Perceptions of Techniques Used to Personalize Messages Across Media in Real Time

Abstract

The amount of online messages that are personalized based on people's characteristics and interests is growing. Due to technological advancements, it has become possible to personalize messages across media in real time. However, little is known about people's perceptions of these different personalization techniques, while this can have important implications for message effectiveness and the privacy debate. A survey with U.S. adults (N = 1,008) showed that in the context of real-time personalization, all personalization techniques are seen as unacceptable and they are all associated with perceptions of surveillance. This applies to all generations, but younger generations are more likely to accept and to perceive less surveillance than older generations. Furthermore, we found that, of all predictors, perceived surveillance and attitudes toward personalization were the strongest predictors of acceptance of all personalization techniques. The results advance theory by differentiating between personalization techniques and introducing privacy cynicism and mobile device dependency as factors that positively relate to acceptance of personalization techniques. Practically, the results contribute to the debate on consumer agency related to people's personal data and inform media literacy programs.

Introduction

An increasing amount of messages is personalized based on people's characteristics and interests.^1,2 Because personalized messages are more relevant to people and generate less waste (i.e., reach less people outside the target group), they are perceived as an effective messaging strategy.^3,4 Personalized messages are realized using available personal data about individuals to create messages that align with their characteristics and interests. However, previous surveys showed that U.S. adults are concerned about ways companies use their personal data⁵; they experience data mining practices as creepy,⁶ and they feel being watched (i.e., perceived surveillance).⁶ However, at the same time research indicates that people are often unaware of how personalization works⁷ and also with regards to the types of data that could be used as input for personalized messages. People seem to have a general awareness about the existence of cookies and about the usage of browsing activity data as input for personalized messages.^7,8 However, when people are asked about their acceptance of personalized messaging (i.e., advertisements), they generally are aware that personalization occurs, but they do not seem to make any distinction between personalization techniques.⁸

Personalization based on concurrent media usage is currently developing in fast pace, for which increasingly sophisticated and distinctive personalization techniques are being used (Table 1). Due to technological advancements, it has become possible to personalize messages across media in real time, known as synced advertising (i.e., receiving ads on a mobile device based on TV viewing habits).⁹ Yet, little is known about the extent to which people accept the techniques to make this form of personalization happen. Especially when personalization occurs in real time across media or devices, the extent to which some personalization techniques elicit feelings of being watched might increase. Furthermore, in line with developments in privacy regulations, such as the General Data Protection Regulation¹⁰ and the California Consumer Privacy Act,¹¹ in which it is aimed to give people more control over their data,¹² it is important to study people's perceptions of these personalization techniques. Therefore, the aim of the current study is to examine people's perceptions (i.e., acceptance and perceived surveillance) of personalization techniques utilized to personalize message across media in real time and the predictors of acceptance.

Table 1.

Overview of Personalization Techniques Used to Personalize Across Media in Real Time (i.e., Synced Advertising)

Psychographic segmentation	Collecting people's demographic and psychographic information to divide the market into smaller groups (segments) based on this information.^43,44
Social media analytics	Tracking hashtags on popular social media sites.⁹
Geofencing	Companies can make use of location data on mobile devices and advertise based on people's current location.⁴⁵
IP matching	When people own multiple devices (e.g., smartphone, laptop, tablet) companies are able to relate these different devices to one single user by matching the IP addresses of the devices.
Keywords	Some installed apps “listen” to predefined words that are mentioned out loud. Then, they can trigger an ad on your mobile device based on that word (e.g., you mention brand X, and receive an ad of brand X).⁴⁶
Watermarking	Some installed apps “listen” to what you are watching on TV or listening to on the radio and they pick up a sound (watermark) from the media content.⁹

IP, Internet protocol.

The results advance theory by differentiating between personalization techniques. Although these personalization techniques are used to personalize messages across media in real time (i.e., synced advertising), some of these techniques are also used for other types of personalization (e.g., behavioral or demographic targeting). Therefore, the findings are relevant to the personalization literature in general. Moreover, we study predictors of acceptance, which will provide a better understanding of why people are more likely to accept certain personalization techniques. In addition, insights into people's perceptions of personalization techniques will contribute to the debate about consumer agency and ethical questions related to collection of people's data. Finally, insights into the predictors of acceptance will help to identify potential vulnerable groups that may need to be protected or informed. Hence, the results may inform literacy programs.

