Personalized Advertising on Social Media: Its attributes and user perceptions
Personalized Advertising on Social Media: Its attributes and user perceptions
Yanran Wang
Teachers College, Columbia University
Abstract
As personalized advertising becomes increasingly prevalent on social media with the development of internet-based behavior-tracking and database technology, it is necessary to learn about audience’s perception as a guidance of the industry. This study applies in-depth interview and observation to examine audience’s perception, attitude and behavior towards personalized advertising and their evaluation on different attributes of these advertisements. Findings are that users show positive attitudes toward personalized advertising and tend to look into them as a reference when need and otherwise ignore them, while they would not directly buy things via the platform, nor would they totally block them. Generally, personalized ads prevail non-personalized ones and privacy concerns are mitigated with the benefits of relevance. As for the attributes, time, location, preference, interest, brand, promotion, privacy control and frequency are valued by the users. Personal information is boycotted. Price and search history are controversial.
Keywords: personalized advertising, advertising attributes, customer perception
Introduction
When scrolling down your Instagram timeline, sponsored posts appear from time to time, which are linked directly to the websites of these sponsors. You may feel surprised that the brands are those you just searched yesterday and that was with your browser, not Instagram at all. Well, you are the target audience for personalized advertising. It is written in the Instagram Ads page (shown in the appendix) that they use what users do on Instagram and Facebook and on third-party sites and apps to show business information that is interesting and relevant to the users. Although you can react clicking that certain ads are irrelevant or you see them too often, you cannot make them disappear or stop them from being “relevant” to you, and such prevalence of personalized advertising continues to grow (Internet Retailer, 2013).
Advertising on social media is struggling as the cost of massive communication is high and research such as the report of MediaMind (2012) shows low click-through rates of traditional banner advertisements (as cited in Bleier and Eisenbeiss, 2015). Thus, retargeting becomes an important strategy for social media, which is to personalize the ads based on individual consumers’ recent online shopping behaviors and other information (Bleier and Eisenbeiss, 2015) since firstly this is more efficient for companies to conduct customer care (Yu and Cude, 2009) and also many researchers and marketers believe that personalized advertisements are less likely to cause ad avoidance (Baek and Morimoto, 2012; Bang and Wojdynski, 2016) and hence the audience attention could be transferred into profits.
Personalization is not a new concept in marketing. The reason it is such a heat in social media industry nowadays lies in the development of internet-based behavior-tracking and database technology (Bang and Wojdynski, 2016). However, these advanced technologies may frighten some users exerting their privacy concerns. The controversy around personalized advertising, especially on social media never ceases. Current research around personalized advertising mainly focus on building the function model of it or examining the effectiveness of personalization, or discussing its relationship with privacy concern, while through interview and observation, this study aims to listen to the customers, depict their attitudes toward personalized advertising and find out the attributes of personalized advertising which would make it easily accepted by the audience. With these contributions of customers’ perception and attributes that matter, both sponsors and social media executives would gain a much clearer concept of their following moves.
Literature Review
Personalized Advertising
Researchers give different definitions of personalized advertising. Baek and Morimoto (2012) define personalized advertising as making customized promotional messages to deliver to individual consumer through paid media based on personal information (such as consumers’ names, past buying history, demographics, psychographics, locations, and lifestyle interests). Bang and Wojdynski (2016) put personalization as tailoring of message content and delivery based on data collection or covert observation of users, to increase the personal relevance of message. Wirtz & Göttel and Daiser (2017) define it as targeted advertising based on user’s personal characteristics and preferences. All the definitions talk about customers’ personal information, while the components of the information set may be different. Also, the application of such information could be to merely target the relevant audience, or to target and personalize the content of the advertisements. This difference can result in customers’ different degree of awareness of personalized advertising. Tucker’s research (2014) found that the privacy policy change of Facebook significantly affects the click-through rate of certain targeted and personalized advertisements, while merely targeted advertisements have no prominent difference. Since merely targeted advertisements still make use of personal information and customers’ opinions toward it is valuable, both types of personalized advertising are discussed in this study.
