Known case from a new perspective:
A study into public debate and experts’ views on personalization in marketing and related concerns
Joanna Strycharz Student number: 10864083
Master’s Thesis
Supervisor: prof. dr. Edith G. Smit
Graduate School of Communication
Research Master’s programme Communication Science University of Amsterdam
Introduction
Personalization can be defined as a communication strategy used to make messages more relevant for the recipient (Masłowska, 2013). This technique has been widely applied by marketers: with collected (online) data they aim to communicate with consumers as single individuals instead of a homogenous group (Kalyanaraman & Sundar, 2006). Past studies have identified multiple ways to personalize communication in marketing, such as inclusion of identification clues (i.e., the name of the consumer) or adaptation of content to the
recipient’s needs, interest, preferences or beliefs (Smit, van Noort, & Voorveld, 2014). The use of personalization strategies by marketers has been growing. For example, a study conducted in 2007 showed that 90% of regular commercial emails (e.g., newsletters) contain personalized material (Jupiter Research, 2007). Another survey conducted in 2013 showed that as much as 90% of advertising networks uses data to personalize online ads (eMarketers, 2013).
Marketers apply personalization as this technique is believed to make the message more relevant for the recipient, which in turn is expected to lead to higher persuasion. Indeed, multiple studies have confirmed this expectation (e.g., Tam &Ho, 2005; Postma & Brokke, 2002). However, academics have found multiple moderators that lower the persuasiveness of personalized messages (Dijkstra, 2008). These include preference matching (Maslowska, van den Putte, & Smit, 2011), attitude towards advertising (Baek & Morimoto, 2012), level of cognitive demand (Bang & Wojdynski, 2015), and finally, privacy concerns (Maslowska, Smit, & van den Putte, 2011; Eastin, Brinson, Doorey, & Wilcox, 2015).
Especially privacy concern is a source of major controversies connected to
personalized marketing. This strategy, also called data-driven marketing, is based on profiles created by using personal information about consumers, who tend to feel uncomfortable with it. Numerous academic studies and whitepapers have shown that the concern stemming from
data use for marketing purposes is high both in the US and in Europe (see TRUSTe, 2016; Eurobarometer, 2015).
In sum, one can conclude that personalization has been widely investigated from the perspective of the consumer; their reactions to it and their opinions have been a subject of multiple academic studies. In other words, it is a known case. The aim of this research master thesis is to look at personalization in marketing and related concerns from another angle, namely by investigating the public debate in the media on the one hand and experts’ opinions on the other hand.
This project addresses personalized marketing issues by conducting two studies presented in separate papers. Paper one applies the methods of computer-assisted content analysis to investigate the amount of coverage, the topics present and the sentiment of the coverage of online privacy related issues in general as well as in relation to marketing in two Dutch newspapers, namely Volkskrant and NRC. Paper two is based on a series of qualitative interviews conducted with experts coming from marketing agencies, market research
companies, advertising networks, and from privacy specialized organizations. The aim of this qualitative study is to draft a descriptive map of the personalization phenomenon in
marketing in the Netherlands.
In sum, the combination of a quantitative study based on a large dataset with qualitative interviews with a small group of experts allows to draw a picture of
personalization use in marketing and controversies around it, which will contribute to the literature on personalized marketing communication, offering new insights that allow to look at personalized marketing issues from a new perspective. The papers add to the knowledge coming from consumer research by presenting the perspective of the media and of
References
Baek, T. H., & Morimoto, M. (2012). Stay away from me. Journal of advertising, 41(1), 59-76.
Bang, H., & Wojdynski, B. W. (2016). Tracking users' visual attention and responses to personalized advertising based on task cognitive demand. Computers in Human Behavior, 55, 867-876.
Eastin, M. S., Brinson, N. H., Doorey, A., & Wilcox, G. (2016). Living in a big data world: Predicting mobile commerce activity through privacy concerns. Computers in Human Behavior, 58, 214-220.
eMarketer (2013). Third-party data guides campaign targeting. Retrieved from: http://
www.emarketer.com/Article/Third-Party-Data-Guides-Campaign-Targeting/ 1009894 Eurobarometer. (2015). Data protection. European Commission. Retrieved from:
http://ec.europa.eu/COMMFrontOffice/publicopinion/index.cfm/Survey/getSurveyDe tail/instruments/SPECIAL/surveyKy/2075
Jupiter Research. (2006). Best Marketing Practices. Email Address Manager. Retrieved from: http://www.emailaddressmanager.com/email-marketing/best-marketing-practices.html Kalyanaraman, S., & Sundar, S. S. (2006). The psychological appeal of personalized content
in web portals: does customization affect attitudes and behavior?. Journal of Communication, 56(1), 110-132.
Masłowska, E. H. (2013). " Just for you!" A study into the effectiveness and the mechanism of customized communication.
Masłowska, E., Putte, B. V. D., & Smit, E. G. (2011). The effectiveness of personalized e-mail newsletters and the role of personal characteristics. Cyberpsychology, Behavior, and Social Networking, 14(12), 765-770.
Masłowska, E., Smit, E., & van den Putte, B. (2011). Is Personalized Communication Superior? the Effectiveness of Personalization and the Role of Consumers’ Characteristics. AP-Asia-Pacific Advances in Consumer Research Volume 9. Postma, O. J., & Brokke, M. (2002). Personalisation in practice: The proven effects of
personalisation. Journal of Database Marketing & Customer Strategy Management, 9(2), 137-142.
Smit, E. G., Van Noort, G., & Voorveld, H. A. (2014). Understanding online behavioural advertising: User knowledge, privacy concerns and online coping behaviour in Europe. Computers in Human Behavior, 32, 15-22.
Tam, K. Y., & Ho, S. Y. (2005). Web personalization as a persuasion strategy: An
elaboration likelihood model perspective. Information Systems Research, 16(3), 271-291.
TRUSTe. (2016). 2016 TRUSTe/NCSA Consumer Privacy Infographic – US Edition. Retrieved from: https://www.truste.com/resources/privacy-research/ncsa-consumer-privacy-index-us/
Paper 1
Public debate on online privacy and marketing: Attention, topics and the sentiment of media coverage
Abstract
Online privacy and concerns related to it in the context of marketing have been widely investigated in the light of public opinion. Research on the public debate concerning those topics is limited. This study investigates the public debate concerning online privacy and marketing over the period of ten years (2006 – 2015). Following the first- and second-level agenda setting theory, media attention given to online privacy as well as topics concerning online privacy and online privacy and marketing and finally, the media sentiment are investigated. The results show that media attention given to online privacy has been increasing and this issue has been covered from the perspective of worry. In addition, the results show a strong focus of the public debate on intercultural differences concerning this topic. The sentiment of media coverage on online privacy is negative; coverage on this topic in relation to marketing shown more negativity. The paper contributes to the studies
concerning public opinion on online privacy and online privacy and marketing in describing the public debate on those issues.
