See discussions, stats, and author profiles for this publication at: [Link]

net/publication/318033558

Online Behavioral Advertising: A Literature Review and Research Agenda

Article  in  Journal of Advertising · June 2017

DOI: 10.1080/00913367.2017.1339368

CITATIONS READS

146 2,851

3 authors, including:

Sophie C. Boerman
University of Amsterdam
31 PUBLICATIONS   1,417 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Sophie C. Boerman on 14 November 2017.

The user has requested enhancement of the downloaded file.

 Journal of Advertising

ISSN: 0091-3367 (Print) 1557-7805 (Online) Journal homepage: [Link]

Online Behavioral Advertising: A Literature Review

and Research Agenda

Sophie C. Boerman, Sanne Kruikemeier & Frederik J. Zuiderveen Borgesius

To cite this article: Sophie C. Boerman, Sanne Kruikemeier & Frederik J. Zuiderveen Borgesius
(2017) Online Behavioral Advertising: A Literature Review and Research Agenda, Journal of
Advertising, 46:3, 363-376, DOI: 10.1080/00913367.2017.1339368

To link to this article: [Link]

© 2017 The Author(s). Published with

license by Taylor & Francis, LLC© Sophie C.
Boerman, Sanne Kruikemeier, and Frederik
J. Zuiderveen Borgesius

Published online: 29 Jun 2017.

Submit your article to this journal

Article views: 720

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

[Link]

Download by: [UVA Universiteitsbibliotheek SZ] Date: 01 August 2017, At: 07:59
 Journal of Advertising, 46(3), 363–376
Copyright Ó 2017, Sophie C. Boerman, Sanne Kruikemeier, and Frederik J. Zuiderveen Borgesius.
Published with license by Taylor & Francis, LLC.
ISSN: 0091-3367 print / 1557-7805 online
DOI: 10.1080/00913367.2017.1339368

Literature Review Corner

Online Behavioral Advertising: A Literature Review

and Research Agenda

Sophie C. Boerman, Sanne Kruikemeier, and Frederik J. Zuiderveen Borgesius

University of Amsterdam, Amsterdam, the Netherlands
Downloaded by [UVA Universiteitsbibliotheek SZ] at 07:59 01 August 2017

searched for with a search engine. This phenomenon is called

Advertisers are increasingly monitoring people’s online online behavioral advertising (OBA). In a simple example of
behavior and using the information collected to show people
OBA, an advertising network (i.e., a company that serves
individually targeted advertisements. This phenomenon is called
online behavioral advertising (OBA). Although advertisers can advertising on thousands of websites) tracks a consumer’s
benefit from OBA, the practice also raises concerns about website visits. If a consumer visits several websites about cars,
privacy. Therefore, OBA has received much attention from the network assumes the consumer is interested in cars. The
advertisers, consumers, policymakers, and scholars. Despite this network can then display ads for cars only to people (presumed
attention, there is neither a strong definition of OBA nor a clear
to be) interested in cars. Consequently, when two people visit
accumulation of empirical findings. This article defines OBA and
provides an overview of the empirical findings by developing a the same website at the same time, one may see car ads while
framework that identifies and integrates all factors that can the other (who had visited websites about furniture) may see
explain consumer responses toward OBA. The framework furniture ads.
suggests that the outcomes of OBA are dependent on advertiser- Advertisers see OBA as one of the most important new ways
controlled factors (e.g., the level of personalization) and
consumer-controlled factors (e.g., knowledge and perceptions
of reaching targeted audiences. Online advertising revenues are
about OBA and individual characteristics). The article also growing rapidly and setting records every year (Interactive
overviews the theoretical positioning of OBA by placing the Advertising Bureau 2016), and it is believed that OBA will be a
theories that are used to explain consumers’ responses to OBA in part of this growth (eMarketer 2010; Chen and Stallaert 2014).
our framework. Finally, we develop a research agenda and The industry claims that OBA creates more relevant and effi-
discuss implications for policymakers and advertisers.
cient ads and boosts ad effects (Beales 2010; Chen and Stallaert
2014). However, the practice also involves collecting, using,
In today’s digital world, advertisers have seized the oppor- and sharing personal data, and thus raises consumer privacy con-
tunity to use online data about consumers to personalize and cerns. Therefore, OBA has received much attention from regula-
target advertisements. Such data can include websites visited, tors, such as the U.S. Federal Trade Commission (FTC 2012)
articles read, and videos watched, as well as everything and European Data Protection Authorities (Article 29 Data Pro-
tection Working Party 2010), and consumer organizations. In
response, industry alliances, such as the Digital Advertising Alli-
Address correspondence to Sophie C. Boerman, University of ance in the United States and the European Interactive Digital
Amsterdam, Amsterdam School of Communication Research, P.O. Box
15791, Amsterdam, the Netherlands. E-mail: [Link]@[Link] Advertising Alliance, set up self-regulatory programs to protect
Sophie C. Boerman (PhD, University of Amsterdam) is an assis- consumer privacy and describe how to inform consumers about
tant professor of persuasive communication, Amsterdam School of data collection and usage.
Communication Research, University of Amsterdam. OBA is believed to be part of the future of advertising. It is
Sanne Kruikemeier (PhD, University of Amsterdam) is an assis- one of the new options advertisers can choose to use in their
tant professor of political communication, Amsterdam School of
Communication Research, University of Amsterdam. campaigns that allows for more precise targeting (Keller 2016;
Frederik J. Zuiderveen Borgesius (PhD, University of Amsterdam) Kumar and Gupta 2016). Leading scholars argue that advertis-
is a postdoctoral researcher, Institute for Information Law, University ing will become more personalized and targeted and will
of Amsterdam. involve more individual communication, where advertisers
This is an Open Access article. Non-commercial re-use, distribu- can iterate messages based on consumer behavior and needs
tion, and reproduction in any medium, provided the original work is
properly attributed, cited, and is not altered, transformed, or built (Kumar and Gupta 2016; Schultz 2016; Rust 2016). This
upon in any way, is permitted. The moral rights of the named emphasizes the relevance of the topic not only in practice but
author(s) have been asserted. also in the academic field.

363
 364 S. C. BOERMAN ET AL.

Despite the growing interest in OBA, a clear understanding To track consumers’ browsing behavior, companies
of OBA is lacking. First, there have been various definitions often use (tracking) cookies, but other technologies include
of OBA, making the concept ambiguous. Second, OBA flash cookies and device fingerprints (Altaweel, Good, and
research has examined a wide range of independent, mediat- Hoofnagle 2015). Recently, researchers found that the 100
ing, moderating, and outcome variables without a clear most popular sites collect more than 6,000 cookies, of
accumulation of knowledge. This is partly due to the interdis- which 83% are third-party cookies, with some individual
ciplinary nature of the field and the various interested parties, websites collecting more than 350 cookies (Altaweel,
including advertisers, consumers, computer scientists, and Good, and Hoofnagle 2015). These cookies allow compa-
policymakers. nies to collect detailed information about millions of con-
To address these issues, we first define OBA. Second, we sumers, partly for use in OBA. To illustrate the magnitude
provide an overview of empirical findings by developing a of this business, Facebook has individual profiles of 1.65
framework that identifies and integrates all factors that can billion people (Facebook 2016) and AddThis has profiles
explain consumer responses to OBA. The proposed framework of 1.9 billion people (AddThis 2016).
provides an up-to-date review of this new, and still developing, OBA differs from other types of online advertising because
type of advertising. In addition, the framework identifies the it aims at personal relevance, which often happens covertly.
Downloaded by [UVA Universiteitsbibliotheek SZ] at 07:59 01 August 2017