Perceptions of personalization techniques

To personalize messages across media in real time, data on people's current media usage are collected or purchased. Research into personalized message strategies showed that people have an aversion to their information being collected online¹³ and they are concerned about the ways companies use their personal information to personalize advertising messages.⁵ These privacy concerns often result in advertising avoidance, especially when people have higher levels of persuasion knowledge.¹⁴ Because synced advertising, like other personalized message strategies, makes use of personal data, and it happens in real time, it could evoke perceived surveillance (i.e., feeling of being watched⁶).⁹ The current study further explores people's perceptions of personalization techniques by examining to what extent the different personalization techniques are considered as acceptable and to what extent they evoke perceptions of surveillance (RQ1 and RQ2; Table 2).

Table 2.

Overview of Research Questions and Hypotheses

No.	Research question/hypothesis	Result
Research questions
RQ1	To what extent do different personalization techniques used in real-time personalization differ in terms of acceptance?	Most acceptable: social media analytics and segmentation
		Medium acceptable: geofencing, followed by IP matching and keywords
		Least acceptable: watermarking
		All significantly below the midpoint of the scale^a
RQ2	To what extent do different personalization techniques used in real-time personalization evoke perceived surveillance?	Most perceived surveillance: watermarking
		Medium perceived surveillance: geofencing, IP matching, Keywords
		Least perceived surveillance: Segmentation followed by social media analytics
		All significantly above midpoint of the scale^a
RQ3	To what extent are acceptance and perceived surveillance of different personalization techniques different between generations (a) and gender (b)?	(a) The younger the generation the more they accept and the lower the perceptions of surveillance
RQ3		(b) No gender differences
Hypotheses
H1	Higher perceived surveillance will lead to less acceptance	Confirmed
H2	More privacy concerns will lead to less acceptance	Rejected
H3	More negative privacy experiences will lead to less acceptance	Rejected; opposite was found
H4	More privacy cynicism will lead to more acceptance	Confirmed
H5	More positive attitudes toward personalization will lead to more acceptance	Confirmed
H6	Higher mobile device dependency will lead to more acceptance	Confirmed

p < 0.001.

Moreover, these perceptions may differ depending on generation or gender. Different generations (e.g., Gen X, Baby boomers) grew up under different circumstances, such as having varying adoption rates of media,¹⁵ which affect (media) behavior later in life.¹⁶ For example, previous research has found that older people were more negative about personalized advertising and more inclined to protect themselves than younger people.⁷ Regarding gender, some studies found that women are more concerned^17–20 and more inclined to protect their privacy,²⁰ while other studies found that men were more inclined to protect their privacy²¹ or found no differences.²² Therefore, we ask (RQ3) to what extent do acceptance and perceived surveillance of personalization technologies differ per generation and gender?

Predictors of acceptance

To get a better understanding about real-time personalization techniques' acceptance, it is important to study predictors of acceptance. The personalization paradox and privacy trade-off may shed some light on this. According to the personalization paradox, personalized messages could have both costs and benefits.²³ Costs include, for example, perceived surveillance,^6,9,24 privacy concerns,²⁵ and privacy infringement.^24,26 In addition, benefits include personal discounts²⁷ and receiving more relevant messages.^25,28,29 According to the privacy trade-off, people balance potential costs and benefits in light of possible outcomes when making decisions on, for example, sharing of personal data.^30–33 By doing this, people want to maximize positive and minimize negative outcomes.^31,32 In line with this trade-off, it is expected that costs lead to less acceptance of personalization techniques. For example, the more people have the feeling of being watched by devices that collect and process their data, the less likely they will accept a personalization technique.⁹ Conversely, potential benefits will lead to more acceptance of the personalization techniques. For example, when people have more positive attitudes toward personalized messages because it could lead to personal discounts²⁷ and more relevant messages^25,28 (Table 2).

The privacy trade-off assumes that people make rational decisions and weigh potential costs and benefits. However, this would not explain the contradicting relationship between people's privacy concerns but lack of protection behaviors³⁴—also known as the privacy paradox.^33,35–37 Two factors that may provide an explanation for the privacy paradox are privacy cynicism and mobile device dependency. First, privacy cynicism is defined as “an attitude [toward online privacy] accompanied by frustration, hopelessness, and disillusionment,”³⁸ with the sense of cynicism mainly generated from unmet expectations. Privacy cynicism represents “a cognitive coping mechanism for users, allowing them to overcome or ignore privacy concerns and engage in online transactions and self-disclosure without ramping up privacy protection efforts”.³⁴ Thus, we argue that privacy cynicism could lead to more acceptance of personalization techniques because people feel cynical to what extent they can do anything to protect their personal data.