Keywords: online privacy, privacy concerns, data-driven marketing, agenda-setting theory, computer-assisted content analysis, topic modeling
Privacy and particularly online privacy is said to be strongly affected by the recent changes observed in the field of information technology (van Noort, Polegato, Smit, & Vliegenthart, 2015). A consequence that is often presented in academic research and whitepapers is the growing concern of the public about how this influences their right to privacy in the information society. For example, in marketing, one can include in a banner or a newsletter the name of the target, information about demographics or the behavior of the target online (Smit, van Noort, & Voorveld, 2014). This gives many advantages to the marketers, but also carries the danger of misuse of data, privacy breach, information asymmetry, unfair practices when it comes to segmentation and manipulation (Zuiderveen Borgesius, 2015), and in turn, concerns among consumers.
A substantial body of literature deals with explanations of concern about privacy related to the use of personal data in marketing (for example, Turow et al., 2009; McDonald & Cranor, 2010; Smit, van Noort & Voorveld, 2014). Interestingly, even though some authors have stressed the role of external factors in shaping those concerns (Malhora, Kim & Agarwal, 2004), most studies focus on investigating them mainly among consumers with the use of surveys. However, these studies do not provide a more general understanding of the public debate – surveys using pre-defined concepts might not capture the debate to the fullest.
In communication science, studies have demonstrated that media content matters for public opinion about such topics as European integration (de Vreese & Boomgaarden, 2006) or political attitudes (Lenart, 1994). Moreover, media coverage has been proven to play a crucial role in the public understanding of science and technology (Scheufele & Tewksbury, 2007). Thus, one can expect that uncovering reporting in the media is crucial also for
understanding concerns.
In sum, the aim of the current study is to look into the public debate in the Netherlands and uncover topics treated in the media when it comes to online privacy in
general as well as in context of marketing. The findings of this study fill the gap in research that currently strongly focuses on public opinion, but leaves out the public debate on the topic. In particular, the current study integrates three different methods of automated content analysis, namely frequency counts, sentiment analysis and topic modeling to fully describe the public debate on privacy related topics.
Theoretical framework Privacy and research on privacy concerns
Before moving into uncovering the public debate on privacy and concerns related to it, a definition what privacy and the concerns are, is needed. Following Zuiderveen Borgesius one can identify three perspectives on privacy that overlap in many ways: privacy understood as the control one has over personal information, as limited access, and as freedom from constrains when it comes to identity construction. Privacy as control can be summarized following a definition by Westin (1968): it is “the claim of individuals, groups or institutions to determine when, how and to what extent information about them is communicated to others.” (p. 7) Next, following Warren and Brandeis (1890) privacy as limited access can be defined as “right to be let alone.” (p. 195) In the context of the current study it is worth noting that the authors constructed this definition in relation to media, more specifically
photography. They argued that “photographs and newspaper enterprise have invaded the sacred precincts of private and domestic life.” (p. 195) Central in this case is the private sphere of an individual and how it shall not be interrupted, also taking into account new technologies and possibilities they give. Finally, privacy is not only related to information, but also to the right to construct one’s identity – managing one’s image and impressions they make.
Privacy related questions have been investigated for decennia (see for example abovementioned Westin (1967) or Thomson (1975)). However, in the information age
privacy has become even more a widely discussed topic: “Information privacy has been called one of the most important ethical issues of the information age.” (Smith, Milberg, & Burke, 1996) This can be observed in users who show concerns about their privacy, which can be defined as the subjective perspective of fairness related to privacy (Campbell, 1997). The concerns differ depending on, on the one hand, individual characteristics and
experiences, and, on the other hand, on external influences (Donaldson & Dunfee, 1994). Mechanisms used to measure privacy concerns reveal its relation to privacy definition presented above: the widely applied concern for information privacy scale reflects three dimensions (Smith, Milberg, & Burke, 1996): collection, which can be seen in relation to the right to be let alone, but also to the right to withdraw information in the process of identity construction; unauthorized secondary use, which can be related to control over personal information; improper access, also connected to control, but also add another important aspect, namely errors related to storing information.
Privacy and marketing
Privacy research is especially related to marketing and marketing communications. Using personal data and monitoring people’s behavior is the basis of personalization, which is used frequently by marketers (eMarketer, 2013). It is one of the main developments in targeting smaller audience segments, which has been the aim of marketers for decennia (Turow, 2012). Nowadays, they have access to different kind of data, what allows them to aim communication at the consumer as a single individual (Kalyanaraman & Sundar, 2006).
A vast body of research has investigated people’s opinions and worries related to targeted marketing. Already in 1998 Business Week/Harris poll revealed that privacy is the biggest reason for concern among Internet users. Turow et al. (2009) asked American internet users about their opinions and concerns related to targeting: 55% of users between 18 and 24 years old did not want to be targeted based on their personal data. This number increased to
87% when respondents were made aware that data about them collected by organizations is used for targeting. A more recent survey conducted by McDonald and Cranor (2010) only 12% of respondents did not mind being tracked online for the purpose of data collection. 64% agreed with the following statement: “Someone keeping track of my activities online is invasive.” (p. 69) 46% of respondents judged behavioral advertising as “creepy.” Through the means of in-depth interviews Ur, Leon, Cranor, Shay and Wang (2012) found that
participants were “scared about being tracked and monitored” (p. 7) and felt lack of control over their personal data online. Moreover, the respondents also mentioned information asymmetry between consumers and organizations. On the other hand, they also noticed the advantages that more targeted advertising offers.
Results obtained by American researchers clearly show concern and disagreement among users. Similar results have been found in Europe. Eurobarometer (2015) studies conducted by the European Union showed privacy concerns and the feeling of lack of control over personal information – 67% of Europeans are concerned about it. Seven out of ten people also believed that commercial organizations may misuse the data they collect online. Similarly, Smit, van Noort and Voorveld concluded that: “The respondents worried about misuse of their personal data and they were especially negative about the idea of privacy violations.” (p. 20) However, the Eurobarometer study (2015) showed that 86 % of the Dutch believe that sharing data is part of modern way of living.
Media coverage and public opinion
Malhora, Kim and Agarwal (2004) note that external factors such as culture or regulatory laws influence the privacy concerns described above. One more important factor that has been identified by past research to influence individual opinions and possibly concerns is the information provided on the media. However, the public debate on the media concerning privacy concerns and marketing has not been widely investigated yet.
The underlying assumption of this influence of media can be traced back to the agenda-setting theory (Carroll & McCombs, 2003). According to this theory, topics that are salient on the media agenda are transferred to the public agenda. Correspondingly, it can be assumed that the topics related to personalized marketing and data use that are covered in the media are more likely to be present among individual’s perceptions: “elements prominent in the media picture become prominent in the audience’s picture.” (McCombs, Llamas, Lopez-Escobar, & Rey, 1997, p. 703) First, people use news media to acquire factual information about the world around them; second, they determine the importance they attach to a topic based on the frequency of its coverage and the angle the coverage takes (McCombs, 2002). Thus, the angle from which data use by commercial organizations is reported in the media might be seen as contributing or reducing the concerns of the consumers discussed above, but also to the fact that Dutch respondents are more open towards sharing their data. Looking at the current privacy concerns and the role of the media according to the first-level agenda setting theory, the following research questions are raised:
(RQ1): How much attention is given to online privacy related topics in Dutch newspapers? (RQ2a): What topics does media reporting on online privacy cover?
