consumers’ knowledge to make effective privacy decisions. In this second study by McDonald and Cranor (2010), another gap is found
between consumer’s willingness to pay to protect their privacy and the willingness to accept discounts in exchange for private information, while this actually should be in balance. This can be related to the previously mentioned privacy calculus (Dinev & Hart, 2006).
As stated earlier, matching an Ad to the content of the website and increased Ad obtrusiveness increase Ad effectiveness in isolation (Goldfarb & Tucker, 2011a), however, these strategies are not effective in combination. This failure appears to be related to privacy concerns, because this failure is more pronounced for people with higher privacy concerns (Goldfarb & Tucker, 2011a). In a study by Baek & Morimoto (2012) privacy concerns are found to have a direct positive effect on Ad avoidance, when consumers have not given their permission. The question now becomes how firms should take care of this. In another study by Tucker (2011b), users were twice as likely to click on personalized Ads after improved privacy controls, which suggests that giving consumers more control over their private information can be a useful strategy. Giving users more control over their private information may therefore mitigate the tradeoff between the informativeness of advertising and the degree of privacy intrusion (Tucker, 2011b).
This tradeoff is important in a study by Goldfarb and Tucker (2011c), in which the economic effects of privacy regulation for online advertising are studied. Higher consumer privacy concerns have led governments to introduce privacy regulation. This privacy regulation restricted the use of online tracking techniques by websites to target Ad campaigns, which led display advertising to become far less effective. This loss in effectiveness was the greatest for websites with more general content (such as news sites), where it is hard to target without such online tracking techniques (Goldfarb and Tucker, 2011c). This suggests that privacy regulation may therefore lead to more intrusive Ads and
advertisers may shift their focus away from sites that are difficult to match with relevant Ads (Tucker, 2012). This development led researchers to investigate ways to target consumers without compromising privacy (Toubiana et al., 2010), (Backes et al., 2012).
RELATIONAL COMMITMENT
As discussed previously, consumers may have privacy concerns when being confronted with BT and/or ST Ads (Turow et al., 2009), (McDonald & Cranor, 2009). Another possibility is that consumers may have the feeling that the company is involved and caring and tries to build a relationship with them. In this paper, the focus is on offering Ads based on certain status updates on SNS. Nowadays, more and more firms are responding to posts on SNS (Fournier and Avery, 2011), which allows companies to interact with customers personally (Kotler & Armstrong, 2010). Responding to SNS posts provides great opportunities for extremely targeted advertising, since Ads can be personalized according to individual customer attributes (Kotler & Armstrong, 2010). Using ST techniques may therefore give consumers the feeling that the company is involved and committed to them.
Offering highly relevant products that result from the use of personal information in communications can lead to customer relationships (Ansari & Mela, 2003). Firms using ST Ads may therefore be better evaluated than firms using standard online Ads because consumers feel a (most likely positive) relationship with the company. The personalization of Ads plays a central role in customer relationship management (CRM) (Baek & Morimoto, 2012). BT can be seen as a form of CRM, which enables companies to develop unique, long-‐ term relationships (Montgomery & Chester, 2009). Based on extensive literature review, Ngai et al. (2009) developed a CRM cycle (Appendix 2), in which BT can best be related to the customer identification phase. This phase involves targeting and analyzing people that
are most likely to become customers (Ngai et al., 2009), which is in line with the previously discussed findings of Evans (2009) and McDonald & Cranor (2009).
Vesanen states: “the urge to personalize is largely driven by the expected benefits of one-‐to-‐one marketing and CRM (Vesanen 2007, p. 409). In CRM, personalization enables e-‐ business providers to execute strategies to lock in customers (Mulvenna et al., 2000). In a study by Vlasic & Kesic (2007) consumers are found to have a more positive attitude toward relationship personalization than toward classical transactional relationships. Goldsmith (1999) sees personalization as the opposite of one-‐size-‐fits-‐all and proposes that personalization is very important in marketing strategy and should therefore be featured as one of the elements of the marketing mix.
The personalization and targeting of Ads can be viewed as building customer loyalty by building one-‐to-‐one relationships with individuals (Riecken, 2000). Willingness to promote a company is a strong indicator of customer loyalty, because customers would only recommend a company when they are very loyal (Reichheld, 2003). Reichheld (2003) introduced the Net-‐Promoter Score (NPS), which tracks how customers represent a company to their friends, colleagues, etc. ST Ads may lead to more loyal customers and therefore higher NPS scores.
CONSUMER RESPONSES
Current research investigates the effects of ST Ads on consumer responses. These consumer responses involve Ad evaluation, CTR, firm evaluation and NPS. These consumer responses will be discussed below. It is expected that ST Ads have more positive consumer responses than standard online Ads.
Ad evaluation
Online behavior is shown to indicate which Ads will catch consumer’s attention (Turow et al., 2009), which in theory means that advertising can be more informative to consumers as opposed to standard online advertising (Tucker, 2012), since Ads are targeted to the consumer’s individual interests (Shelton, 2012). Therefore, it is expected that ST Ads are better evaluated than standard online Ads.
Click-‐through rates (CTR)
A clear relationship between BT Ads and click performance has been found in earlier research (Beales, 2010), (Yan et al., 2009), (Joshi et al., 2012). Ads using BT techniques were found to be more effective than standard online Ads (Beales, 2010), (Yan et al., 2009). For example, Tucker (2011a) and (2011b) found that ST Ads are effective at generating a higher click-‐through-‐rate for non-‐profit companies with an appealing cause as compared to standard Ads. Therefore, it is expected that ST Ads have higher CTR than standard online Ads.
Firm evaluation
By personalizing Ads, consumers may have the feeling that the company is showing interest and tries to form a relationship with them (Ansari & Mela, 2003), (Montgomery & Chester, 2009). Relevant product offerings that result from the use of personal information in communications may lead to customer relationships (Ansari & Mela, 2003). Therefore, it is expected that firms using ST Ads are better evaluated than firms using standard online Ads.