Abundant research has examined the effectiveness of personalization of advertising while the conclusions are diverse. Supporters such as Bleier and Eisenbeiss (2015) argue that personalization increases click-through; Bang and Wojdynski (2016) conclude that personalized advertisements draw more visual attention than non-personalized ones; Baek and Morimoto (2012) conclude that increased perceived personalization leads directly to decreased ad avoidance. Opponents such as Yu and Cude (2009) argue that customers generally hold negative perceptions of personalized advertising, regardless of the channel of delivery. Then it is proved that the effectiveness of personalized advertising is not all or nothing and we need to look into details of its specific aspects.
Advertising Attributes
Various attributes are discussed and examined in determining successful advertising. With regard to personalized advertising, Xu & Liao and Li (2008) examine and conclude that the most important factor influencing personalization is context (e.g. location, weather, time), followed by user preference and content (e.g. price, discount, brand); Tucker (2014) find privacy control as influencing factor; Bleier and Eisenbeiss (2015) name the credibility of brand and relevance of interest as influences; Zhu and Chang (2016) also emphasize the importance of relevance. Above all, Chen and Hsieh (2012) carefully go through previous literature about advertising attributes and divide the attributes into content, context and personal profile. Context includes weather, user activity, location, time and device type. Content includes promotion, price, and brand name. Personal profile includes background information, preference, interests, search history and virtual community. Their survey with the experts finds that price, preference, promotion, interest, brand, and type of mobile device can be defined as important contributors. Among these aforementioned attributes, relevance is a general idea and almost cover all the others. The context dimension is usually used to target audience and content dimension is to personalize the ads. User preference of Personal profile can be used to target or personalize and privacy control is to soothe users’ concerns as a whole.
The role of relevance
Since almost all the attributes are related to relevance, the literature on the role of relevance in consumer research could serve as the framework for the hypotheses that certain attributes would actually affect customer behavior. Zhu and Chang (2016) define the relevance in personalized advertising as the extent to which consumers see the advertisements as self-related or can be functional in achieving their goals and meeting their needs. Previous research claims the effects of relevance throughout the perception, attitude and behavior dimension of customers, such as decreasing ad avoidance (Baek and Morimoto, 2012), soothing privacy concerns (Zhu and Chang, 2016), paying closer attention to the ads (Bang and Wojdynski, 2016), showing positive attitudes towards the ads (Campbell & Wright, 2008), showing higher purchase willingness (Pavlou and Stewart, 2000), and promoting advertising effectiveness (Drossos and Giaglis, 2005) (as cited in Zhu and Chang, 2016). Zhu and Chang (2016) draw on two theories to explain the mechanism: rational choice theory and self-awareness theory. They prove their hypotheses that relevance by contributing additional perceived benefits to customers motivates them to accept it with privacy as a tradeoff. What’s more, personalization through successfully activating self-awareness would in turn alleviates users’ perception of privacy invasion.
Research Method
The research of this study is conducted in two stages. Observation and in-depth interview are combined to gain comprehensive insights of what social media users’ perception, attitude and behavior are like towards personalized advertising as well as what attributes of these advertisements they care about.
Observation: How Users React to Personalized Ads
The core question illustrated via observation is that through accurately targeting and adding personalized content, whether personalized advertisements are treated equally as the posts actively followed by users.
Popular and incorporating a large number of personalized ads, Instagram is chosen to be the representative of social media. Five students from Columbia University joined in the observation who have different genders, and cultural and academic backgrounds, while all of them are active on Instagram reporting that they would use it for 10 minutes at least every day. So, they do have some insightful thoughts on the issue, and their reactions and feelings are real.
Approaches including field notes and instrumented systems are applied during the observation, which works like this: each participant is required to watch their Instagram feeds on their phone for 5 minutes without any specific instruction, during which they use the application on the phone to record the screen. At the same time, researcher takes notes of the participants’ conditions and expressions. After the observation, data like the proportion of ads, the proportion of ads that participants perceive relevant, stickiness of non-ads and ads posts, responses to non-ads post and responses to ads post are collected from the screen record to reflect how users’ attitude and behavior toward such ads are. Instead of merely interviewing participants of their perceptions and attitudes, observation helps in generating a more authentic and objective description. Instrumented systems as an observational method, can provide accurate information on specific factors while can hardly address contextual information (Cash et al., 2015) and hence taking field notes is necessary to record other useful information other than users’ executions.