Second, previous research into the privacy trade-off found that perceived value of an object, person, or activity can trump the costs when it comes to sharing personal data.³³ In line with this finding, we argue that people, who value their mobile device, are more willing to accept personalization techniques because not accepting it may have consequences for people's mobile device use. Therefore, we expect that the more dependent people are on their mobile device, the more likely they are willing to accept personalization techniques (Table 2).

Method

Participants and procedure

An online survey was distributed between March 31 and June 8, 2019. A request, with a link to the survey, was sent to an online panel of Dynata. Informed consent was needed to progress to the survey. The respondents received an incentive through the panel company for completing the questionnaire. A total of 1,008 respondents (55.8 percent female, Age: M = 50.26, standard deviation [SD] = 17, range 18–94 years) participated and passed at least three of the four attention checks (Table 3). We recoded the variable age in line with previous research to create generations.¹⁵

Table 3.

Demographics of the Sample (N = 1,008)

Question	Answer option	Result
Gender	Female	55.8 percent
	Male	43.8 percent
	Other	0.5 percent
Age^a	Numerical	M = 50.26 (SD = 17), range 18–94
Generations^a	Gen Z (1996>)	7.3 percent
	Gen Y/Millennials (1977–1995)	28.2 percent
	Gen X (1965–1976)	18.8 percent
	Baby boomers (1946–1964)	39.2 percent
	Prewar/silent generation (<1945)	6.2 percent
Education	No schooling	0.4 percent
	Primary school	0.5 percent
	High school	18.2 percent
	Some college, but no degree	20.5 percent
	Associate's degree	12.0 percent
	Bachelor's degree	29.5 percent
	Master's degree	12.7 percent
	Professional degree	2.6 percent
	Doctoral degree	3.1 percent
	Other	0.6 percent
Ethnicity/race^b	American Indian/Alaska native	2.0 percent
	Arab/Middle Eastern	0.2 percent
	Asian/Pacific Islander	5.7 percent
	Black, African, African American	9.3 percent
	Hispanic, Latino	8.1 percent
	Native Hawaiian/other Pacific Islander	0.1 percent
	White/Caucasian	78.6 percent
	Other	0.5 percent
Job situation	Full-time	37.5 percent
	Part-time	12.1 percent
	Unemployed	17.1 percent
	Retired	33.3 percent
Work experience	Self	13.3 percent
in ad industry	Someone close to you	9.8 percent

Note: ^aThree participants did not indicate their age.

Percentages do not add up to 100 percent because people could check multiple boxes as an answer to the ethnicity/race question.

M, mean; SD, standard deviation.

The respondents were shown descriptions of the different personalization techniques, representing the techniques in Table 1. Then, they were asked to fill out whether they accept each personalization technique (1 = Totally unacceptable, 7 = Totally acceptable), and perceived surveillance of each technique was measured with four items generated based on previous research.⁶ We asked the participants “To what extent do you believe that [technique] to sync advertisements stimulates the feeling that companies are… (a) looking over your shoulder, (b) entering your private space, (c) watching your every move, and (d) checking up on you”.

In addition, the questionnaire asked about people's privacy concerns (e.g., “I am concerned about misuse of my personal information”; Cronbach's α = 0.87, M = 5.58, SD = 1.09),^39,40 privacy cynicism (e.g., “I doubt the significance of online privacy issues more often”; Cronbach's α = 0.76, M = 3.50, SD = 1.33),³⁴ attitudes toward personalization (e.g., “I prefer that ads shown on my devices are personalized to my interests”; Cronbach's α = 0.80, M = 3.32, SD = 1.15), mobile device dependency with the Extended Mind Questionnaire (e.g., “I am very dependent on my smartphone, tablet, or computer”; Cronbach's α = 0.91, M = 4.67, SD = 1.38),⁴¹ and privacy infringement (e.g., “How often have you personally been victim of what you felt was an improper invasion of online privacy?”; Cronbach's α = 0.76, M = 4.09, SD = 1.49).⁴²

Results

People's perceptions of the personalization techniques

First, a repeated measure analysis of variance with the personalization techniques as within factor, and generation and gender as between factors, was conducted to test to what extent respondents accept the personalization techniques. We found a main effect of personalization techniques, F(1, 989) = 113.67, p < 0.001, η² = 0.10. Overall, social media analytics (M = 3.47, SD = 1.70) and segmentation (M = 3.47, SD = 1.69) are the most accepted personalization techniques, and the least acceptable is watermarking (M = 2.43, SD = 1.66). Moreover, One-Sample t-tests show that all acceptance scores were significantly below the midpoint of the scale (p's < 0.001), indicating that none of the personalization techniques was perceived as acceptable to the respondents. In addition, we found a main effect of generations, F(4, 988) = 35.10, p < 0.001, η² = 0.12. Younger generations find the personalization techniques more acceptable compared to older generations. We did not find a main effect of gender (p = 0.412) nor any interaction effects (Fig. 1; Table 4).