most important factors in examining, predicting, and evaluat- Similar to other forms of personalized advertising, such as
ing the outcomes of OBA, making the framework relevant not location-based advertising (e.g., Ketelaar et al. 2017) and ads
only to academic research but also to advertisers. Third, we that include people’s names (Bang and Wojdynski 2016),
establish the theoretical positioning of OBA by reviewing the- OBA uses personal information to tailor ads in such a way that
ories used to explain people’s responses in the context of our they are perceived as more personally relevant. A new dimen-
framework. Finally, our framework and the theoretical posi- sion to this personalization is the fact that the tracking of
tioning of OBA help identify gaps in the literature and facili- online activities, collection of behavioral data, and dissemina-
tate the development of a research agenda. tion of information often happen covertly (Ham and Nelson
2016; Nill and Aalberts 2014). This covertness may be harm-
ful and unethical, as consumers are unaware of the persuasion
DEFINING ONLINE BEHAVIORAL ADVERTISING mechanisms that entail OBA; it has led to a call for
There are many definitions of OBA, which is also called transparency.
“online profiling” and “behavioral targeting” (Bennett 2011).
Examples include “adjusting advertisements to previous online
surfing behavior” (Smit, Van Noort, and Voorveld 2014, p. LITERATURE SEARCH
15), “a technology-driven advertising personalization method We performed a keyword search of the most important
that enables advertisers to deliver highly relevant ad messages electronic databases in advertising and communication sci-
to individuals” (Ham and Nelson 2016, p. 690), and “the prac- ence (i.e., PsycINFO, Web of Science, Communication and
tice of collecting data about an individual’s online activities Mass Media Complete, Academic Search Premier, database
for use in selecting which advertisement to display” of the World Advertising Research Center). The keywords
(McDonald and Cranor 2010, p. 2). These definitions and used were “online behavioral/behavioural advertising,”
others have two common features: (1) the monitoring or track- “online behavioral/behavioural targeting,” “customized/cus-
ing of consumers’ online behavior and (2) use of the collected tomised advertising,” “personalized/personalized (online)
data to individually target ads. Therefore, we define OBA as advertising,” and “online profiling.” The search period cov-
the practice of monitoring people’s online behavior and using ered all manuscripts available by the end of September 2016.
the collected information to show people individually targeted After identifying a study, we examined its references to find
advertisements. Online behavior can include web browsing further studies. In addition, we contacted several experts in
data, search histories, media consumption data (e.g., videos the field to inquire about other relevant manuscripts.
watched), app use data, purchases, click-through responses to We included studies that reported on empirical data while
ads, and communication content, such as what people write in leaving out nonempirical studies, such as legal studies. Fur-
e-mails (e.g., via Gmail) or post on social networking sites thermore, by definition, OBA involves tailoring advertising
(Zuiderveen Borgesius 2015a). based on online behavior. Therefore, we excluded studies that
OBA could be considered a type of personalized or custom- addressed personalized advertising based on personal data that
ized advertising—concepts which refer to tailoring advertising were not inferred from online behavior. In total, 32 manu-
to individuals. However, these concepts have a broader scope scripts fit the criteria, and the earliest study was published in
than OBA and could include advertising amended to personal 2008. Among these, 21 were from academic journals, six from
data that are not based on online behavior. OBA refers only to conference proceedings, one was a book chapter, and four
advertising that is based on people’s online behavior. were published online.
 ONLINE BEHAVIORAL ADVERTISING: A LITERATURE REVIEW AND RESEARCH AGENDA 365
Downloaded by [UVA Universiteitsbibliotheek SZ] at 07:59 01 August 2017

FIG. 1. Proposed framework of online behavioral advertising (OBA).

Based on these studies, we developed a framework that ADVERTISER-CONTROLLED FACTORS

identifies the factors that explain consumer responses to OBA
Ad Characteristics
and illustrates their interconnectedness (see Figure 1). The
framework focuses on relationships with empirical support. To Level of personalization. The data used to create OBA
develop the framework, we first identified all variables that vary widely. Because advertisers typically do not use all these
were studied with regard to OBA and grouped them into three data for one ad, the levels of personalization differ. We pro-
main factors based on the interactive advertising model (Rodg- pose that the level of personalization is based on (a) the types
ers and Thorson 2000). This model explains how consumers of personal data that are used to target the ad (e.g., browsing
perceive and process online ads and distinguishes three main data or search history) and (b) the amount of information that
types of factors: advertiser-controlled factors, consumer- is used (e.g., just one search term or a combination of brows-
controlled factors, and advertising outcomes. Within these ing data and search history). Several studies compared differ-
factors, we distinguished separate aspects. The advertiser- ent levels of personalization in OBA. The types of information
controlled factors include (a) the ad characteristics, or the used included age, gender, location (Aguirre et al. 2015), edu-
factors which are part of the ad itself and which can differ cation level (Tucker 2014), interests (Aguirre et al. 2015;
among different online behavioral ads, and (b) the forms of Tucker 2014), online shopping behavior (Bleier and Eisen-
transparency which advertisers use to communicate that an ad beiss 2015), and search history (Van Doorn and Hoekstra
is based on online behavior. These forms of transparency 2013). The researchers created various levels of personaliza-
involve information that often accompanies an ad (e.g., an tion by combining one or more types of information. Their
additional logo or privacy statement), which some self- findings suggest that the level of personalization influences
regulatory bodies require for OBA. consumer-related factors, such as feelings of intrusiveness
The consumer-controlled factors include (a) a cognitive (Van Doorn and Hoekstra 2013), feelings of vulnerability
aspect, including people’s knowledge and abilities with (Aguirre et al. 2015), the ad’s perceived usefulness, reactance,
respect to OBA; (b) an affective aspect, including people’s and privacy concerns (Bleier and Eisenbeiss 2015). The level
perceptions of OBA in general or of a specific ad; and (c) of personalization also influences OBA outcomes, such as
personal characteristics, such as a person’s age or desire click-through intentions and rates (Aguirre et al. 2015; Bleier
for privacy. Finally, the outcomes include consumers’ and Eisenbeiss 2015; Tucker 2014).
responses to OBA with respect to (a) the actual advertising Negative responses to higher levels of personalization cor-
effects, such as purchases and click-through rates, and (b) respond to choice theory, psychological ownership theory, and
the degree to which people accept or avoid OBA. Table 1 psychological reactance theory (Aguirre et al. 2015; Bleier
provides an overview of the literature addressing these and Eisenbeiss 2015; Tucker 2014). These theories propose
factors. that people want to restore their freedom of choice, control,
 366 S. C. BOERMAN ET AL.

TABLE 1
Summary of Framework-Related Literature
Factors References

Advertiser-controlled factors
Ad characteristics
Level of personalization Aguirre et al. 2015; Bleier and Eisenbeiss 2015; Tucker 2014; Van Doorn and
Hoekstra 2013
Accuracy Summers, Smith, and Reczek 2016
OBA transparency
Privacy statements and informed Marreiros et al. 2015
consent requests
Disclosure Aguirre et al. 2015; Jai, Burns, and King 2013; Leon, Cranshaw, et al. 2012;
Miyazaki 2008; Schumann, von Wangenheim, and Groene 2014; Ur et al. 2012;
Van Noort, Smit, and Voorveld 2013
Downloaded by [UVA Universiteitsbibliotheek SZ] at 07:59 01 August 2017

Consumer-controlled factors
Knowledge and abilities
OBA knowledge Marreiros et al. 2015; McDonald and Cranor 2010; Smit, Van Noort, and Voorveld
2014; Ur et al. 2012; Ham and Nelson 2016
(Knowledge about) Balebako et al. 2012; Cranor 2012; Leon, Ur, et al. 2012
self-protection
Perceptions*
Privacy concerns (dv, med) Aguirre et al. 2015; Ant
on, Earp, and Young 2010; Baek and Morimoto 2012; Bleier
Attitude toward OBA and Eisenbeiss 2015; Ham and Nelson 2016; Jai, Burns, and King 2013; Lambrecht
Feelings of vulnerability (med) and Tucker 2013; Lee et al. 2015; Marreiros et al. 2015; McDonald and Cranor
Perceived usefulness (med) 2010; Moore et al. 2015; Phelan, Lampe, and Resnick 2016; Schaub et al. 2016;
Reactance (med) Stanaland, Lwin, and Miyazaki 2011; Smit, Van Noort, and Voorveld 2014;
Trust in advertiser (dv, mod) Sutanto et al. 2013; Turow et al. 2009; Turow, Carpini, and Draper 2012; Ur et al.
Feelings of intrusiveness (med) 2012; Van Doorn and Hoekstra 2013; Yang 2013
Perceived personalization (iv)
Ad skepticism (iv)
Perceived risk (med)
Perceived fairness (med)
Consumer characteristics
Privacy concerns Ant

on, Earp, and Young 2010; Baek and Morimoto 2012; Lee et al. 2015; Smit, Van
Noort, and Voorveld 2014; Yang 2013
Desire for privacy Miyazaki 2008; Stanaland, Lwin, and Miyazaki 2011
Decision stage, needs Lambrecht and Tucker 2013; Van Doorn and Hoekstra 2013
Online experience Lee et al. 2015; Miyazaki 2008
Education Smit, Van Noort, and Voorveld 2014
Age Smit, Van Noort, and Voorveld 2014; Turow et al. 2009
Self-perceptions Summers, Smith, and Reczek 2016
Outcomes
Advertising effects
Click-through intention and Lambrecht and Tucker 2013; Van Doorn and Hoekstra 2013; Summers, Smith, and
behavior Reczek 2016
Purchase intention and behavior Aguirre et al. 2015; Bleier and Eisenbeiss 2015; Tucker 2014
Brand recall Van Noort, Smit, and Voorveld 2013
Perceived relevance of ad Van Noort, Smit, and Voorveld 2013
OBA acceptance and resistance
OBA acceptance Schumann, von Wangenheim, and Groene 2014
Advertising avoidance Baek and Morimoto 2012
Note. *The abbreviations between parentheses clarify the role of this factor in the studies (if applicable): dv D dependent variable, iv D
independent variable, med D mediator, mod D moderator. To avoid repetition, the References cell sums up all articles that studied one or more
of the perceptions.
 ONLINE BEHAVIORAL ADVERTISING: A LITERATURE REVIEW AND RESEARCH AGENDA 367

and ownership when they feel they are threatened. Highly per- understand the language of such statements (Jensen and Potts
sonalized ads lead people to perceive a loss of choice, control, 2004; Milne, Culnan, and Greene 2006). People tend to agree
or ownership, and thus cause negative feelings and responses. with almost all requests, or they simply ignore them (Mar-
Accuracy. Another key characteristic of OBA is accuracy. reiros et al. 2015; Zuiderveen Borgesius 2015b). Thus,
Based on self-perception theory, Summers, Smith, and Reczek informed consent requests seem to be a valuable approach to
(2016) propose that OBA can act as an implied social label. give people control, yet they fail to inform or empower people.
When consumers know an ad is based on their past online Disclosure. The online marketing industry has developed
behavior, they understand that the marketer has made infer- self-regulatory approaches to improve transparency that entail
ences about them. Thus, OBA provides an external characteri- explicit disclosure of data collection, usage, and distribution.
zation of the self, leading consumers to adjust their self- The current disclosure methods used by the industry involve
perceptions and draw on these perceptions to determine their icons, logos, and taglines. The Digital Advertising Alliance in
purchase behavior (Summers, Smith, and Reczek 2016). Inter- the United States and the European Interactive Digital Adver-
estingly, these effects seem to occur only when OBA is accu- tising Alliance have developed a standard icon that consists of
rately connected to past behavior. the letter i in a blue triangle. However, advertisers can use
other form of disclosure, such as pop-ups or text that explain
Downloaded by [UVA Universiteitsbibliotheek SZ] at 07:59 01 August 2017