(RQ2b): What topics does media reporting on marketing and online privacy cover? Moreover, according to the second-level agenda setting theory not only the topics salient in the media, but also the way they are presented influences public opinion (McCombs et al., 1997). “While the agenda-setting metaphor states that the media may not tell us what to think but are successful in telling us what to think about, the metaphor for second-level agenda setting suggests that the media tell us how to think about some objects.” (Golan & Wanta, 2001, p. 249) One can differentiate between four different dimensions of second-level agenda setting: subtopics, framing mechanisms, affective and cognitive agenda setting
privacy and marketing, the affective dimension arouses special interest. It foresees that positive coverage influences how positively a specific topic is viewed by the general
audience. An example coming from a study on coverage of presidential candidates would be the positive link readers have with a specific candidate when the media portray him or her in a positive light even though the reporting concerns questionable moral issues (Golan & Wanta, 2001). Concluding from this example, it can be assumed that a negative portrayal of online privacy and marketing is more likely to cause negative attitude of public opinion towards it. Thus, looking at the crucial role of media sentiment, the following research question is posed:
(RQ3a): What is the sentiment of media coverage on online privacy?
(RQ3b): What is the sentiment of media coverage on marketing and online privacy?
Methods Data collection
An automated content analysis has been conducted for the period ranging from January 2006 until December 2015. This period incorporates the time when targeting and data use have become a common practice (see Sanje & Senol, 2012). Two Dutch quality newspapers, namely Volkskrant and NRC, have been chosen for the analysis. To retrieve relevant articles from the LexisNexis database three strategies were use. First, a generic search string was used to capture all possibly relevant articles. Next, a number of the retrieved articles was randomly selected and examined for relevance. This examination allowed to narrow down the search string in order to higher precision of retrieval and lower the number of irrelevant articles. Third, the final, narrowed search string was constructed, namely: “BODY((privacy w/p online) OR ((veiligheid w/50 (online OR internet OR web)) OR (databeveiliging w/p (online OR internet OR web)) OR (persoonsgegevens w/p (online
OR internet OR web))) AND NOT "privacy@nrc.nl").” In total, 431 NRC articles and 433 Volkskrant articles were retrieved.
Data preparation
As there is no standard software package available to process text files downloaded from LexisNexis, a series of Python codes was written for this purpose. The first collection of codes retrieved relevant part of the articles including title, date and the article text itself. In this step multiple date formats used by LexisNexis were unified and metadata was removed from the body of the article.
Another collection of codes was written with the purpose of data cleaning. In the first step, all non-alphabetic characters were removed from the texts and the data was lowercased. Next, a text-tagging tool was used to first tag parts of speech (with the help of the pattern.nl part-of-speech tagger by Smedt & Daelemans, 2012) and all words, but nouns, adjectives and adverbs were removed. Following, stopwords were removed from the data and the remaining words were tokenized and stemmed, all using the nltk package for the Dutch language. Moreover, the removed stopwords included terms used in the search-string itself, as they are expected to be mentioned in all articles selected for the analysis.
Construction of the dataset
The retrieved and preprocessed articles contained only text, accompanied by metadata such as title, source and publication date. The variables useful for analyses necessary to answer research questions had to be created using techniques of automated content analysis for journalistic texts (see Boumans & Trilling, 2016).
Media attention. This variable was operationalized as the number of articles retrieved in the data collection processed based on the count method in Python. The scores per newspaper were added to obtain one general score per year.
Topics. Topics of the preprocessed articles were identified based on Latent Dirichlet Allocation (LDA). The gensim package for Python was used for this purpose (Řehuřek & Sojka, 2010). LDA is an algorithm based on unsupervised machine learning that assumes that texts (in this case articles) consist of a number of topics. Each topic in turn consists of
number of words. The importance of a specific word for a topic is determined by number of co-occurrences with other topic-related words and the relevancy of the word for the specific topic (Boumans & Trilling, 2016). Following the approach proposed by Tsur, Calacci, and Lazer (2015), after 50 topics were generated similar topics were coded and merged to reduce their number. The coding procedure presented in detail in Appendix 1 foresaw first looking at the first word in the specific topic and attributing an initial code to it. To confirm the fit of the initial code, second up to the fifth word were taken into account. Next the final topic code was given. The final codes were compared across the 50 topics for similarities, which allowed merging them.
To identify more detailed topics related to marketing and privacy, a second step was undertaken – so called hierarchical topic model was constructed. First, articles concerned with this topic (identified in step one of topic modelling procedure) were selected from the sample. The selection was based on the presence of the topic in an article based on topic loading. Out of 864 retrieved articles 207 with topic loading higher than 5 were selected. Next, 50 topics were generated using the gensim package. Those topics were coded by the author to merge them and the same procude as in step one was followed to identify more specific topics (see Tsur, Calacci, & Lazer. 2015).
Sentiment. In the last step, the average sentiment score of all articles retrieved, as well as of the 207 articles dealing with marketing and privacy were determined. The
SentiStrength module for Python (Thelwall et al., 2010) was used to determine the sentiment of the articles. Sentistrength returns two values: positivity (ranging from 1 – not positive to 5
– extremely positive) and negativity (ranging from -1 – not negative to -5 – extremely negative). Using those two scores, the average sentiment per article was calculated in the following way: positivity (i.e. the positive value) + negativity (i.e. the negative value), resulting in a scale ranging from -4 to 4.
Results
To answer the first research question, descriptive information on dynamics of media coverage is provided. Figure 1 shows the yearly number of articles published on online privacy. It clearly shows an increase in the number of articles over the years – particularly, after 2010 the number of articles was almost twice as high as in the years before and it remained at this level.
Figure 1. Number of articles on privacy between January 2006 and December 2015. This figure illustrates the number of articles concerning privacy published in Volkskrant and NRC over the course of 10 years.
Research question 2a concerns the topics in the media related to online privacy. To answer this research question 50 topics identified in the first step of topic modelling were coded and merged. Out of the 50 topics, 4 were not coherent enough to interpret them.
0 20 40 60 80 100 120 140 160 2004 2006 2008 2010 2012 2014 2016
Interpretative coding allowed to reduce the number of topics to eight, which are presented in Figure 2.
Figure 2. Topics in the public debate on online privacy. This figure illustrates the topics that construct public debate on online privacy in the Netherlands.
The topic about surveillance concerns this practice executed by such organs as FBI and NSA, and is connected to such actors as Edward Snowden and Glenn Greenwald. Safety of medical data concerns in general information about patients. Next, data safety online deals mostly with hacker attacks. Safety of children on the Internet concerns not only the children
themselves, but also the role of parents. When it comes to protection by authorities, Internet users as seen as citizens whose personal data should be protected by the state and its organs such as police. Cookie law is related to the so called “cookiewet” that was introduced in the Netherlands in February 2015. Next, privacy and commercial organizations relates to data use by such companies as Google or Facebook as well as companies collecting information on their clients. This topic is further investigated in the second step of hierarchical topic modelling. Finally, privacy on social networks mostly concerns Facebook also in relation to the fact that the company is based in the US.
To answer research question 2b a similar coding was undertook in the second step of the hierarchical topic modelling. In this step, out of the 50 topics, 16 were not coherent
enough to interpret them. The rest of the topics was merged to identify overarching topics, which are presented in Figure 3.