FIG. 1.

Perceived acceptance of real life personalization techniques per generation.

Table 4.

Acceptance and Perceived Surveillance Per Personalization Technique Across Generations

	Social media		Segmentation		Geofencing		IP matching		Keywords		Watermarking
	M	SD	M	SD	M	SD	M	SD	M	SD	M	SD
Acceptance
Gen Z (1994>)	4.58^a	1.62	4.15^b	1.72	3.50^c	1.75	3.18^c	1.81	3.5^bc	1.88	3.12^c	1.85
Gen Y (1977–1995)	4.23^a	1.80	4.03^a	1.74	3.59^b	1.75	3.43^b	1.80	3.37^b	1.92	3.00^c	1.90
Gen X (1965–1976)	3.54^a	1.83	3.51^a	1.78	3.07^b	1.70	2.89^b	1.64	2.85^b	1.66	2.51^c	1.58
Baby boomers (1946–1964)	2.82^b	1.73	3.01^a	1.78	2.67^b	1.65	2.37^c	1.50	2.15^d	1.50	1.93^e	1.31
Prewar (<1945)	2.55^abc	1.72	2.82^a	1.84	2.60^ab	1.70	2.11^cd	1.36	2.23^bcd	1.65	1.87^d	1.31
Total	3.47^a	1.89	3.47^a	1.83	3.07^b	1.75	2.82^c	1.70	2.74^c	1.79	2.43^d	1.66
Perceived surveillance
Gen Z (1994>)	4.69^a	1.79	4.67^a	1.84	5.29^ab	1.49	5.34^b	1.58	4.98^ab	1.72	5.44^b	1.60
Gen Y (1977–1995)	4.8^a	1.70	4.8^a	1.61	5.35^b	1.42	5.31^b	1.54	5.33^b	1.52	5.64^c	1.44
Gen X (1965–1976)	4.93^ac	1.87	4.82^a	1.86	5.32^d	1.80	5.42^bd	1.70	5.25^cd	1.70	5.69^b	1.61
Baby boomers (1946–1964)	5.58^b	1.53	5.33^a	1.58	5.81^c	1.38	5.87^c	1.42	5.80^c	1.42	6.06^d	1.33
Prewar (<1945)	5.65^ab	1.50	5.16^a	1.77	5.61^ab	1.70	5.90^b	1.35	5.51^b	1.47	6.13^b	1.22
Total	5.18^b	1.70	5.03^a	1.69	5.54^c	1.52	5.59^c	1.53	5.51^c	1.55	5.83^d	1.45

Note: The M and SD are presented in the table.

Different superscripts indicate significant differences between personalization techniques within one generation.

Second, we conducted the same analysis for perceived surveillance. We found a main effect of personalization techniques, F(1, 988) = 63.36, p < 0.001, η² = 0.06. Overall, perceptions of surveillance are the highest for watermarking (M = 5.83, SD = 1.45) and the least for segmentation (M = 5.03, SD = 1.69). Again, One-Sample t-tests show that all personalization techniques score significantly above the midpoint of the scale (p's < 0.001), which indicates that all personalization techniques increase the feelings of being watched. In addition, we found a main effect of generation, F(4, 988) = 11.45, p < 0.001, η² = 0.04. Younger generations experienced lower levels of perceived surveillance compared to older generations. We did not find a main effect of gender (p = 0.381) nor any interaction effects (Fig. 2; Table 4).

FIG. 2.

Perceived surveillance of real life personalization techniques per generation.

Predictors of personalization technique acceptance

A regression analysis per personalization technique was conducted with acceptance as the dependent variable (Table 5). In addition, a correlation matrix of all variables is presented in Table 6. The regression models explain 35.0 percent–39.9 percent of the variance in acceptance. The strongest predictors of acceptance of all personalization techniques are perceived surveillance and attitude toward personalization. As predicted, the higher the perceived surveillance, the less likely people are willing to accept the personalization technique (βs = −0.23 to −0.34, p's < 0.001), and the more positive people are toward personalization, the more likely they are willing to accept the personalization technique (βs = 0.22–0.035, p's < 0.001). Furthermore, the results show that people are more willing to accept the personalization technique when they are younger (βs = −0.08 to −0.21), are more dependent on their mobile device (βs = 0.07–0.13), are more cynical about privacy issues (βs = 0.07–0.17), or score higher on perceived privacy infringement (βs = 0.08–0.12). The latter is contrary to the expectations. Finally, education level is a positive predictor of social media analytics acceptance (β = 0.05, p = 0.042) and geofencing acceptance (β = 0.07, p = 0.006).