OBA.
OBA Transparency Different academic studies have compared the effectiveness
Research and regulations also indicate the importance of of OBA icons. Leon, Cranshaw, et al. (2012) found that disclo-
transparency about OBA. Consumers want openness and to be sures are rarely noticed: Only one-quarter of the respondents
informed about the collection, usage, and sharing of personal remembered OBA disclosure icons (the standard icon and an
data (Gomez, Pinnick, and Soltani 2009; Turow et al. 2009). “asterisk man” icon), and only 12% remembered seeing a tag-
Privacy laws require companies to be transparent about their line (e.g., “Why did I get this ad?” or “AdChoices”) and cor-
data processing practices (EU Data Protection Directive 1995; rectly selected the tagline they had seen from a list. In addition,
EU General Data Protection Regulation 2016). Similarly, the none of the taglines were understood to be links to pages where
U.S. FTC (2012) calls for transparency regarding OBA. you can make choices about OBA, nor did they increase knowl-
Privacy Statements and Informed Consent Requests. edge about OBA. Other studies show that consumers are unfa-
Companies typically publish privacy statements on their web- miliar with the icons (Ur et al. 2012; Van Noort, Smit, and
sites to comply with transparency requirements. In addition, Voorveld 2013), do not understand their purpose (Ur et al.
some laws aim to improve transparency by requiring compa- 2012), and rarely notice them (Van Noort, Smit, and Voorveld
nies to obtain consent before using OBA. The FTC (2012) also 2013). However, the standard icon could effectively increase
emphasizes that companies should offer consumers choices OBA awareness and understanding when accompanied by an
regarding OBA. According to the privacy guidelines from the explanatory label stating, “This ad is based on your surfing
Organisation for Economic Co-Operation and Development behavior” (Van Noort, Smit, and Voorveld 2013).
(OECD 2013), personal data should be obtained where appro- Consumers do seem to appreciate companies’ transparency
priate, with the knowledge or consent of the data subject. One initiatives (Van Noort, Smit, and Voorveld 2013). When firms
of the goals in these efforts is empowering the consumer. Pri- do not openly state that they use personal data to personalize
vacy laws typically aim to enable consumers to make informed ads and then present highly personalized ads, consumers feel
decisions about privacy and personal data. For instance, some more vulnerable (Aguirre et al. 2015). However, when compa-
consumers might like OBA and thus allow companies to track nies are open about data collection, it does not affect perceived
them, while others may prefer more privacy and decline vulnerability. In addition, Miyazaki (2008) found that explic-
tracking. itly disclosing the usage of cookies in a privacy statement and
A privacy statement is a document on a website that disclo- a pop-up can increase consumers’ trust toward the website and
ses which personal data are collected through the website, as their intentions to use and recommend it. Hence, advertisers
well as how and why. In theory, privacy statements should benefit from transparency about OBA.
help reduce the information asymmetry between companies The effects of transparency on consumer trust can be
and consumers through companies disclosing information to explained by social contract theory (Miyazaki 2008) and expec-
consumers (McDonald and Cranor 2008). Although inclusion tancy violation theory (Moore et al. 2015). According to social
of privacy statements disclosing the usage of cookies has contract theory, advertisers form an implied social contract with
increased (Miyazaki 2008), such statements are seldom read consumers by explicitly disclosing the collection and use of per-
and thus fail to inform consumers (Cranor 2003; McDonald sonal information. Under such an implied contract, consumers
and Cranor 2008; Milne and Culnan 2004). It would take a expect advertisers to collect and care for their personal informa-
person approximately 201 hours per year to read all privacy tion in a responsible manner. When companies do not disclose
statements for the websites he or she visits (McDonald and the collection and use of personal information, or do not use the
Cranor 2008); in addition, people are highly unlikely to information responsibly, they are violating this contract. In
 368 S. C. BOERMAN ET AL.

addition, expectancy violation theory (see Moore et al. 2015) do not know why they do this (McDonald and Cranor 2010).
suggests that a violation of personal space will drive subsequent Protective behavior does seem to depend on consumer charac-
reactions. Thus, when an advertiser collects and uses informa- teristics: the more concerned people are about their privacy,
tion without disclosing it and without consent, this may lead to a the more they try to protect their online privacy (Smit, Van
violation of the social contract, a violation of personal space, Noort, and Voorveld 2014).
and, as a result, lowers trust. Moreover, not all available tools and tactics are effective
for protecting privacy. Research has shown that certain tools
that block third-party cookies can effectively limit OBA
CONSUMER-CONTROLLED FACTORS (Balebako et al. 2012). Opt-out options limit receiving
behaviorally targeted ads but may not limit being tracked. The
OBA Knowledge and Awareness. Several academic stud- “do-not-track” option in browsers limits OBA only slightly.
ies show that consumers have little knowledge about OBA and Furthermore, consumers do not seem to understand the
hold misconceptions (Marreiros et al. 2015; McDonald and available tools and thus have difficulties protecting their online
Cranor 2010; Smit, Van Noort, and Voorveld 2014). More- privacy (Cranor 2012; Leon, Ur, et al. 2012).
over, people have little insight into the extent to which their
Downloaded by [UVA Universiteitsbibliotheek SZ] at 07:59 01 August 2017

online behavior is tracked (Ur et al. 2012). Interestingly, the

perception of having knowledge about OBA makes people Consumer Perceptions
more likely to perceive the effects of OBA as larger on others Many empirical studies on OBA consider consumer percep-
than on themselves (Ham and Nelson 2016). Even more confu- tions either as a consumer characteristic in a survey or as mod-
sion arises in regard to legal protections: A substantial major- erator, mediator, or dependent variable in experimental
ity of Americans have false beliefs about companies’ rights to research. Overall, consumers’ perceptions of OBA appear to
share and sell information about them online (McDonald and be mixed. Some see the benefits of targeted ads (McDonald
Cranor 2010; Turow et al. 2009). These findings suggest there and Cranor 2010; Ur et al. 2012), but the majority seem to be
is information asymmetry: Companies know much about con- skeptical toward OBA and find it invasive and “creepy” (Smit,
sumers, yet consumers know little about what happens to their Van Noort, and Voorveld 2014; Ur et al. 2012). Most adults in
personal data. It seems nearly impossible for people to deter- the United States do not want advertising to be tailored to their
mine which companies collect which personal data online and personal information (Turow, Carpini, and Draper 2012). Con-
what happens to the data. sumers believe that invasive tactics, such as using and gather-
These findings indicate that consumers’ mental models (i.e., ing personal data, tracking, and invading a consumer’s
their beliefs about how a system works, interacts, or behaves; personal space, can be considered “creepy marketing” (Moore
see McDonald and Cranor 2010) and persuasion knowledge et al. 2015). These negative perceptions and privacy concerns
(i.e., consumers’ knowledge and beliefs about persuasive tac- can be explained by social presence theory. Social presence
tics; see Baek and Morimoto 2012; Van Noort, Smit, and describes the feeling of being with another in mediated com-
Voorveld 2013; Ham and Nelson 2016) are rarely well devel- munication (see Phelan, Lampe, and Resnick 2016). When a
oped in the context of OBA. In addition, there seems to be an computer collects your data, it generates the same negative
important relationship between persuasion knowledge and feelings as when another person looks over your shoulder as
people’s third-person perceptions about OBA. The more peo- you browse (Phelan, Lampe, and Resnick 2016).
ple think they know about how OBA works (i.e., subjective Over the past decade, consumers have become more
persuasion knowledge), the more they tend to overestimate the concerned about OBA practices and especially about their
effects of OBA on others and underestimate its effects on privacy (e.g., Ant
on, Earp, and Young 2010; McDonald
themselves (Ham and Nelson 2016). This can be a problem and Cranor 2010). Researchers found signs for a chilling
because incorrect mental models, low persuasion knowledge, effect: People report that they change their online behavior
and an underestimation of OBA effects may undermine careful when they know their data are being collected (McDonald
and educated decision making. and Cranor 2010). Perceptions of OBA also depend on
(Knowledge about) Self-Protection. Consumers’ lack of consumer characteristics such as age. Younger people are
knowledge about OBA impedes them from having control less likely to oppose OBA compared to older people,
over their personal data (Cranor 2012). This is especially inter- although the majority of young people do not want OBA
esting because consumers want to have control over the collec- (Turow et al. 2009).
tion and use of their personal data (Gomez, Pinnick, and The notion of privacy calculus is often used to describe the
Soltani 2009; Turow, Carpini, and Draper 2012). Research has process in which consumers assess the benefits and risks of
shown that a minority of consumers do try to control their per- OBA (Jai, Burns, and King 2013; Phelan, Lampe, and Resnick
sonal data by deleting cookies, by not letting cookies save to 2016; Schumann, von Wangenheim, and Groene 2014). Pri-
the hard drive, and by using software that deletes cookies. vacy calculus is rooted in theories such as social exchange the-
However, despite taking such actions, it appears that people ory (Schumann, von Wangenheim, and Groene 2014) and the
 ONLINE BEHAVIORAL ADVERTISING: A LITERATURE REVIEW AND RESEARCH AGENDA 369