Figure 3. Topics in the public debate on online privacy and marketing. This figure illustrates the topics that construct public debate on online privacy in relation to marketing in the Netherlands.
Those include the differences between US and Europe in data use by organizations, particularly concerning privacy standards. In this context, it is mentioned frequently how Facebook and Google use and protect data of their users. Next, how banks treat the data of their clients comes up in the articles. Banking sector is also brought in relations with surveillance, namely how banks share data with governmental organs. Data that mobile applications collect is another topic in the debate. This includes social media apps and mobile baking apps. Moreover, when it comes to mobile data, also data collection by media
companies such as RTL is discussed. Next, the debate covers the use of particular data including biomedical data and electronic health records. Next, it is also discussed how data can be a source of income for organizations. Moreover, the debate also covers the protection that companies offer. In particular, a big leak of personal data of clients of the Dutch telecom provider KPN that took place in February 2012 is discussed in detail. Finally, data collection
and use by various commercial organizations is discussed including Google, Facebook, Wikipedia and Dropbox.
To answer the third research question, sentiment scores per article were used to compute an overall average sentiment score. When it comes to all retrieved articles (i.e. articles concerning privacy in general it amounts -0.89, which on a scale ranging from -4 to 4 is slightly negative. On the other hand, looking into articles concerning specifically to online privacy and marketing it amounts -1.37. It can be concluded that the tone of the articles concerning both online privacy and online privacy and marketing is negative with the later showing more negativity.
Discussion and conclusion
The aim of this research was to examine the public debate on online privacy in general as well as online privacy and marketing in Dutch media. Automated content analysis demonstrated that first of all, the number of articles concerning online privacy in general has been increasing over the past years. Second, online privacy is associated with multiple topics in the Dutch media, namely surveillance, safety of particular data, children safety online, protection offered to citizens by the authorities, cookie law in the Netherlands, privacy and commercial organizations and finally, privacy on online social networks. Fourth, online privacy in marketing context has been found to be associated with such topics as American and European privacy issues, data safety, use of cookies for data collection, data of app users, particular personal data and income from data collection. Finally, through the means of sentiment analysis, a negative overall sentiment of -0.89 for articles about online privacy in general and of -1.37 for articles concerning privacy and marketing was shown.
First, the fact that the number of articles published concerning online privacy has been growing over the years lies in line with what could have been expected. Over the same period of time, privacy online has become a concern among users and companies have started
collecting and using data. Similarly, according to Slegg (2006) in 2006 only 8% of
companies advertising online did so with the use of data. In comparison, eXelate in a yearly survey conducted among practitioners from the digital marketing industry found that in 2013 94% of agencies and 83% of advertisers used audience targeting in their digital display ads (eMarketer, 2013).
In the light of agenda setting theory that states that topics that are salient on the media agenda are transferred to the public agenda (Carroll & McCombs, 2003) one could conclude that the frequent coverage of privacy related issues in Dutch newspapers contributes to the opinion among the Dutch that sharing data is a natural part of modern way of living (Eurobarometer, 2015).
Second, looking at the topics present in the Dutch media in relation to online privacy, they lie in line with past research on dimensions of privacy (privacy as control over
information, as limited access, and as unconstrained identity construction). Privacy as control over one’s personal information is most prevalent in the topics identified – the focus lies on data collection by commercial organization, mostly big corporations such as Facebook and Google. Moreover, the fourth dimension introduced by Smith, Milberg and Burke (1996), namely errors related to information storing, is strongly present in public debate as one of the most prominent topics both in the general context as well as in the marketing context is data protection by companies and avoiding hacker attacks. Privacy as control over personal data is also present in research on public opinion about privacy – in surveys, such items are used as “I am concerned about the potential misuse of personal data” (Smit, van Noort & Voorveld, 2014) or “No one should use data from Internet history.” (McDonald & Cranor, 2010, p. 68) Thus, the findings of the current study lie in line with past survey research into online privacy and privacy concerns as both highlight one’s right to control over their personal data.
Next, investigating topics present in the public debate, articles often treat both the situation in Europe and in the US. Interestingly, the Dutch media do not focus on the
differences between the Netherlands and the US, but between Europe in general and the US. Indeed, past research shows that when it comes to privacy protection, the Anglo-Saxon world differs significantly form its continental European counterparts (Milberg, Smith, & Burke, 2000). European countries have decided to implement strict legal framework for privacy protection while UK and US opt for case-by-case ruling about online privacy issues. These differences result in different standards in Europe and the US.
Fourth, privacy related to sensitive data comes back in the public debate – medical and biometrical data are mentioned both in general context and in the context of marketing. This lies in line with law practice as the so called “particular personal data” is under special protection in the Dutch law (de Zwart, 2015). The reason for it is the higher hazard of manipulation and higher information asymmetry between the consumers and organizations, which are also one of the main ethical concerns when it comes to biomedical data use by commercial organizations (Zuiderveen Borgesius, 2015). Thus, it is not surprising that the media treat those data as a separate topic related to online privacy.
Moreover, data safety of app users is another occurring topic. This lies in line with research into this issue as “app stores are not free from privacy- invasive apps that collect personal information without sufficient disclosure or user consent.” (Choe, Jung, Lee & Fisher, 2013) Apps are also often seen as intrusive – through push notification they breach privacy understood as the right to be left alone. Thus, the presence of this topic is not surprising.
When it comes to the sentiment of the articles, both public debate about online privacy in general as well as about privacy in the context of marketing are negative. Newspaper articles show privacy in a negative light – this suggests a perspective of worry
and lack of protection. The Dutch are afraid to lose control over their personal data (Eurobarometer, 2015) and so are Dutch newspapers. Interestingly, Dutch newspapers are more negative about online privacy and marketing than they are about online privacy in general. As according to the second-level agenda setting theory media sentiment influences public sentiment (McCombs et al., 1997), this negativity can be seen as one of possible explanations why the Dutch are more negative towards data collection by corporations in comparison to data collected online by the Government or news websites (Smit, van Noort & Voorveld, 2014).
As with all studies, the current study has some limitations. As the current study has a very explorative character, unsupervised approach to topic modelling was chosen. In other words, possible topic categories were not defined beforehand based on existing research and researcher’s expectations. Maybe the openness of the topic modelling approach led to the fact that particularly in the second step of the analysis, multiple topics were difficult to code and 17 were not coherent. An alternative way for future studies would be a supervised approach, namely predefining the number of topics one expects, which would possibly allow better topic coding and evaluation. The decision on the expected topics could be based on past findings or on interviews with experts on this topic. Next, the chosen method for sentiment computation is not without flaws. SentiStrengh is designed to work with short texts, such as Tweets. It takes the value of the most positive sentence as the positivity value and the value of the most negative sentence as the negativity value (Thelwall et al., 2010). For longer texts, such approach can be seen as less appropriate. Future research could develop another
approach that would be better able to represent the sentiment of a longer text.
Although public opinion on privacy issues has been investigated multiple times, not much research has focused on the media coverage of such issues. Future research could investigate how these issued are covered in the online media as online news is not only more
up-to-date comparing to offline news, but is also more and more taking over the role of print media within the daily information routines of many individuals. Moreover, as differences between countries and cultures when it comes to privacy protection have been found as one of the determined topics, a cross-national analysis that includes Anglo-Saxon countries could be the next step in exploring those issues.