Table 5.

Multiple Regression Analysis Showing the Predictors of Acceptance of Each Personalization Technique

	Segmentation	Social media	Geofencing	IP matching	Keywords	Watermarking
	b^*	b^*	b^*	b^*	b^*	b^*
Gender (1 = female)	−0.04	−0.02	−0.02	0.01	−0.04	−0.05
Age	−0.12^***	−0.21^***	−0.08^**	−0.11^***	−0.14^***	−0.15^***
Education	0.02	0.05^*	0.07^**	0.05	0.04	0.01
Perceived surveillance^a	−0.34^***	−0.31^***	−0.23^***	−0.27^***	−0.31^***	−0.33^***
Privacy concern	0.01	0.00	−0.01	−0.06^*	−0.01	−0.05
Privacy infringement	0.11^***	0.12^***	0.08^**	0.10^***	0.08^**	0.10^***
Privacy cynicism	0.10^***	0.07^*	0.14^***	0.15^***	0.16^***	0.17^***
Personalization attitude	0.29^***	0.28^***	0.35^***	0.34^***	0.29^***	0.22^***
Mobile device dependency	0.10^***	0.13^***	0.12^***	0.07^*	0.09^**	0.08^**
F value	66.55	70.19	60.39	74.12	70.80	62.68
Degrees of freedom	9, 983	9, 983	9, 983	9, 983	9, 983	9, 982
p-Value	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001
Adjusted R²	37.3%	38.6%	35.0%	39.9%	38.8%	35.9%

Note: The model shows standardized regression coefficients (b^*). VIF diagnostics showed no multicollinearity issues (VIF's <1.45).

Perceived surveillance of the corresponding technique (e.g., perceived surveillance of segmentation for model 1, perceived surveillance of social media analytics for model 2, and so on).

p < 0.05, ^**p < 0.01, ^***p < 0.001.

VIF, variance inflation factor.

Table 6.

Correlation Matrix

	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20
1. Gender	1	−0.10^***	−0.19^***	0.04	−0.01	0.02	0.03	0.03	0.06	0.01	−0.00	−0.03	−0.03	0.13^***	−0.04	0.00	−0.03	−0.01	−0.04	−0.05
2. Age		1	0.12^***	0.14^***	0.22^***	0.14^***	0.16^***	0.17^***	0.15^***	0.15^***	−0.07^*	−0.18^***	−0.19^***	−0.27^***	−0.26^***	−0.37^***	−0.22^***	−0.28^***	−0.30^***	−0.29^***
3. Education			1	−0.05	−0.01	0.00	−0.00	0.04	0.04	−0.00	0.07^*	−0.01	−0.08^*	−0.08^*	0.01	0.01	0.03	0.00	−0.02	−0.04
4. Perceived surveillance – segmentation				1	0.66^***	0.59^***	0.58^***	0.51^***	0.47^***	0.35^***	0.22^***	−0.08^*	−0.36^***	−0.11^***	−0.45^***	−0.38^***	−0.35^***	−0.29^***	−0.18^***	−0.12^***
5. Perceived surveillance—social media					1	0.54^***	0.56^***	0.52^***	0.44^***	0.35^***	0.23^***	−0.07^*	−0.29^***	−0.11^**	−0.36^***	−0.42^***	−0.29^**	−0.27^***	−0.15^***	−0.13^***
6. Perceived surveillance—geofencing						1	0.63^***	0.55^***	0.56^***	0.42^***	0.24^***	−0.11^***	−0.37^***	−0.10^**	−0.30^***	−0.26^***	−0.38^***	−0.31^***	−0.26^***	−0.25^***
7. Perceived surveillance—IP matching							1	0.58^***	0.66^***	0.43^***	0.26^***	−0.11^***	−0.36^***	−0.09^**	−0.26^***	−0.26^***	−0.34^***	−0.43^***	−0.31^***	−0.30^***
8. Perceived surveillance—keywords								1	0.64^***	0.37^***	0.27^***	−0.12^***	−0.33^***	−0.10^**	−0.19^***	−0.18^***	−0.23^***	−0.29^***	−0.43^***	−0.30^***
9. Perceived surveillance—watermarking									1	0.42^***	0.28^***	−0.13^***	−0.31^***	−0.03	−0.15^***	−0.13^***	−0.23^***	−0.29^***	−0.35^***	−0.44^***
10. Privacy concern										1	0.32^***	−0.07^*	−0.36^***	0.07^*	−0.21^***	−0.20^***	−0.24^***	−0.30^***	−0.24^***	−0.27^***
11. Privacy infringement											1	−0.01	−0.13^***	0.18^***	0.03	0.06	0.00	−0.01	−0.02	−0.01
12. Privacy cynicism												1	0.27^***	0.20^***	0.23^***	0.22^***	0.28^***	0.30^***	0.31^***	0.30^***
13. Personalization attitude													1	0.28^***	0.47^***	0.44^***	0.50^***	0.52^***	0.48^***	0.42^***
14. Mobile Device dependency														1	0.29^***	0.32^***	0.28^***	0.25^***	0.27^***	0.23^***
15. Acceptance—segmentation															1	0.79^***	0.70^***	0.63^***	0.46^***	0.45^***
16. Acceptance—social media																1	0.64^***	0.61^***	0.46^***	0.44^***
17. Acceptance—geofencing																	1	0.74^***	0.56^***	0.58^***
18. Acceptance—IP matching																		1	0.65^***	0.65^***
19. Acceptance—keywords																			1	0.75^***
20. Acceptance—watermarking																				1