acquisition-transaction utility theory (Baek and Morimoto construed preferences (construal level theory; Lambrecht and
2012). The social exchange theory stems from psychology and Tucker 2013). Furthermore, responses to OBA seem to be
proposes that people evaluate social exchanges in terms of related to age, education, and online experience (Lee et al.
costs and rewards. People are supposed to alter their behavior 2015; Miyazaki 2008; Smit, Van Noort, and Voorveld 2014;
according to their evaluation and are expected to participate in Turow et al. 2009).
social exchanges only when the rewards outweigh the costs
(Schumann, von Wangenheim, and Groene 2014). Acquisi-
tion-transaction utility theory is often used to understand ethi- OBA OUTCOMES
cal issues in marketing and suggests that the probability of Several academic studies show that the outcomes of OBA
consumers purchasing a product or service depends on the per- are determined by factors controlled by the advertisers and by
ceived benefits compared with the perceived costs (Baek and the consumer. Outcome measures have included advertising
Morimoto 2012). Based on these theories, consumers should effects (i.e., click-through intentions and rates, actual pur-
accept OBA only if the benefits (e.g., personal relevance) out- chases, and purchase intentions) and measures of OBA accep-
weigh the costs or risks (e.g., privacy invasions). tance and avoidance. Overall, the findings are much more
The information boundary theory (Sutanto et al. 2013) nuanced than the industry’s promise that OBA boosts ad
Downloaded by [UVA Universiteitsbibliotheek SZ] at 07:59 01 August 2017

provides some insight into the actual weight that consumers effects.
give to the benefits and risks of OBA. This theory suggests
people find the collection and use of personal information
intrusive and thus perceive it as a risk or cost that does not out- Advertising Effects
weigh the possible benefits of OBA when people consider
using this information as crossing a boundary. When a person Click-through intention and click-through rates
considers the collected information as harmful or too uncom- Several studies demonstrated that the level of personaliza-
fortable, the costs do not outweigh the benefits of OBA. tion in OBA influences click-through intentions and click-
In line with these theories, research has shown that privacy through rates. Tucker (2014) found that Facebook ads target-
concerns and trust play important roles in consumer accep- ing a person’s interests (e.g., a celebrity of whom a person is a
tance and the effectiveness of OBA. For instance, more trusted fan) led to higher click-through rates than ads targeting back-
retailers can increase the perceived usefulness of their ads by ground characteristics (i.e., the college a person is attending).
developing ads that reflect consumers’ interests in a complete In addition, Aguirre et al. (2015) showed that moderately per-
way (Bleier and Eisenbeiss 2015). Trust can be enhanced by sonalized Facebook ads (based on interest in a subject)
including a “privacy trustmark” (i.e., a symbol explaining that increased click-through rates compared to nonpersonalized
the website is involved in a program that protects consumers ads, whereas highly personalized ads (based on interest in a
privacy), which positively affects consumers’ perceptions of subject, age, gender, and location) decreased click-through
the trustworthiness of the advertiser, lowers privacy concerns rates. Bleier and Eisenbeiss (2015) found that highly personal-
about the advertiser, and leads to more positive behavior inten- ized banner ads (showing items consumers placed in their vir-
tions (Stanaland, Lwin, and Miyazaki 2011). Furthermore, tual shopping cart during a recent shopping session) increased
Sutanto and colleagues (2013) note that if consumers’ personal click-through rates compared to ads with a lower level of
data are not transmitted to third parties, consumers are less personalization (showing items that consumers viewed during
concerned about their privacy and more satisfied with the con- a shopping session). This effect occurred only when the ad
tent of a smartphone app about products. concerned a trusted retailer.
Furthermore, several studies indicate that transparency and
consumer awareness of OBA alter consumers’ responses to
Consumer Characteristics online behavioral ads. For instance, when companies overtly
Responses to OBA also differ among consumers, and indi- inform people about the collection and use of data to personal-
vidual levels of privacy concerns are especially important. ize ads, OBA increases click-through rates. However, when
Multiple studies suggest that people with low levels of privacy companies covertly collect information, click-through rates
concerns or less desire for privacy tend to be more positive are not influenced or are even reduced (Aguirre et al. 2015).
toward OBA (e.g., Baek and Morimoto 2012; Miyazaki 2008; Moreover, the difference in click-through intentions between
Smit, Van Noort, and Voorveld 2014; Stanaland, Lwin, and overt and covert ads disappeared when an OBA icon was
Miyazaki 2011). Moreover, the level of privacy concern included (Aguirre et al. 2015). Likewise, an OBA icon
appears to moderate the effects of OBA on consumers’ adver- improves brand recall and the perceived relevance of the
tising responses (e.g., Lee et al. 2015; Miyazaki 2008). advertised brand and the online ad (Van Noort, Smit, and
The effects of OBA on purchase intentions and behavior are Voorveld 2013). Hence, people seem to appreciate company
more positive when the ad fits consumers’ needs (Van Doorn transparency, and an icon can function as a cue to trust the
and Hoekstra 2013) and when consumers have narrowly advertiser and even positively affect advertising outcomes.
 370 S. C. BOERMAN ET AL.

Thus, this self-regulatory approach does benefit advertisers but to OBA, which is in accordance with the stimulus-organism-
does not help consumers make informed decisions. response model. Van Doorn and Hoekstra (2013) conclude that
Research also specifies various moderators and mediators more personalization increases feelings of intrusiveness and thus
that influence the effects of OBA on click-through rates and negatively affects purchase intentions. These negative effects
intentions. Aguirre et al. (2015) demonstrated that the effects cannot be offset by offering a discount, but they can be partly
of OBA were mediated by the consumers’ experience of vul- mitigated by ensuring the ad fits the consumer’s current needs.
nerability. Overall, personalization appears to increase click- However, an ad with high fit also increases the perceived intru-
through rates and intentions, but only when consumers know siveness of the ad. Moreover, although privacy concerns did not
that data are collected. When consumers are unaware that data alter the effects of OBA, the study indicated that people with
are collected, they feel more vulnerable when confronted with high concerns had lower intentions to purchase.
personalized advertising, which decreases their intention to Furthermore, Summers, Smith, and Reczek (2016) emphasize
click on an ad (Aguirre et al. 2015). the importance of consumers’ self-perceptions as a mediator of
This idea of vulnerability is in line with the finding that trust the effects of OBA on purchase intentions. They found that
in a retailer is a major determinant, and that the perceived use- OBA can influence consumers’ self-perceptions as it reflects
fulness of the ad, the reactance, and privacy concerns are impor- their past online behavior. When the ad accurately fits their
Downloaded by [UVA Universiteitsbibliotheek SZ] at 07:59 01 August 2017