All in all, from a methodological perspective, the current study has illustrated how the use of computational techniques of automated content analysis can improve understanding of the public debate on the topic of online privacy. On the other hand, from a theoretical
perspective, the findings show that privacy in the media is portrayed in a similar way that has been found in survey research. Moreover, the study fills a gap by complementing past
References
Boumans, J. W., & Trilling, D. (2016). Taking Stock of the Toolkit: An overview of relevant automated content analysis approaches and techniques for digital journalism
scholars. Digital Journalism, 4(1), 8-23.
Boyd, D., & Ellison, N. (2010). Social network sites: definition, history, and scholarship. IEEE Engineering Management Review, 3(38), 16-31.
Bruno, O. (2014). DATALEK: KLANTGEGEVENS KPN TOEGANKELIJK DOOR HACK. [Web log article]. Retrieved from: https://www.bof.nl/2012/02/15/datalek-klantgegevens-kpn-toegankelijk-door-hack-2/
Campbell, A. J. (1997). Relationship marketing in consumer markets: A comparison of managerial and consumer attitudes about information privacy. Journal of Interactive marketing, 11(3), 44-57.
Choe, E. K., Jung, J., Lee, B., & Fisher, K. (2013, September). Nudging people away from privacy-invasive mobile apps through visual framing. In IFIP Conference on Human-Computer Interaction (pp. 74-91). Springer Berlin Heidelberg.
Coleman, R., & Banning, S. (2006). Network TV news' affective framing of the presidential candidates: Evidence for a second-level agenda-setting effect through visual framing. Journalism & Mass Communication Quarterly, 83(2), 313-328.
Carroll, C. E., & McCombs, M. (2003). Agenda-setting effects of business news on the public’s images and opinions about major corporations. Corporate Reputation Review, 6(1), 36–46.
De Vreese, C. H., & Boomgaarden, H. G. (2006). Media effects on public opinion about the enlargement of the European Union. JCMS: Journal of Common Market
Golan, G., & Wanta, W. (2001). Second-level agenda setting in the New Hampshire primary: A comparison of coverage in three newspapers and public perceptions of
candidates. Journalism & Mass Communication Quarterly, 78(2), 247-259.
de Zwart, D. (2015). WIE WEET OF JE NAAR DE KERK OF MOSKEE GAAT, OF DAT JE THUIS BLIJFT?. [Web log article]. Retrieved from: https://bof.nl/2015/08/21/wie-weet-of-je-naar-de-kerk-of-moskee-gaat-of-dat-je-thuis-blijft/
Donaldson, T., & Dunfee, T. W. (1994). Toward a unified conception of business ethics: Integrative social contracts theory. Academy of management review, 19(2), 252-284. eMarketer (2013). Third-party data guides campaign targeting. Retrieved from: http://
www.emarketer.com/Article/Third-Party-Data-Guides-Campaign-Targeting/ 1009894 Eurobarometer. (2015). Data protection. European Commission. Retrieved from:
http://ec.europa.eu/COMMFrontOffice/publicopinion/index.cfm/Survey/getSurveyDe tail/instruments/SPECIAL/surveyKy/2075
Hoofnagle, C. J., Soltani, A., Good, N., Wambach, D. J., & Ayenson, M. D. (2012).
Behavioral advertising: the offer you cannot refuse. Harvard Law & Policy Review, 6(2), 273.
Kalyanaraman, S., & Sundar, S. S. (2006). The psychological appeal of personalized content in web portals: does customization affect attitudes and behavior?. Journal of
Communication, 56(1), 110-132.
Leenes, R. (2015). The Cookiewars: From regulatory failure to user empowerment? In M. van Lieshout, & J-H. Hoepman (Eds.), The Privacy & Identity Lab: 4 years later. (pp. 31-49). [3] Nijmegen: The Privacy & Identity Lab.
Lenart, S. (1994). Shaping political attitudes: The impact of interpersonal communication and mass media. Thousand Oaks, CA: Sage.
Malhotra, N. K., Kim, S. S., & Agarwal, J. (2004). Internet users' information privacy concerns (IUIPC): The construct, the scale, and a causal model. Information systems research, 15(4), 336-355.
McCombs, M., Llamas, J. P., Lopez-Escobar, E., & Rey, F. (1997). Candidate images in Spanish elections: Second-level agenda-setting effects. Journalism & Mass Communication Quarterly, 74(4), 703-717.
McDonald, A., & Cranor, L. F. (2010). Beliefs and behaviors: Internet users' understanding of behavioral advertising. TPRC.
Milberg, S. J., Smith, H. J., & Burke, S. J. (2000). Information privacy: Corporate management and national regulation. Organization science, 11(1), 35-57.
Řehůřek, R., & Sojka, P. (2010). Software framework for topic modelling with large corpora. In Proceedings of the LREC 2010 Workshop on New Challenges for NLP
Frameworks (pp. 45–50). Valletta, Malta: ELRA.
Sanje, G., & Senol, I. (2012). The importance of online behavioral advertising for online retailers. International Journal of Business and Social Science, 3(18).
Scheufele, D. A., & Tewksbury, D. (2007). Framing, agenda setting, and priming: The evolution of three media effects models. Journal of communication, 57(1), 9-20. Slegg, J. (2006). What's the Buzz Behind Behavioral Advertising?. Search Engine Watch.
Retrieved from: https://searchenginewatch.com/sew/news/2066269/whats-buzz-behind-behavioral-advertising
Smedt, T. D., & Daelemans, W. (2012). Pattern for python. Journal of Machine Learning Research, 13(Jun), 2063-2067.
Smit, E. G., Van Noort, G., & Voorveld, H. A. (2014). Understanding online behavioural advertising: User knowledge, privacy concerns and online coping behaviour in Europe. Computers in Human Behavior, 32, 15-22.
Smith, H. J., Milberg, S. J., & Burke, S. J. (1996). Information privacy: measuring individuals' concerns about organizational practices. MIS quarterly, 167-196. Thelwall, M., Buckley, K., Paltoglou, G., Cai, D., & Kappas, A. (2010). Sentiment strength
detection in short informal text. Journal of the American Society for Information Science and Technology, 61(12), 2544–2558.
Thomson, J. J. (1975). The right to privacy. Philosophy & Public Affairs, 295-314.
Turow, J., Hennessy, M., & Draper, N. A. (2015). The tradeoff fallacy: How marketers are misrepresenting American consumers and opening them up to exploitation. Available at SSRN.
Turow, J., King, J., Hoofnagle, C. J., Bleakley, A., & Hennessy, M. (2009). Americans reject tailored advertising and three activities that enable it. Available at SSRN 1478214. TRUSTe. (2016). 2016 TRUSTe/NCSA Consumer Privacy Infographic – US Edition.
Retrieved from: https://www.truste.com/resources/privacy-research/ncsa-consumer-privacy-index-us/
Tsur, O., Calacci, D., & Lazer, D. (2015). A frame of mind: Using statistical models for detection of framing and agenda setting campaigns. In Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics and the 7th International Joint Conference on Natural Language Processing (pp. 1629–1638). ACL.