p < 0.05, ^**p < 0.001, ^***p < 0.001.

Privacy concerns did not appear to be a significant predictor of acceptance for most personalization technique except for acceptance of Internet protocol (IP) matching (β = −0.06, p = 0.036); the more concerned people are, the less likely to accept IP matching. A closer look at the data showed that privacy concerns are correlated with perceived surveillance (r = 0.35–0.43, p's < 0.001). Privacy concerns is a significant predictor in all models when removing perceived surveillance (Segmentation β = −0.07, p = 0.025; social media analytics β = −0.07, p = 0.023; geofencing β = −0.09, p = 0.005; IP matching β = −0.14, p < 0.001; keywords β = −0.08, p = 0.006; watermarking β = −0.15, p < 0.001). Then, the more concerned people are about their privacy, the less likely they will accept the personalization techniques.

Discussion

The aim of the study was to examine people's perceptions of personalization techniques used to personalize messages in real-time across media or devices, namely psychographic segmentation, social media analytics, geofencing, IP matching, keywords, and watermarking. First, the study found that people did not perceive any of the personalization techniques as acceptable, and all personalization techniques evoked perceived surveillance. Moreover, we found that younger generations were more willing to accept personalization techniques and perceived less surveillance compared to older generations. Furthermore, we found that perceived surveillance and attitudes toward personalization were the strongest predictors of acceptance. All predictors showed relationships in the expected direction except for privacy infringement. The results showed that more negative past experiences led to more acceptance of personalization techniques. Future research is needed to further examine this.

The results of the current study have important implications for theory and practice. Theoretically, the study advances our knowledge on people's perceptions of personalization techniques. Previous research showed that most people feel uncomfortable and concerned that personal data are utilized to personalize messages.¹³ The current study extends this work by showing differences per personalization technique, differences among generations, and by taking a closer look at predictors of acceptance. Although we focused on techniques of synced advertising, some of the techniques included are also used for other personalized advertising strategies (e.g., behavioral or demographic targeting). Therefore, the results of the study are relevant to a broader stream of literature as well. Furthermore, the results indicate the importance of concepts that are relatively understudied in this context, namely privacy cynicism and mobile device dependency. These concepts provide us with more insights in people's perceptions, as well as offer a possible explanation for why people with high privacy concerns do not behave accordingly (i.e., privacy trade-off). Future research should further examine the relationship of these concepts and personalized advertising strategies.

Practically, the results help us to identify vulnerable groups. For example, younger generations and people who are more dependent on their mobile device are more likely to accept the personalization techniques. This information helps us to identify high priority groups when it comes to educating media literacy. Furthermore, the high scores on perceived surveillance on all personalization techniques and the role of privacy cynicism is a reason for concern; people experience feelings of being watched but at the same time they give up because they experience a lack of agency. The results of this study contribute to the societal debate on the topic of ethics, user agency, and empowerment.

Although the study has valuable contributions, the results need to be carefully interpreted because survey data do not allow us to make causal inferences. Furthermore, synced advertising is still relatively unknown. Perceptions may change with more information available. It would be interesting to examine whether people would become more or less acceptable over time, for example, because they know more about the topic or because it is no longer new and therefore less creepy. In addition, the lack of knowledge could have resulted in respondents not clearly understanding the difference between personalization technologies. Although the correlation matrix indicated moderate relationships among personalization techniques, future research (e.g., focus groups) is necessary to further examine people's perceptions of personalization technologies.