tant underlying mechanisms for the effects of OBA on click- behavior and perceptions, it can increase their purchase behavior.
through intentions (Bleier and Eisenbeiss 2015). For banner ads
from trusted retailers, click-through intentions appear particu-
larly high when ads reflect a combination of high depth (using OBA Acceptance and Resistance
items previously placed in an online shopping cart) and narrow In addition to these advertising effects, empirical
breadth (only one out of three items shown in the ad personal- research has also examined which factors can explain OBA
ized) of personalization. People perceive such ads to be more acceptance and avoidance. Baek and Morimoto (2012)
useful and do not elicit increased reactance or privacy concerns found that privacy concerns and ad irritation both increase
compared to low-depth ads. For ads from less-trusted retailers, a ad skepticism, which consequently leads to more avoidance
higher depth of personalization decreased click-through inten- of OBA. In addition, the more consumers feel the ad is
tions, irrespective of the breadth of personalization. personalized for them (i.e., perceived personalization), the
The lower click-through intention can be explained by peo- less they avoid the ad.
ple experiencing lower usefulness, more reactance, and more Transparency about the reason why a company collects data
privacy concerns. These findings are in line with the stimulus- also influences responses to ads. Schumann, von Wangenheim,
organism-response model, which posits that stimuli influence and Groene (2014) compared people’s responses to two argu-
individuals’ cognitive and affective responses, which then ments for data collection: a reciprocity argument which argues
translate into specific behavior (see Bleier and Eisenbeiss that the service on the website is for free in return for personal
2015; Jai, Burns, and King 2013). Indeed, OBA seems to first data, or a relevance argument stating data collection is neces-
trigger affective responses, such as feelings of vulnerability, sary to make advertisements more personally relevant. The
reactance, and privacy concerns, which consequently affect researchers found that the reciprocity argument increased con-
behavior (e.g., lower intentions to click on an ad). sumers’ acceptance of OBA compared to the relevance argu-
ment, and consumers were more likely to opt in and disclose
personal data for OBA purposes. This finding suggests that
Purchases and purchase intention consumers believe that receiving web services for “free” in
Research has also examined the effects of OBA on pur- return for use of their personal data is an acceptable trade-off.
chase intentions and actual purchases. Lambrecht and People’s responses to data collection depend on what hap-
Tucker (2013) compared the effects on purchase intentions pens to the data. Jai, Burns, and King (2013) showed that tell-
and actual purchases between OBA showing an image of ing respondents that the website shared their personal and
previously browsed hotels and random generic ads for a website navigation data with third-party companies increased
travel firm. They found that the effects of the different ads the perceived risk and unfairness, leading to lower repurchase
depended on the decision stage of the consumer, with intentions compared to telling them their data would be shared
OBA being more effective when consumers had narrowly only internally within the corporate family.
construed preferences and thus had a greater focus on spe-
cific and detailed information. When preferences were still
broad and people were in the early stages of a purchase THEORETICAL POSITIONING OF OBA
decision, generic ads led to a higher likelihood of pur- We reviewed the theories used to study OBA to gain an
chases (Lambrecht and Tucker 2013). understanding of its theoretical positioning. Figure 2 shows an
As for click-through behavior, there also seem be important overview of theories used to explain the effects of different
mediating variables that influence purchase behavior in response factors in our framework and the responses to them. This
 ONLINE BEHAVIORAL ADVERTISING: A LITERATURE REVIEW AND RESEARCH AGENDA 371
Downloaded by [UVA Universiteitsbibliotheek SZ] at 07:59 01 August 2017

FIG. 2. Overview of theoretical position of online behavioral advertising (OBA).

model shows that the theoretical background of the research Wangenheim, and Groene 2014), acquisition-transaction util-
regarding OBA is very fragmented. Almost half of the studies ity theory (Baek and Morimoto 2012), and information bound-
(n D 15) did not refer to specific theories. The studies that do ary theory (Sutanto et al. 2013).
use them employ a variety of theories from different disci- Aside from the theories that can be attributed to one of the
plines, such as social and cognitive psychology, communica- factors in our framework, some general communication and
tion, and marketing. There is certainly no single overarching processing theories are also applied to OBA: human-interac-
or reoccurring theory that is used to describe and explain tion theory (Ahrens and Coyle 2011), uses and gratifications
responses to OBA. theory (Sutanto et al. 2013), overload theory (Schumann, von
The theories cited more than twice are the persuasion Wangenheim, and Groene 2014), and dual process theory
knowledge model (Baek and Morimoto 2012; Van Noort, (Phelan, Lampe, and Resnick 2016). These theories are helpful
Smit, and Voorveld 2013; Ham and Nelson 2016), psycho- in explaining how people process and respond to computer-
logical reactance theory (Aguirre et al. 2015; Baek and mediated communication and specific messages, but they do
Morimoto 2012; Bleier and Eisenbeiss 2015; Tucker 2014), not seem to be specific to the concept of OBA or our
and social contract theory (Jai, Burns, and King 2013; framework.
Miyazaki 2008; Yang 2013). These three theories are all
used to explain why people may have negative feelings
toward OBA and resist it. When people develop persuasion RESEARCH AGENDA
knowledge about tactics used in OBA, they can become
more skeptical of them (Baek and Morimoto 2012). In Theoretical Advancement
addition, highly personalized ads can threaten consumers’ Our review of the theoretical positioning of empirical
perceived ability to avoid being closely observed by firms, research regarding OBA demonstrates that the field is frag-
causing reactance (Bleier and Eisenbeiss 2015). And when mented and lacks a solid theoretical basis. To advance the lit-
people feel that advertisers violate a social contract by erature on OBA, it is important to develop more conceptual
wrongly collecting and using their information, it can also coherence between the different theoretical ideas that focus on
cause distrust (Miyazaki 2008). the roles of the advertiser and consumer variables in explain-
Although privacy concerns are an important aspect of OBA, ing consumer responses. Valkenburg and Peter (2013) pro-
many (n D 13) of the studies that discuss people’s privacy con- posed that connecting different conceptual approaches could
cerns never relate these concerns to any kind of theoretical help shed light on media effects. Moreover, they suggested
notions. When they do, privacy concerns are often placed that research should focus on specific models that combine
within a debate about the privacy calculus, which is directly moderating and mediating processes (Valkenburg and Peter
related to social exchange theory (Schumann, von 2013). Thus far, most OBA studies have focused on one
 372 S. C. BOERMAN ET AL.

perspective and singled out specific moderators and mediators. Understanding Acceptance of and Resistance Toward
However, the combinations of certain moderating and mediat- OBA
ing processes are expected to play crucial roles. Prior empirical research shows that consumers avoid and
Therefore, we urge researchers to take a more integrative dislike some ads but accept others. However, it is currently not
look, combine insights from different perspectives, and look at well understood why people reject OBA (Turow et al. 2009).
the reinforcing role of outcomes in affecting future use or Consumers’ most important concerns seem to be privacy and a
exposure to ads. For instance, choice theory, psychological lack of control over personal data (McDonald and Cranor
ownership theory, and psychological reactance theory 2010; Smit, Van Noort, and Voorveld 2014; Turow et al.
(Aguirre et al. 2015; Bleier and Eisenbeiss 2015; Tucker 2009; Ur et al. 2012). People might have a “general antago-
2014) suggest that people perceive advertising as invasive nism to being followed without knowing exactly how or with
when an ad is too personal (which is also a boundary condition what effects” (Turow et al. 2009, p. 4). For instance, Yang
as it depends on the characteristics of the ad). When people (2013) found that consumers who are more concerned about
perceive a lack of freedom of choice, it could lead to height- privacy are less likely to trust online companies to protect their
ened irritation and skepticism toward OBA, as well as more privacy. In addition, people with low levels of privacy con-
concerns about privacy (Baek and Morimoto 2012). This cerns who are willing to share information respond more posi-
Downloaded by [UVA Universiteitsbibliotheek SZ] at 07:59 01 August 2017

might ultimately contribute to a reinforcing process in which tively to OBA (Lee et al. 2015). However, there appears to be
responses to OBA also affect future media use for shorter or a privacy paradox: People say they care about privacy and do
longer periods of time (transactional media effect; Valkenburg not intend to share personal data, but in reality they provide
and Peter 2013). Such reinforcing processes are currently their data in exchange for small benefits or conveniences
neglected in the literature, which mainly focuses on immediate (Norberg, Horne, and Horne 2007). In other words, although
responses while ignoring future consequences. people say they reject OBA, they take few measures to protect
In addition, one of the key features of OBA is that it is often their data from it. It seems that people accept privacy risks but
covert. The perceived costs and benefits of OBA (e.g., the pri- still express concerns when prompted (Phelan, Lampe, and
vacy calculus) may therefore depend on peoples’ knowledge Resnick 2016). Interestingly, consumers even seem to seek out
about the practice of OBA. Those with more knowledge might more relevant advertising messages that fulfill their specific
perceive the cost and benefits differently and might believe that needs (Kumar and Gupta 2016). More work is needed to
the negative consequences are more severe. Our framework understand why consumers like or dislike OBA, specifically
reveals an absence of research on the roles of consumer knowl- because consumers’ negative perceptions of OBA are not in
edge and abilities in OBA perceptions and responses. Combina- line with their behavior to protect themselves or their
tions of theories, such as the privacy calculus notion and the expectations.
persuasion knowledge model, could be useful for attaining Personalization has an important influence on consumers’
deeper understanding of OBA transparency, as well as the ante- responses to OBA. However, studies examining this factor are
cedents and effects of consumer knowledge and abilities. somewhat limited with respect to the varieties in the levels of
Other theories could also be added to the theoretical personalization. We know that consumers may consider tai-
positioning of consumer reactions to OBA. For instance, lored ads to be useful, but they may consider the use of more
the notion of present bias describes tendencies to choose sensitive information to be creepy and inappropriate. Such
immediate gratification and disregard future costs or disad- feelings can lead to reactance and privacy concerns (e.g.,
vantages (Acquisti and Grossklags 2007; O’Donoghue and Bleier and Eisenbeiss 2015; Smit, Van Noort, and Voorveld
Rabin 1999). Online, people often want to read an article, 2014; Ur et al. 2012). However, it is currently not well under-
watch a video, or purchase a product immediately, thus stood what levels of personalization consumers find acceptable
skipping privacy statements and accepting informed con- and what they consider creepy. There is ample room for
sent requests without thinking about long-term consequen- improvement in the understanding of how consumers respond
ces. Despite its relevance, there is a lack of research that to the usage of various types and amounts of personal data
looks at whether the idea of present bias can explain peo- with respect to resistance, acceptance, and the advertising out-
ple’s behavior in response to OBA. The circumstances in comes of OBA. Further research should aim to understand and
which immediate gratification is sought and the motiva- identify the tipping point, including the point where consumers
tions that play a role warrant investigation. Such motiva- feel that data collection for OBA becomes too invasive, what
tions function as boundary conditions. For instance, when they consider acceptable or unacceptable, in what context, and
people need information immediately, certain privacy con- how this affects their responses to different levels of personali-
cerns might be overruled, but perhaps only when the web- zation in OBA.
site is trusted. Thus, we believe that the field could benefit Research suggests it is important to know who is being tar-
not only from a more integrative perspective but also from geted, as there seem to be individual differences in responses
the inclusion of such theories. to OBA. Although our framework delineates some important
 ONLINE BEHAVIORAL ADVERTISING: A LITERATURE REVIEW AND RESEARCH AGENDA 373