In-depth Interview: What Are the Attributes That Matters
Concluded from previous work, four dimensions of attributes are valued by social media users through in-depth interviews. Referring to the work of Xu & Liao and Li (2008), as well as Chen and Hsieh (2012), attributes of personalized advertising can be mainly categorized into three
Figure 2. Attributes of personalized ads that are valued in the interview
dimensions: context, personal profile and content. Since research shows that privacy concern is closely related to personalization and audience acceptance (Tucker, 2014; Zhu and Chang, 2016), and social media as Facebook, Instagram and Weibo tend to enact policies for users’ power of privacy, privacy control is also included as a subject. Specific attributes of the three dimensions: context, personal profile and content, are chosen based on the research results of Xu & Liao and Li (2008), Chen and Hsieh (2012) and Bleier and Eisenbeiss (2015). Some modifications are done in this study, based on personal experience, primitive interview and affordances of the social media, such as involvement of personal information, research history and frequency, and elimination of device type. Figure 2 shows the attributes illustrated in the interview.
In-depth interviews are conducted with some of the participants right after the observation, to let them reflect the motives of their choice and reaction. Questions asked in the interview are listed in the appendix. Chen and Hsieh (2012) valued the attributes with experts throughout the industry value chain, which is very influential and inspiring, while end users’ voice should not be absent. Participants in this current study are not decision makers, but every single decision affects their daily information consumption. Also, plentiful examples show that users’ opinions do push forward reassessment and reform, such as the compromise of Facebook mentioned by Tucker (2014) due to public criticism. Thus, the users’ evaluation and preference of advertising attributes as well as their description of what ideal ads would be like generated from the interview could contribute to the social media advertising as well.
Results and Discussion
During the observation taken on five participants who were told to normally consume the information on their timeline for five minutes without posting anything of themselves, some data are collected and listed in Table 1. In five minutes, each user can read 70-100 posts. Ads Proportion is defined as the result of the number of posts marked as “sponsored” divided by the whole number of posts the individual participant read. Perceived Relevant Ads proportion is based on the self-report of the participants. Stickiness is fuzzily recorded as the time participants spent on each ad or non-ads post. The responses recorded are clicking “like”, swiping to the next photo when a set of photos is posted, reading a post and clicking the owner to check the one’s profile, clicking the outside link of ads and clicking the option “hide ads”.
Table 1
Observation of Users’ Behavior
Ads Proportion
|
Perceived Relevant Ads Proportion
|
Stickiness Per Non-Ads Post (Second)
|
Stickiness Per Ads Post (Second)
|
Non-Ads Post Responses
|
Ads Responses
|
|
Participant 1
|
11.59%
|
75%
|
3.93
|
7.50
|
6 likes
|
1 swipe, 1 click link
|
Participant 2
|
12.20%
|
75%
|
7.97
|
2.60
|
2 likes, 4 check profiles
|
1 swipe
|
Participant 3
|
10.99%
|
100%
|
3.49
|
1.40
|
1 swipe, 5 check profiles
|
/
|
Participant 4
|
8.22%
|
67%
|
4.12
|
4.00
|
3 likes
|
1 hide ads
|
Participant 5
|
10%
|
77.78%
|
2.92
|
4.11
|
1 like
|
1 swipe
|
Besides, other findings are concluded from the field notes with regard to participants’ overall behavior:
§ Compared with non-ads posts, users have more obvious likes and dislikes towards ads, as the time they spent on ads is polarized. Either they stick to the ad or they swipe to the next immediately.
§ Users have no emotional expressions when watching ads while they sometimes appear surprised or pleasant towards some non-ads, of which the most are videos.
§ Users’ stickiness to posts decreases generally as time passes.
§ Overall users spent little time in reading texts like description or comments.
§ Users find it hard to focus on Instagram for five minutes given that they can only receive information, not post.
§ Users tend to interact with people they know offline.
Records of three in-depth interviews are analyzed. All the participants noticed the existence of sponsored posts and the certain relevance to themselves, but none of them are clear that their behaviors outside Instagram (e.g. on Facebook, third-party websites and apps) are also utilized in deciding which advertisements they will see. As for attitudes toward such personalized advertisements, all the participants find themselves between neutral and like. When asked to choose between personalized ads and non-personalized ones, two thirds affirmatively stand with the former while the other one mentioned the consideration about filter bubble. More specific findings combined with the observation results are illustrated in the following discussions.