Turow, J. (2012). The daily you: How the new advertising industry is defining your identity and your worth. Yale University Press.
Ur, B., Leon, P. G., Cranor, L. F., Shay, R., & Wang, Y. (2012). Smart, useful, scary, creepy: perceptions of online behavioral advertising. In proceedings of the eighth symposium on usable privacy and security (p. 4). ACM.
Van Noort, G., Antheunis, M. L., & Verlegh, P. W. (2014). Enhancing the effects of social network site marketing campaigns: If you want consumers to like you, ask them about themselves. International Journal of Advertising, 33(2), 235-252.
Van Noort, G., Polegato, A., Smith, E., & Vliegenthart, R. (2015). Privacy and the information society. Online Privacy covered by Western Newspapers: A cross-national analysis of the USA, UK, Netherlands, France, Spain, and Denmark. Unpublished manuscript.
Warren, S. D., & Brandeis, L. D. (1890). The right to privacy. Harvard law review, 193-220. Westin, A. F. (1968). Privacy and freedom. Washington and Lee Law Review, 25(1), 166. Zuiderveen Borgesius, F. J. (2015). Improving privacy protection in the area of behavioural
Appendix A
Topic Coding Coding decisions:
– I looked at the first word of the topics identified by the LDA, given that they have the highest loading per topic and thus can be considered as most important
– I excluded ‘waar” from the topic coding, as they appear frequently and overlap with other topics. Instead, we look at the following word that loads second highest per topic.
– In case the second word could not be clearly identified as a topic, I looked at the following words
– For the topics that could not be identified based on the first words, I looked at the following words to see whether they could be fit to other (already defined) topics; otherwise, we defined a new topic
Table 1.
Initial and final codes given in the first step of topic modeling. 1st Word in
Topic List
Initial Code Other Words in
Topic List
Second Code Merged Topics
1 Facebook Social media gebruiker, Amerikaanss Privacy on social networks
2 Groot ? google, heel Data use by Google Data use by companies
3 waar ? cookies, groot, minister Cookie law Cookie law
4 waar ? over, burger, steds Protection of citizens Data protection by
authorities
5 Nederland Data protection in NL waar, politie Data protection by
authorities
6 Foto Social media procent, google Data use by Google Data use by companies
7 Amerikaans Data protection in the US ouder, week, heel Protection of children Children safety online 8 Europees Data protection in Europe google, groot, Facebook Data use by companies Data use by companies
authorities authorities 10 Burger Data protection by
authorities over, groot, Amerikaans Data protection by authorities
11 Waar ? informative, politie, rotterdam Data protection by police Data protection by authorities
12 Gegevens ? waar, Amerikaans, week ---
13 Kinder Protection of children ouder, heel, groot Children safety online
14 Gegevens Data saftey Amerikaans, over, bedrijf Data use by US
companies Data use by companies
15 FBI Data protection in the US apple, telefoon, waar Surveillance Surveillance
16 Land ? waar, groot, Amerikaans ---
17 Groot ? europees, groot, heel bedrijv Data use by foreign
companies
Data use by companies
18 Vorig ? procent, hacker, informatie Data safety Data safety
19 Groot ? nsa, Amerikaans, waar Surveillance Surveillance
20 Gegevens ? Google, gebruiker, weer Data use by Google Data use by companies
21 Amerikaans Data protection in the US bedrijf, waar, informatie Data use by US
companies Data use by companies
22 Gegevens informatie, Amerikaanss,
bedrijv Data use by US companies Data use by companies
23 Facebook Social media gegevens, informative, land Privacy on social networks
24 Ouder Protection of children waar, foto, kinder Children safety online
25 Groot ? waar, politie, patient Medical data safety
26 Week ? waar, waarin, Amerikaanss ---
27 Hacker Data safety Americaanss, gegevens, land Data safety
28 Amerikaans Data protection in the US digitaal, over, groot Data protection
29 Burger Data protection by
authorities informate, open, groot Data protection by authorities
30 Google Data use by companies Amerikaans, groot, bedrijf Data use by Google Data use by companies 31 Burger Data protection by
authorities
over, groot, informatie Data protection by
32 Over ? waar, facebook, wereld Privacy on social networks 33 Burger Data protection by
authorities groot, over, naam Data protection by authorities
34 Amerikaans Data protection in the US europees, Facebook, recht Social media Privacy on social networks
35 Informatie ? Amerikaans, week, groot ---
36 Amerikaans Data protection in the US gebruiker, klant, zak Data use by US
companies Data use by companies
37 Informatie ? over, waar, steds ---
38 Facebook Social media informative, groot, waar Social media Privacy on social networks
39 waar ? groot, gegevens, mogelijk,
wachtword Data protection Data protection
40 Facebook social media groot, Amerikaans, nederland Social media Privacy on social networks
41 Groot ? burger, gegevens, mogelijk, Data protection by
authorities Data protection by authorities
42 Groot waar, hacker, informatie Data safety Data safety
43 Media ? kinder, land, procent Children safety Children safety online
44 Facebook Social media Amerikaans, wereld, digitaal Privacy on social networks
45 Facebook Social media waar, week, vrouw Privacy on social networks
46 KPN Data use by companies bedrijv, gegevens, europees Data use by companies
47 Computer ? over, bedrijf, waar Data use by companies
48 Snowden Survillance groot, Facebook, Greenwald Survillance
49 Foto Social media politie, groot, waar Protection by police Data protection by
authorities
50 Informatie Data safety over, groot, bedrijf Data use by companies Data use by companies
Table 2
Initial and final codes given in the second step if topic modelling. 1st Word in
Topic List Initial Code Other Words in Topic List Second Code Merged Topics
1 Militair Surveillance klant, Amerikaans, ING Surveillance Surveillance
2 Amerikaans American companies land, over, burger American companies Privacy standard in the US
3 Google Google data krant, gebruiker, gegevens Google data Privacy of Google clients
4 Google Google data foto, Amerikaans, data Google data Privacy of Google clients
5 Amerikaans American companies Huawei, klant, Chinees - ---
6 Klant Privacy of clients bank, steeds open Privacy in banking Privacy in banking
7 Gegevens ? heel, justitie, nl - ---
8 Leraar ? website, gegevens, bezoeker Data safety Data safety
9 Vrijheid ? journalist, Huawei, beter - ---
10 Europees European companies and
privacy burger, politie, krant European companies and privacy Privacy standards in Europe
11 Facebook Facebook data gebruiker, app, wereld Facebook data Privacy of Facebook users
12 Cookies Use of cookies klant, gegevens, KPN Use of cookies Data collection by cookies
13 Europees European companies and
privacy klant, Amerikaans, Google European and Amerikan rules Differences US - Europe
14 Facebook Data saftey Mededingingsautoriteit, app,
gebruiker Data of app users - rules App users’ privacy 15 Google Google data Facebook, gebruiker, mogelijk Google and Facebook
data Corporations collecting data
16 Geld Earning money on data Euro, extra, website Earning money on data Earning money on data
17 KPN KPN data leak Facebook, hacker, klant KPN data leak Data safety
18 Chinees ? gegevens, wereld, Vledder - ---
19 Over ? politie, gegevens, burger - ---
20 Gegevens Data collection landelijk, dossier, biometrisch Privacy of senstive data Privacy of senstive data
22 Gegevens Data collection gebruiker, media, Nederland Data collection Data collection online 23 Europees European companies and
privacy heel, week, winter - ---
24 RTL Data collection of RTL tijd, kanaal, mobiel RTL mobile data Mobile data
25 Google Google data heel, Facebook, woord Google and Facebook
data Corporations collecting data