consumer characteristics, theory suggests there are more rele- their online privacy seems to be low. Even when people under-
vant characteristics to be examined. For instance, the interac- stand OBA, they still may not be capable of protecting their
tive advertising model (Rodgers and Thorson 2000) indicates privacy. Many different tools can be used to mitigate data col-
that consumer responses to interactive advertising depend on lection, but not all of these tools are equally effective (Leon,
their motives for being online and whether those motives are Ur, et al. 2012). Using ineffective tools would therefore cause
goal oriented. When an ad addresses such motives (e.g., infor- a false sense of safety. Further research could address this
mation seeking or entertainment), it is believed to increase the problem and should aim to develop transparency approaches
attention, memory, and attitudes toward the ad, which conse- and education that could encourage and help consumers to pro-
quently enhance outcomes such as clicks and purchases. This tect their online privacy. Such research could also help identify
seems particularly applicable to OBA because it is based on the extent to which consumers might need additional (legal)
online behavior, which is directly connected to online motives. protection.
Another step is to enhance the current approaches of trans-
parency (e.g., informed consent requests, privacy statements,
Empowering the Consumer icons). People often do not notice most approaches to achieve
In general, the research suggests that consumers lack rele- transparency, or they ignore them. Therefore, transparency
Downloaded by [UVA Universiteitsbibliotheek SZ] at 07:59 01 August 2017

vant knowledge about OBA but do have significant concerns approaches often fail to increase knowledge and awareness of
about the collection and use of personal information online. OBA and encourage self-protection. One such example is the
We believe there is both a theoretically and socially relevant OBA icon. Adding some descriptive information to the icon
gap in our understanding of how we can improve consumers’ might help consumers understand that an ad is based on their
knowledge and empower them to take actions when they think personal online behavior (Leon, Cranshaw, et al. 2012; Van
it is necessary. Our framework shows there are two important Noort, Smit, and Voorveld 2013). Moreover, transparency can
gaps with respect to consumer knowledge: There is an absence positively influence consumers’ perceptions of OBA and ad
of research that investigates how OBA characteristics could effectiveness and thus mostly benefits advertisers. More inves-
influence consumer knowledge and abilities and how knowl- tigation is needed to determine which transparency approach
edge would affect OBA outcomes. could be effective in truly increasing OBA knowledge.
More important, further research is needed to gain insights
into how we can educate people about OBA and empower
them to protect their online privacy. The research has pointed Novel Methodological Approaches
out that consumers might not be able to protect themselves Researchers have used various methods to understand the
mainly because they do not have the knowledge to assess effects and responses to OBA. The most applied method is
whether their protective behavior is effective. Using ineffec- experimental research, often involving scenario-based
tive tools might lead to a false sense of safety. However, con- approaches (e.g., Jai, Burns, and King 2013; Van Doorn and
sumers are generally positive toward the notion of protecting Hoekstra 2013; Van Noort, Smit, and Voorveld 2013). Field
their privacy (Smit, Van Noort, and Voorveld 2014), such as experiments are also common, and some are combined with
by clearing their browsing histories or installing ad blockers. scenario-based experiments (e.g., Bleier and Eisenbeiss 2015;
In addition, Schaub et al. (2016) found that popular extensions Sutanto et al. 2013; Tucker 2014). In addition, surveys are
(e.g., Ghostery, DoNotTrackMe, and Disconnect) can increase often used to examine users’ OBA knowledge and perceptions
consumers’ privacy awareness. (e.g., Baek and Morimoto 2012; McDonald and Cranor 2010;
We thus see an important gap in the literature with respect Turow et al. 2009). Content analysis (Ahrens and Coyle 2011)
to the extent to which people engage in self-protection behav- or qualitative data in the form of interviews (Leon, Ur, et al.
ior to circumvent websites’ online data collection, as well as 2012; McDonald and Cranor 2010; Ur et al. 2012) and focus
how we can encourage self-protection. Future research may groups (Marreiros et al. 2015) are rarely used.
draw on theoretical models from health communication, such Despite the diversity in methods, there is still much to gain.
as protection motivation theory (Rogers 1975) and the Examining OBA empirically is a challenge. It is difficult to
extended parallel processing model (Witte 1994). These mod- measure consumers’ exposure to OBA and to examine its con-
els posit that people’s motivations to protect themselves from sequences. The field could thus benefit from methodological
a specific threat depend on the perceived threat (based on the innovations. For instance, research is often constrained to sce-
perceived severity of the threat and one’s own susceptibility) nario-based experiments, as it is often difficult to manipulate
and the perceived efficacy of dealing with this threat (based on behaviorally targeted ads for participants because such ads are
the efficacy of the response and self-efficacy). When perceived based on personal online behavior. Because this information is
threat and efficacy are high, people are motivated to protect specific for each individual and difficult for researchers to
themselves and adapt their behavior. obtain, it is challenging to develop behaviorally targeted ads
Although consumers do seem to understand the threat of in an experimental design. Just like advertisers, a challenge for
online data collection and OBA, their efficacy in protecting academic researchers is to determine ways to use big data in
 374 S. C. BOERMAN ET AL.

the context of OBA and to combine different data sources rules are needed for OBA. There may be OBA practices that
(Kumar and Gupta 2016). For instance, tracking consumers’ society should not accept, regardless of whether consumers
browsing behavior could allow researchers to observe how consent to the practices. Examples include tracking on web-
consumers are exposed to OBA and how they respond to it. sites aimed at children and the use of OBA data for online
Such an approach could identify when, for whom, and in what price discrimination.
situations OBA is effective.
Furthermore, we believe that research into OBA could ben-
efit from implicit and unobtrusive measures to gain more Implications for Advertisers
insights into how consumers process and respond to it. For The advertising industry claims that OBA is much more
instance, eye-tracking research into the effects of ads that are effective than nontargeted ads (Beales 2010; Chen and Stal-
personalized based on demographic variables (e.g., name or laert 2014). However, our framework reveals that the effects
gender) showed that personalized ads attract more attention are more nuanced. OBA can increase click-through rates and
than nonpersonalized ads (Bang and Wojdynski 2016). For purchases, but these effects depend on factors controlled by
now, we have no insights into whether behaviorally targeted the advertiser (e.g., information used to personalize the ad and
ads also attract more visual attention. Such insights would be advertiser transparency) and factors that are related to the con-
Downloaded by [UVA Universiteitsbibliotheek SZ] at 07:59 01 August 2017