In sum, this is the first study to examine people's perceptions of techniques to personalize messages across media in real time. Overall, these personalization techniques are perceived as unacceptable and increase perceived surveillance (i.e., the feeling of being watched) in people across generations.

Footnotes

Acknowledgment

The authors thank Nathan Casper for his help designing the survey.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

No funding was received for this article.

References

Aguirre

, Mahr

, Grewal

, et al. Unraveling the personalization paradox: the effect of information collection and trust-building strategies on online advertisement effectiveness. Journal of Retailing, 2015; 91:34–49.

AdAge. (2018) How personalization will evolve in 2018. https://adage.com/article/epsilon/personalization-evolve-2018/312048 (accessed December 3, 2019).

Kumar

, Gupta

. Conceptualizing the evolution and future of advertising. Journal of Advertising, 2016; 45:302–317.

Maslowska

, Smit

, van den Putte

. It is all in the name: a study of consumers' responses to personalized communication. Journal of Interactive Advertising, 2016; 16:74–85.

Phelps

, Nowak

, Ferrell

. Privacy concerns and consumer willingness to provide personal information. Journal of Public Policy and Marketing, 2000; 19:27–41.

Phelan

, Lampe

, Resnick

. (2016) It's creepy, but it doesn't bother me. In Proceedings of the 2016 CHI Conference on Human Factors in Computing Systems. New York: Association for Computing Machinery, pp. 5240–5251.

Smit

, Van Noort

, Voorveld

HAM

. Understanding online behavioural advertising: user knowledge, privacy concerns and online coping behaviour in Europe. Computers in Human Behavior, 2014; 32:15–22.

, Leon

, Cranor

, et al. (2012) Smart, useful, scary, creepy: perceptions of online behavioral advertising. In Proceedings of the Eighth Symposium on Usable Privacy and Security. SOUPs'12. New York: ACM, pp. 4.1—4.15.

Segijn

. A new mobile data driven message strategy called synced advertising: conceptualization, implications, and future directions. Annals of the International Communication Association, 2019; 43:58–77.

10.

Hoofnagle

, van der Sloot

, Borgesius

. The European Union general data protection regulation: what it is and what it means. Information & Communications Technology Law, 2019; 28:65–98.

11.

McCreary

. (2018) The California consumer privacy act: what you need to know. https://law.com/njlawjournal/2018/12/01/the-california-consumer-privacy-act-what-you-need-to-know/?slreturn=20190619150345 (accessed December 3, 2019).

12.

van Ooijen

, Vrabec

. Does the GDPR enhance consumers' control over personal data? An analysis from a behavioural perspective. Journal of Consumer Policy, 2018; 42:91–107.

13.

Turow

, King

, Hoofnagle

, et al. (2009) Americans reject tailored advertising and three activities that enable it. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=1478214 (accessed December 3, 2019).

14.

Ham

C-D

. Exploring how consumers cope with online behavioral advertising. International Journal of Advertising, 2017; 36:632–658.

15.

Van der Goot

, Rozendaal

, Opree

, et al. Media generations and their advertising attitudes and avoidance: a six-country comparison. International Journal of Advertising, 2016:1–20.

16.

Mannheim

. (1952) The sociological problem of generations. In Kecsckemeti

, ed. Essays on the sociology of knowledge. New York: Oxford University Press, pp. 276–322.

17.

Moscardelli

, Divine

. adolescents' concern for privacy when using the internet: an empirical analysis of predictors and relationships with privacy-protecting behaviors. Family and Consumer Sciences Research Journal, 2007; 35:232–252.

18.

Bartel Sheehan

. An investigation of gender differences in on-line privacy concerns and resultant behaviors. Journal of Interactive Marketing, 1999; 13:24–38.

19.

Youn

. Determinants of online privacy concern and its influence on privacy protection behaviors among young adolescents. Journal of Consumer Affairs, 2009; 43:389–418.

20.

Fogel

, Nehmad

. Internet social network communities: risk taking, trust, and privacy concerns. Computers in Human Behavior, 2009; 25:153–160.

21.

Milne

, Labrecque

, Cromer

. Toward an understanding of the online consumer's risky behavior and protection practices. Journal of Consumer Affairs, 2009; 43:449–473.

22.