26 Mogelijk ? zak, foto, land - ---
27 Heel ? gegevens, Amerikaans ,
Amsterdam - ---
28 Gegevens Data collection groen, klant, KPN KPN data leak Data safety
29 Gegevens Data collection land, Europees, Amerikaans European and Amerikan
rules Differences US - Europe
30 Wikipedia Data on Wikipedia artikel, vriend, persoon Data on Wikipedia Data on Wikipedia
31 Klant Privacy of clients KPN, zak, gegevens KPN data leak Data safety
32 Den ? dag, best, klant - ---
33 Procent ? proefpersoon, euro, politie - ---
34 Krant ? klant, gegevens, bank - ---
35 Groen ? helder, journalistiek, boek - ---
36 Cookies Use of cookies klant, computer, bezoeker Use of cookies Data collection by cookies
37 NL ? Nederland, cookies, website Use of cookies Data collection by cookies
38 Gegevens Data collection klant, persoon, netwerk Data on social media Privacy on social networks
39 Gegevens Data collection Google, klant, data Google data Privacy of Google users
40 Amsterdam ? Dropbox, gegevens, deel Dropbox data safety Privacy of Dropbox users
41 Gegevens Data collection groen, week, Amerikaans - ---
42 Cookies Use of cookies gegevens, heffing, NPO Use of cookies by NPO Data collection by cookies
43 Google Google data Amerikaans, gegevens,
Facebook Google and Facebook data Corporations collecting data 44 Muziek Online music services steeds, legaal, justitie Legal online music Online music
45 Gegevens Data collection data, nam, heel - ---
46 Dag ? onderzoek, open, vrijheid - ---
48 Google Google data Amerikaans, Facebook, Europees
European and Amerikan rules
Corporations collecting data 49 Gegevens Data collection sigaret, Google, gebruiker Google data Privacy of Google users
Paper 2
“Make communication more relevant”:
An exploratory study of personalized marketing in the Netherlands
Abstract
Last years have brought a rise to the use of personalization in marketing both in America as well as in European countries. A wide scope of research has investigated this phenomenon from the perspective of the consumer, taking such factors as effectiveness and privacy concerns into account. The current study extends those efforts by constructing a descriptive map of this phenomenon specifically in the Netherlands based on a dialogue with multiple experts, namely marketers, market researchers and privacy specialists. Thorough the means of qualitative expert interviews, the current study sought to find out how personalization is defined, applied, if it is seen as effective and how consumers’ concerns are taken into account in practice. The results indicate that experts and academics roughly agree on definition and application of personalization, but the experts put more emphasis on effectiveness. Experts also call for actions from lawmakers and academics. The paper advances previous studies by moving away from the consumer-centered perspective to a perspective that puts the
practitioners from the field central.
Keywords: personalized marketing, effectiveness of personalization, data-driven marketing, privacy concerns, expert interviews
“Half the money I spend on advertising is wasted; the trouble is I don't know which half.” – this saying attributed to an American proponent of advertising John Wanamaker (Maxin, 2007) shows an old marketers’ problem – everything is advertised to everyone, but not everyone wants everything. As McDonald and Cranor (2010) underline, only a portion of the general public is interested in any given product or service. Targeting customers with a specific interest seems like an ideal solution to reduce the amount of the irrelevant and eventually solve the old dilemma. With the rise of digital technology and access to so-called big data, personalized advertising is not a dream any more.
The last years have brought a sudden rise to the use of personalized communication in marketing. With the accessibility of data, marketers can use information that directly refers to the recipient as a single individual (Kalyanaraman & Sundar, 2006). This way advertising is not any longer directed at a broad audience; using profiles created for Internet users based on a wide range of data, advertisers can nowadays include the name of the target and use
demographic characteristics or Internet habits to reach the right audience (Smit, van Noort, & Voorveld, 2014). For example, a user who looks for a hotel in Paris will later be exposed to an advertising message for “the best hotels in Paris” while reading the news or browsing via Google. A survey among professionals working in the digital marketing sector from 2013 showed that ninety percent of advertising platforms and more than eighty percent of advertisers make use of data to personalize their advertising (eMarketer, 2013) and these numbers are expected to grow.
However, even though effective in reaching the right audience and in reducing the amount of irrelevant advertising, personalization is also seen as one of the most controversial practices used by marketers. The fact that the targeting is based on profiles created by using personal information makes consumers feel uncomfortable. In 2016 the TRUSTe report revealed that 92% of US Internet users worry about their privacy online and the worry keeps
growing. Moreover, one can observe so-called chilling effects stemming from the negative attitude towards personalization. According to TRUSTe (2016), 74% of Americans have limited their online activities due to concerns about their data, while 51% refrain from
clicking a personalized ad. Along these lines, past studies have shown mixed findings when it comes to the effectiveness of personalization. Multiple studies have found that privacy
concern moderates the effectiveness of personalized advertising so that higher privacy concern leads to lower effectiveness of personalized advertising (Maslowska, Smit, & van den Putte, 2011; Eastin, Brinson, Doorey, & Wilcox, 2015).
Personalization and data protection are also heavily debated topics within the European Union and in the Netherlands. In Europe the worry of its citizens is not smaller comparing to Americans – 67% of Europeans are concerned about the control they have over their personal data online, while 69% believe that commercial organizations may misuse the personal information they collect (Eurobarometer, 2015).
One can conclude that with the use of surveys public opinion on privacy and concerns associated with it has been thoroughly investigated. However, what is missing is “the other side of the coin,” namely the opinions of experts who are involved in designing and
executing personalized marketing activities. Thus, the aim of the current study is to look deeper into the use of personalization and to construct a descriptive map of this phenomenon specifically in the Netherlands. Through the means of expert interviews, a dialogue with professionals who are involved into designing and applying personalization strategies on a daily basis – namely, marketers, market researchers and privacy specialists, will allow unveiling another perspective on this issue.
The paper opens with a brief introduction to personalization and research on the topic, followed by controversies around this issue. Second, it presents findings from the expert
interviews and puts them in relation to academic research. Third, it discusses the implications of these findings and draft trends for the future of personalization in marketing.
The development of personalized marketing
When it comes to online activities, advertising has been one of the most important revenue sources for many organizations for decades (Langheinrich, Nakamura, Abe, Kamba, & Koseki, 1999). However, the amount of advertising, which is often irrelevant for the customer, makes it more difficult for marketers to reach their audience. Thus, targeting advertising to users who are most likely to be interested in a particular product or service has been gaining popularity (McDonald & Cranor, 2010).