relevant information, as people’s attention to ads targeted on sumer (e.g., trust in the advertiser, perceived usefulness of the
the basis of their personal data could explain their responses to ad, feelings of intrusiveness, privacy concerns). Our frame-
the ad. work could help predict the outcomes of OBA.
Finally, OBA research lacks a longitudinal focus, making it Advertisers should consider the level of personalization.
difficult to observe developments over several years. Hence, Ads perceived to be too personal can seem intrusive and
we do not know how consumers’ knowledge, perceptions, and lower click-through rates and purchases. In addition, adver-
responses toward OBA change over time. Because technology tisers should be transparent. Overtly informing people
develops quickly and consumers’ knowledge and attitudes about the collection and use of data to personalize ads can
toward OBA might change, research combining panel studies benefit the response and outcomes of OBA. Furthermore,
and longitudinal (big) data could offer important new insights. OBA may not be beneficial in every situation or audience.
OBA seems to have the most positive outcomes for people
who are younger, have high levels of online experience,
IMPLICATIONS have low levels of privacy concerns, and have narrowly
construed preferences.
Implications for Privacy Law in the Area of OBA
Because of OBA’s privacy implications, policymakers
around the world have taken an interest in OBA (e.g., Article OBA AND THE FUTURE
29 Data Protection Working Party 2010; FTC 2012; Hong Although our research agenda is not exhaustive, it
Kong Office of the Privacy Commissioner for Personal Data shows important gaps in the literature and fruitful areas for
2014; Office of the Australian Privacy Commissioner 2011). further work. Personalized and targeted advertising is seen
Data privacy laws vary from country to country. However, as the future of advertising (Kumar and Gupta 2016;
despite the differences, many national data privacy laws also Schultz 2016; Rust 2016). OBA is still far from mature,
have common features (Bygrave 2014). Many privacy laws and it could be seen as an early example of ambient intelli-
aim to empower the consumer by requiring companies to offer gence—technology that senses and anticipates people’s
transparency and choices regarding data use. Our review of behavior to adapt the environment to their inferred needs
academic empirical studies shows that consumers understand (Hildebrandt 2010). Currently, behavioral targeting mostly
little about OBA and the related data use, and current transpar- occurs when using computers or smartphones, but the bor-
ency approaches are not very effective in increasing under- ders between offline and online are fading. Phrases such as
standing. Consumers who do not understand how data are the Internet of things, ubiquitous computing (Weiser 1993),
used for OBA cannot make meaningful privacy decisions. and ambient intelligence (Aarts and Wichert 2009) have
To defend privacy, we propose that policymakers should been used to describe such developments. If objects are
not merely aim for consumer empowerment but also for pro- connected to the Internet, companies could use the data
tection. Most privacy laws have elements that aim to protect collected through those objects for OBA. Google (2013)
consumers. For instance, many laws require companies to predicts, “A few years from now, we and other companies
secure the data they collect against data breaches, and in many could be serving ads and other content on refrigerators, car
countries the law has stricter rules for certain types of sensitive dashboards, thermostats, glasses, and watches, to name just
data, such as health-related data (EU Data Protection Directive a few possibilities.” These developments have important
1995; EU General Data Protection Regulation 2016; OECD implications for advertisers, consumers, scholars, and pub-
2013). However, to defend privacy, perhaps more and stricter lic policy, and they open up a whole new field of research.
 ONLINE BEHAVIORAL ADVERTISING: A LITERATURE REVIEW AND RESEARCH AGENDA 375

ACKNOWLEDGMENTS Individuals with Regard to the Processing of Personal Data and on the Free
Movement of Such Data,” Official Journal of the European Communities,
This project is supported by the Personalised Communica-
38 (281), 31–50, [Link]
tion Project at the University of Amsterdam ([Link] LEX:31995L0046.
[Link]/). The authors would like to thank EU General Data Protection Regulation (2016), “Regulation (EU) 2016/679 of
Prof. Dr. Natali Helberger and Prof. Dr. Claes de Vreese for the European Parliament and of the Council of 27 April 2016 on the Pro-
their valuable comments on an earlier version of this article. tection of Natural Persons with Regard to the Processing of Personal Data
and on the Free Movement of such Data, and Repealing Directive 95/46/
EC (General Data Protection Regulation),” Official Journal of the Euro-
pean Union, 119 (1), [Link]
REFERENCES HTML/?uriDCELEX:32016R0679&fromDEN.
Aarts, Emile, and Reiner Wichert (2009), “Ambient Intelligence,” in Technol- Facebook (2016), “Company Info,” [Link]
ogy Guide: Principles, Applications, Trends, Hans-J€org Bullinger, ed., Federal Trade Commission (2012), Protecting Consumer Privacy in an Era of
Berlin: Springer, 244–49. Rapid Change: Recommendations for Businesses and Policymakers,
Acquisti, Alessandro, and Jens Grossklags (2007), “What Can Behavioral Eco- March, [Link]/sites/default/files/documents/reports/federal-trade-
nomics Teach Us about Privacy?,” in Digital Privacy: Theory, Technologies, commission-report-protecting-consumer-privacy-era-rapid-change-recom
and Practices, Alessandro Acquisti, Stefanos Gritzalis, Costos Lambrinouda- mendations/[Link].
kis, and Sabrina di Vimercati, eds., New York: Auerbach, 363–77. Gomez, Joshua, Travis Pinnick, and Ashkan Soltani (2009), “KnowPrivacy,”
Downloaded by [UVA Universiteitsbibliotheek SZ] at 07:59 01 August 2017

AddThis (2016), “About Us,” [Link] June 1, [Link]

Aguirre, Elizabeth, Dominik Mahr, Dhruv Grewal, Ko de Ruyter, and Martin Google (2013), “Letter to United States Securities and Exchange
Wetzels (2015), “Unraveling the Personalization Paradox: The Effect of Commission,” December 20, [Link]
Information Collection and Trust-Building Strategies on Online Advertise- 1288776/000128877613000074/[Link].
ment Effectiveness,” Journal of Retailing, 91 (1), 34–49. Ham, Chang-Dae, and Michelle R. Nelson (2016), “The Role of Persuasion
Ahrens, Jan, and James R. Coyle (2011), “A Content Analysis of Registration Pro- Knowledge, Assessment of Benefit and Harm, and Third-Person Percep-
cesses on Websites: How Advertisers Gather Information to Customize Mar- tion in Coping with Online Behavioral Advertising,” Computers in Human
keting Communications,” Journal of Interactive Advertising, 11 (2), 12–26. Behavior, 62, 689–702.
Altaweel, Ibrahim, Nathaniel Good, and Chris J. Hoofnagle (2015), “Web Pri- Hildebrandt, Mireille (2010), “Privacy en Identiteit in Slimme Omgevingen”
vacy Census,” Technology Science, December 15, [Link] [Privacy and Identity in Smart Environments], Computerrecht, 172, 273–
2015121502. 82.
Ant
on, Annie I., Julia B. Earp, and Jessica D. Young (2010), “How Internet Hong Kong Office of the Privacy Commissioner for Personal Data (2014),
Users’ Privacy Concerns Have Evolved since 2002,” IEEE Security and “Online Behavioural Tracking,” April, [Link]
Privacy, 8 (1), 21–27. resources_centre/publications/files/online_tracking_e.pdf.
Article 29 Data Protection Working Party (2010), “Opinion 2/2010 on Online Interactive Advertising Bureau (2016), “First Quarter U.S. Internet Ad Reve-
Behavioural Advertising,” working paper 171, [Link] nues Hit Record-Setting High at Nearly $16 Billion, According to IAB,”
policies/privacy/docs/wpdocs/2010/wp171_en.pdf. June 9, [Link]
Baek, Tae H., and Mariko Morimoto (2012), “Stay Away from Me: Examining hit-record-setting-high-nearly-16-billion-according-iab.
the Determinants of Consumer Avoidance of Personalized Advertising,” Jai, Tun-Min C., Leslie D. Burns, and Nancy J. King (2013), “The Effect of
Journal of Advertising, 41 (1), 59–76. Behavioral Tracking Practices on Consumers’ Shopping Evaluations and
Balebako, Rebecca, Pedro G. Leon, Richard Shay, Blase Ur, Yang Wang, and Repurchase Intention toward Trusted Online Retailers,” Computers in
Lorrie F. Cranor (2012), “Measuring the Effectiveness of Privacy Tools Human Behavior, 29 (3), 901–09.
for Limiting Behavioral Advertising,” Web 2.0 Security and Privacy Work- Jensen, Carlos, and Colin Potts (2004), “Privacy Policies as Decision-Making
shop, [Link] Tools: An Evaluation of Online Privacy Notices,” Proceedings of the SIG-
Bang, Hyejin, and Bartosz W. Wojdynski (2016), “Tracking Users’ Visual CHI Conference on Human Factors in Computing Systems, New York:
Attention and Responses to Personalized Advertising Based on Task Cog- Association for Computing Machinery, 471–78.
nitive Demand,” Computers in Human Behavior, 55, 867–76. Keller, Kevin L. (2016), “Unlocking the Power of Integrated Marketing Com-
Beales, Howard (2010), “The Value of Behavioral Targeting,” [Link] munications: How Integrated Is Your IMC Program?,” Journal of Advertis-
[Link]/pdfs/Beales_NAI_Study.pdf. ing, 45 (3), 286–301.
Bennett, Steven C. (2011), “Regulating Online Behavioral Advertising,” John Ketelaar, Paul E., Stefan F. Bernritter, Jonathan van’’t Riet, Arief Ernst
Marshall Law Review, 44, 899–62. H€uhn, Thabo J. van Woudenberg, Barbara C.N. M€uller, and Loes
Bleier, Alexander, and Maik Eisenbeiss (2015), “The Importance of Trust for Janssen (2017), “Disentangling Location-Based Advertising: The
Personalized Online Advertising,” Journal of Retailing, 91 (3), 390–409. Effects of Location Congruency and Medium Type on Consumers’ Ad
Bygrave, Lee A. (2014), Data Privacy Law: An International Perspective, Attention and Brand Choice,” International Journal of Advertising, 36
Oxford: Oxford University Press. (2), 356–67.
Chen, Jianqing, and Jan Stallaert (2014), “An Economic Analysis of Online Kumar, V., and Shaphali Gupta (2016), “Conceptualizing the Evolution and
Advertising Using Behavioral Targeting,” MIS Quarterly, 38 (2), 429–49. Future of Advertising,” Journal of Advertising, 45 (3), 302–17.
Cranor, Lorrie F. (2003), “P3P: Making Privacy Policies More Useful,” IEEE Lambrecht, Anja, and Catherine Tucker (2013), “When Does Retargeting
Security and Privacy, 1 (6), 50–55. Work? Information Specificity in Online Advertising,” Journal of Market-
——— (2012), “Can Users Control Online Behavioral Advertising ing Research, 50 (5), 561–76.
Effectively?,” IEEE Security and Privacy, 10 (2), 93–96. Lee, Seungsin, Younghee Lee, Joing-In Lee, and Jungkun Park (2015),
eMarketer (2010), “How Should Marketers Address Concerns about Ad “Personalized e-Services: Consumer Privacy Concern and Information
Targeting?,” February 16, [Link] Sharing,” Social Behavior and Personality: An International Journal, 43
Marketers-Address-Concerns-About-Ad-Targeting/1007514. (5), 729–40.
EU Data Protection Directive (1995), “Directive 95/46/EC of the European Leon, Pedro G., Justin Cranshaw, Lorrie F. Cranor, Jim Graves, Manoj Hastak,
Parliament and of the Council of 24 October 1995 on the Protection of Blase Ur, and Guzi Xu (2012), “What Do Online Behavioral Advertising
 376 S. C. BOERMAN ET AL.