Boerman

, Kruikemeier

, Zuiderveen Borgesius

. Exploring motivations for online privacy protection behavior: insights from panel data. Communication Research, 2018; [Epub ahead of print]; DOI: 10.1177/0093650218800915.

23.

Awad

, Krishnan

. The personalization privacy paradox: an empirical evaluation of information transparency and the willingness to be profiled online for personalization. MIS Q, 2006; 30:13–28.

24.

, Gupta

, Rosson

, et al. (2012) Measuring mobile users' concerns for information privacy. In Proceedings of the 33rd International Conference on Information Systems (ICIS) Paper 10. Orlando, FL: ICIS.

25.

Strycharz

, van Noort

, Smit

, et al. (2019) Consumer view on personalized advertising: Overview of self-reported benefits and concerns. In Bigne

, Rosengren

, eds. Advances in Advertising Research X. Wiesbaden, Germany: Springer Gabler, pp. 53–66.

26.

Wottrich

, van Reijmersdal

, Smit

. App users unwittingly in the spotlight: a model of privacy protection in mobile apps. Journal of Consumer Affairs, 2018; 53:12218.

27.

Treiblmaier

, Pollach

. (2007) Users' perceptions of benefits and costs of personalization. In ICIS 2007 Proceedings, pp. 1–15.

28.

Jari

. What is personalization? A conceptual framework. European Journal of Marketing, 2007; 41:409–418.

29.

Kalyanaraman

, Sundar

. The psychological appeal of personalized content in web portals: does customization affect attitudes and behavior?. Journal of Communication, 2006; 56:110–132.

30.

Acquisti

, John

, Loewenstein

. What is privacy worth?. The Journal of Legal Studies, 2013; 42:249–274.

31.

Culnan

, Armstrong

. Information privacy concerns, procedural fairness, and impersonal trust: an empirical investigation. Organization Science, 1999; 10:104–115.

32.

Dinev

, Hart

. An extended privacy calculus model for E-commerce transactions. Information Systems Research, 2006; 17:61–80.

33.

Wottrich

, van Reijmersdal

, Smit

. The privacy trade-off for mobile app downloads: the roles of app value, intrusiveness, and privacy concerns. Decision Support Systems, 2018; 106:44–52.

34.

Hoffmann

, Lutz

, Ranzini

. Privacy cynicism: a new approach to the privacy paradox. Cyberpsychology: journal of Psychosocial Research on Cyberspace, 2016; 10:article 7.

35.

Norberg

, Horne

. The privacy paradox: personal information disclosure intentions versus behaviors. Journal of Consumer Affairs, 2007; 41:100–126.

36.

Taddicken

. The ‘privacy paradox’ in the social web: the impact of privacy concerns, individual characteristics, and the perceived social relevance on different forms of self-disclosure. Journal of Computer-Mediated Communication, 2014; 19:248–273.

37.

Kokolakis

. Privacy attitudes and privacy behaviour: a review of current research on the privacy paradox phenomenon. Computers and Security, 2017; 64:122–134.

38.

Choi

, Park

, Jung

. The role of privacy fatigue in online privacy behavior. Computers in Human Behavior, 2018; 81:42–51.

39.

Baek

, Morimoto

. Stay away from me. Journal of Advertising, 2012; 41:59–76.

40.

Kang

, Shin

. Do smartphone power users protect mobile privacy better than nonpower users? Exploring power usage as a factor in mobile privacy protection and disclosure. Cyberpsychology, Behavior, and Social Networking, 2016; 19:179–185.

41.

Nijssen

SRR

, Schaap

, Verheijen

. Has your smartphone replaced your brain? Construction and validation of the Extended Mind Questionnaire (XMQ). PLoS One, 2018; 13:e0202188.

42.

Smith

, Milberg

, Burke

. Information privacy: measuring individuals' concerns about organizational practices. MIS Q, 1996; 20:167–196.

43.

Blattberg

, Buesing

, Sen

. Segmentation strategies for new national brands. Journal of Marketing. American Marketing Association, 1980; 44:59–67.

44.

Smith

. Product differentiation and market segmentation as alternative marketing strategies. Journal of Marketing, 1956; 21:3–8.

45.

Rodriguez Garzon

, Deva

. (2014) Geofencing 2.0: taking location-based notifications to the next level. In Proceedings of the 2014 ACM International Joint Conference on Pervasive and Ubiquitous Computing. UbiComp’14. New York: ACM, pp. 921–932.

46.

Garrity

. (2018) How does… a TV synced ad work? https://mumbrella.com.au/tv-synced-ad-work-501593 (accessed December 3, 2019).