Personalization, also called targeting, is an old phenomenon that can be traced back to direct marketing. Personalized letters sent to a pre-defined group of receivers were used by marketers as early as the 19th century (Ross, 1992). Nowadays, personalization can generally be defined as “advertising that incorporates information about the individual, such as
demographic information, personally identifying information (e.g., name, residence, and job) and shopping-related information (e.g., purchase habit or history and brand preference)” (Bang & Wojdynski, 2015). However, there is lack of agreement on a single definition of personalization (Kemp, 2001)
Research into personalization has investigated multiple areas where this method can be applied: the abovementioned direct mail or telemarketing can be seen as personalized communication (Baek & Morimoto, 2012) as can the online behavioral advertising based on past online behavior (Smit, Van Noort, & Voorveld, 2014).
As Dijkstra (2008) underlines, the personalization online is limited to the possibilities given to it by computer technology. Only concrete aspects of the message can be adopted. Dijkstra (2008) names them ‘tailoring-ingredients’ and distinguishes three main types: adaptation, when the content of the message is adjusted to the single individual;
personalization that refers to including recognizable elements about the target in the message (such as the name); and finally, feedback related directly to ‘something about the target’, i.e. their assessed psychological or behavioral state (for example a direct reference to their eating habits). In case of adaptation the target may not be aware that the message is tailored to their needs.
In sum, personalization involves various techniques and is used in many different contexts, thus not all aspects can be included in this paper; the focus is laid on marketing communication, defining personalization, the tailoring ingredients and personalization techniques described above. This choice is motivated by the high controversies associated with personalization in marketing that are treated in more detail below.
Controversies around personalized marketing
As shown in the previous section, personalized communication is more and more applied by marketers as they expect broad benefits; however, it is not free of controversies. Several studies have proven that consumers are in general rather negative towards
personalization (Sheehan & Hoy, 1999; Turow, King, Hoofnagle, Bleakley & Hennessy, 2009). Turow et al. (2009) asked American internet users about their opinions and concerns related to targeting and concluded that 55% of users between 18 and 24 years old did not want to be targeted based on their personal data. In a more recent survey conducted by McDonald and Cranor (2010) 46% of respondents claimed that they see behavioral
advertising as “creepy.” Moreover, in Europe 69% of internet users believe that commercial organizations may misuse their data they collect online (Eurobarometer, 2015).
According to White, Zahay, Thorbjørnsen and Shavitt (2008) multiple factors play a role in defining consumers’ attitudes towards personalization: level of personalization, justification for applying it and the utility of the message. The most common controversy concerns privacy: consumers find personalization invasive (McDonald and Cranor, 2010) and
feel lack of control over their data (Ur, Leon, Cranor, Shay & Wang, 2012). Thus, another focus of this study involves examination of the perceptions experts have on these
controversies among consumers.
Methodology
To match the exploratory character of the study, qualitative expert interviews were conducted with individuals representing organizations that are involved in designing and applying personalized marketing communication in the Netherlands, or can be seen as experts in the field of privacy.
Participants
To recruit participants, purposive sampling was used, which can be seen as
appropriate as the recruited participants had to strictly fulfill certain characteristics (Riffe, Lacy & Fico, 2014). With regard to this study, experts had to come from organizations that are either applying personalized marketing or are specialized in privacy issues.
11 members of different organizations were interviewed: six decision makers from digital marketing agencies that are engaged in personalization processes, one from a market research agency, one from an advertising network, one legal practitioner from an inter-branch organization and finally, two legal practitioners from privacy-specialized firms (see Table 1). Two practitioners decided to take part in the study on an anonymous basis.
Table 1.
Organizations participating in the expert interviews
Category Organizations
Digital marketing agency Stan & Stacy, Lead Today, WebPower, Yourzine, Storm Digital, Anonymous
Market research agency Veylinx Advertising network Anonymous Inter-branch organization DDMA
To determine the optimal number of participants the principle of data saturation was applied – data collection stopped when no new themes that would add to the existing results would appear (Glaser & Strauss, 1967).
Procedure
The interviews were conducted face-to-face. Before each interview the purpose of the study was explained to the participants and they were given the right to withdraw from the study at any time and to stay anonymous. The consent text can be found in Appendix A.
All interviews happened at the location chosen by the interviewee, being it either the company or the University of Amsterdam. The interviews lasted 45 minutes on average and were semi-structured. The questions were developed based on past research, but were adapted based on initial results. Figure 1 presents the sampling and interviewing procedure.
Figure 1. Sampling and interviewing procedure. Flowchart presents how topics of inquiry for the interviews and sampling were defined.
An interview side with a topic list that was used (see Appendix A) started with general questions about the organization (e.g. it’s activities, size, clients), definition of
personalization (e.g., typology, justification, expectations), their experiences with the practice (e.g., applied strategies, example campaigns), effectiveness of personalization (e.g., way of
measuring it) risks and thresholds experienced by the organization (e.g., economic or legal considerations), their perceived user’s attitude towards personalization (e.g. privacy concerns) and finally, trends for the future in the field of personalization.
The expert-interviews were recorded, transcribed and analyzed. Data analysis
After transcribing, the interviews were analyzed in two steps. First, the transcripts were initially read and open codes were assigned to bits of data. Initial properties of
categories were defined in this step. During the process of open coding, multiple memos were created by the author to describe how codes from different transcripts were related to each other and to reflect the first results against the existing literature on personalization (Corbin & Strauss, 1990). In the second step, with the help of the initial codes and the memos, axial codes were assigned in order to group the initial codes under the main five categories from the topics list: definition of personalization, practice of personalization, effectiveness of personalization, experienced thresholds, user’s attitudes and future developments.
Quotations from the interviews were consulted with the interviewees and are introduced in italics in the current paper.
How to define personalization?
Looking at the literature, finding a single definition of personalization is not easy – past studies provide us with numerous definitions, often dramatically different from each other. Following Peppers, Rogers and Dorf (1999) one could define personalization as differentiation of a product or service for the benefit of the consumer. On the other hand, Allen, Kania and Yaeckel (2001) limit personalization to web experience. Wind and Rangaswamy (2001) include customization initiated by the consumer in the definition of personalization. What connects all those and other definitions is the fact that personalized
messages are targeted to a specific person and that marketers see the customer as a single individual (see also Imhoff, Loftis & Geiger, 2001).
According to Kemp (2001) practitioners stumble upon the same issues trying to define personalization – it means something different to each marketer who claims to apply it. However, looking at experts opinions one can derive a definition consisting of four main elements: To reach the right person, with the right content at the right time (Jasper Dijkstra, Storm Digitaal) (see Figure 2).
Figure 2. Personalization definition. Graph presents four elements of personalization mentioned by the experts.
First, the corresponding target group has to be identified; second, identification of the
demand is necessary – the sender should be able to produce the appropriate content; third, the content has to be provided at the appropriate time when the receiver is interested in it and needs it. Meeting those three conditions is the ultimate goal of marketers who apply personalization. Moreover, when it comes to commercial organizations a fourth condition came up – namely, a measurable return on investment (ROI). Experts all agree that the positive effect is an important aspect, but disagreement was discovered when it comes to the type of ROI – as also past research shows, outcome of advertising can be either to increase sales or to create more difficult to measure awareness (Niazi, Siddiqui, Alishah, & Hunjra, 2012). However, even though all those four elements define personalization they do not all