Privacy Disclosures Communicate to Users?,” Proceedings of the Annual [Link]

ACM Workshop on Privacy in the Electronic Society, 19–30. [Link].
Leon, Pedro G., Blase Ur, Richard Shay, Yang Wang, Rebecca Balebako, and Schultz, Don (2016), “The Future of Advertising or Whatever We’re Going to
Lorrie F. Cranor (2012), “Why Johnny Can’t Opt Out: A Usability Evalua- Call It,” Journal of Advertising, 45 (3), 276–85.
tion of Tools to Limit Online Behavioral Advertising,” Proceedings of the Schumann, Jan H., Florian von Wangenheim, and Nicole Groene (2014),
SIGCHI Conference on Human Factors in Computing Systems, 589–98. “Targeted Online Advertising: Using Reciprocity Appeals to Increase
Marreiros, Helia, Richard Gomer, Michael Vlassopoulos, and Mirco Tonin Acceptance among Users of Free Web Services,” Journal of Marketing, 78
(2015), “Exploring User Perceptions of Online Privacy Disclosures,” Pro- (1), 59–75.
ceedings of IADIS International Conference WWW/Internet. IADIS Press. Smit, Edith G., Guda Van Noort, and Hilde A. Voorveld (2014),
[Link] “Understanding Online Behavioural Advertising: User Knowledge, Pri-
McDonald, Aleecia M., and Lorrie F. Cranor (2008), “The Cost of Reading vacy Concerns, and Online Coping Behaviour in Europe,” Computers in
Privacy Policies,” Information System: A Journal of Law and Policy for Human Behavior, 32, 15–22.
the Information Society, 4 (3), 543–67. Stanaland, Andrea J., May O. Lwin, and Anthony D. Miyazaki (2011), “Online
———, and ——— (2010), “Beliefs and Behaviors: Internet Users’ Under- Privacy Trustmarks,” Journal of Advertising Research, 51 (3), 511–23.
standing of Behavioral Advertising,” TPRC 2010, [Link] Summers, Christopher A., Robert W. Smith, and Rebecca Walker Reczek
authors-drafts/[Link]. (2016), “An Audience of One: Behaviorally Targeted Ads as Implied
Milne, George R., and Mary J. Culnan (2004), “Strategies for Reducing Online Social Labels,” Journal of Consumer Research, 43 (1), 156–78.
Privacy Risks: Why Consumers Read (or Don’t Read) Online Privacy Sutanto, Juliana, Elia Palme, Chuan-Hoo Tan, and Chee W. Phang (2013),
Downloaded by [UVA Universiteitsbibliotheek SZ] at 07:59 01 August 2017

Notices,” Journal of Interactive Marketing, 18 (3), 15–29. “Addressing the Personalization–Privacy Paradox: An Empirical Assess-
———, ———, and Henry Greene (2006), “A Longitudinal Assessment of ment from a Field Experiment on Smartphone Users,” MIS Quarterly, 37
Online Privacy Notice Readability,” Journal of Public Policy and Market- (4), 1141–64.
ing, 25 (2), 238–49. Tucker, Catherine E. (2014), “Social Networks, Personalized Advertising, and
Miyazaki, Anthony D. (2008), “Online Privacy and the Disclosure of Cookie Privacy Controls,” Journal of Marketing Research, 51 (5), 546–62.
Use: Effects on Consumer Trust and Anticipated Patronage,” Journal of Turow, Joseph, Michael X.D. Carpini, and Nora Draper (2012), “Americans
Public Policy and Marketing, 27 (1), 19–33. Roundly Reject Tailored Political Advertising,” [Link]
Moore, Robert S., Melissa L. Moore, Kevin J. Shanahan, and Britney Mack edu/asc_papers/398.
(2015), “Creepy Marketing: Three Dimensions of Perceived Excessive ———, Jennifer King, Chris J. Hoofnagle, Amy Bleakley, and Michael
Online Privacy Violation,” Marketing Management, 25 (1), 42–53. Hennessy (2009), “Americans Reject Tailored Advertising and Three
Nill, Alexander, and Robert J. Aalberts (2014), “Legal and Ethical Challenges Activities That Enable It,” [Link]
of Online Behavioral Targeting in Advertising,” Journal of Current Issues cfm?abstract_idD1478214.
and Research in Advertising, 35 (2), 126–46. Ur, Blase, Pedro G. Leon, Lorrie F. Cranor, Richard Shay, and Yang Wang
Norberg, Patricia A., Daniel R. Horne, and David A. Horne (2007), “The Pri- (2012), “Smart, Useful, Scary, Creepy: Perceptions of Online Behavioral
vacy Paradox: Personal Information Disclosure Intentions versus Behav- Advertising,” Proceedings of the Eighth Symposium on Usable Privacy
iors,” Journal of Consumer Affairs, 41 (1), 100–26. and Security, art. 4.
O’Donoghue, Ted, and Matthew Rabin (1999), “Doing It Now or Later,” Valkenburg, Patti M., and Jochen Peter (2013), “The Differential Susceptibil-
American Economic Review, 89 (1), 103–24. ity to Media Effects Model,” Journal of Communication, 63 (2), 221–43.
Office of the Australian Privacy Commissioner (2011), “Privacy Fact Sheet 4: Van Doorn, Jenny, and Janny C. Hoekstra (2013), “Customization of Online
Online Behavioural Advertising: Know Your Options,” December, https:// Advertising: The Role of Intrusiveness,” Marketing Letters, 24 (4), 339–
[Link]/individuals/privacy-fact-sheets/general/privacy-fact- 51.
sheet-4-online-behavioural-advertising-know-your-options. Van Noort, Guda, Edith G. Smit, and Hilde A. Voorveld (2013), “The Online
Organisation for Economic Co-Operation and Development (2013), The Behavioural Advertising Icon: Two User Studies,” in Advances in Adver-
OECD Privacy Framework, [Link] tising Research. Vol. 4: The Changing Roles of Advertising, Sara Rose-
cy_framework.pdf. ngren, Micael Dahl
en, and Shintaro Okazaki, eds., Wiesbaden, Germany:
Phelan, Chanda, Cliff Lampe, and Paul Resnick (2016), “It’s Creepy, But It Springer Fachmedien Wiesbaden, 365–78.
Doesn’t Bother Me,” Proceedings of the 2016 CHI Conference on Human Weiser, Mark (1993), “Ubiquitous Computing, Hot Topics.” IEEE Computer,
Factors in Computing Systems, New York: Association for Computing 26 (10), 71–72.
Machinery, 5240–51. Witte, Kim (1994), “Fear Control and Danger Control: A Test of the Extended
Rodgers, Shelly, and Esther Thorson (2000), “The Interactive Advertising Parallel Process Model (EPPM),” Communication Monographs, 61 (2),
Model: How Users Perceive and Process Online Ads,” Journal of Interac- 113–34.
tive Advertising, 1 (1), 41–60. Yang, Hongwei (2013), “Young American Consumers’ Online Privacy Con-
Rogers, Ronald W. (1975), “A Protection Motivation Theory of Fear Appeals cerns, Trust, Risk, Social Media use, and Regulatory Support,” Journal of
and Attitude Change,” Journal of Psychology, 91 (1), 93–114. New Communications Research, 5 (1), 1–30.
Rust, Roland T. (2016), “Comment: Is Advertising a Zombie?,” Journal of Zuiderveen Borgesius, Frederik J. (2015a), Improving Privacy Protection in
Advertising, 45 (3), 346–47. the Area of Behavioural Targeting, Alphen aan de Rijn, the Netherlands:
Schaub, Florian, Aditya Marella, Pranshu Kalvani, Blase Ur, Chao Pan, Emily Kluwer Law International.
Forney, and Lorrie F. Cranor (2016), “Watching Them Watching Me: ——— (2015b), “Informed Consent: We Can Do Better to Defend Privacy,”
Browser Extensions’ Impact on User Privacy Awareness and Concern,” IEEE Security and Privacy, 2, 103–107.

View publication stats