Users’ Attitudes and Behaviors: Diversity and Similarity
What is shown firstly in Table 1 and the interviews is the diversity of users. Given that the data of ads proportion and relevance are relatively congruent among different participants, compare stickiness of ads and non-ads and we can divide users into at least three categories: the one sticking to ads longer, the one sticking to non-ads longer and the one treating them barely equally. Thus, it seems not always that due to natural ads resistance users will neglect ads. Participants also acknowledged in the interview that when they happen to have purchase intention and the ads are congruent with their preference, they would pay attention to them:
Previously I decided to buy sneakers and followed several relative accounts. Then I noticed quite a few sneaker brands appeared in my feeds marked as sponsored which actually helped me get more information and made the decision (Participant 3, interview, 2017).
When asked why would they ignore or even resist thus sponsored posts, participants also have different opinions. Participant 1 said that when she has completed the purchase, she would be annoyed of continuously seeing relevant selling in Instagram, which is to say purchase intention plays an important role. Participant 3 considered frequency as an essential problem, her expectation is seeing ads two to five times per day. If the ads proportion increases and bothers her from reading her own feeds, she would take measures to block ads. Ads quality is also a factor in consumers’ ads acceptance. Participant 2 said that his motive to use Instagram is to appreciate high-quality photographs, thus if the ads are shoot delicately and not break the harmony, he would not mind seeing them. However, if the ads are too commercial and cheesy, he would feel violated.
Although attitudes are diverse among different people or during different stages of an individual, a similarity reflected through the observation is that people are not willing to interact with or respond to personalized ads. In the observation, the ads link is only clicked once and the rest responses are just swiping the photos. no likes. No comments. In the interview, similarly, only participant 3 reported had clicked the link “shop now” once, while that shopping was not realized at last. Social media try to transfer from merely advertising platforms to the promoter of direct purchase of the sponsors, while it seems that users just don’t buy it.
Evaluation of different attributes
The context attributes including time and location are generally welcomed by users. Time attribute can be defined as seasonal time or personal schedule. The ads applying the former would be like, for example, promoting different products according to holiday or events, while not much individual information is utilized in it. Personal schedule can be studied to cater to one’s purchase intention and participants feel fine about it. The condition of location attribute would be more complicated. On one hand, participants feel more sensitive about their location information, considering its relation to their privacy and safety. On the other hand, information catering to their location is more relevant and appealing to them. Once again, different individuals hold different opinions for this trade-off. Participant 2 showed trust in the social media, saying that media platforms would take duties to protect users’ privacy thus their location information cannot be used wrongly. Participant 3 said that if the ads aim at promoting products then the location doesn’t have to be used since people can buy almost anything online. One thing all participants agree with is that randomly posted ads regardless of their location or time information would not be relevant to them, and watching such ads would be a waste of time.
Conditions for the attributes in personal profile dimension are widely divergent. Firstly, all the participants hate the idea of their personal information being used in the ads. They would find the ads showing their names or institutions a serious violation:
When using Weibo, I sometimes saw my username in the posts that are obviously ads, I admit they would catch my attention but I’m really pissed at this. It feels like I’m deceived (Participant 3, interview, 2017).
I’m fine with that being used by the medium itself, but cannot tolerate it being used by its sponsors because I didn’t allow them to do so (Participant 2, interview, 2017).
At the same time, other attributes like interest and preference are welcomed to be used, since users would either have searched similar information on their own or they just don’t mind seeing things congruent with their senses. Search history is a bit tricky, participants reported that they were startled by the appearance of exactly the same products with their search somewhere else, and since then they realized the personalization of these ads. Aside from privacy concern, there is also the problem of filter bubble:
I’m tired of seeing the same products once and once again and being informed of what I already knew. It’s like the shopping assistants following me everywhere persuading me to buy them. I want to be the master of myself and learn about other good things offered in the market (Participant 1, interview, 2017).
Contradictorily, the only one clicking the “shop now” link is conducted by Participant 1, and the product was exactly personalized with her search history. Thus, it is evident that customers sometimes struggle with the balance of their purchase desire and their privacy or information control.
Then there come content attributes. Promotion and brand information is generally preferred by the customers, which directly affect whether an advertisement is appealing or not. As for price, participant 2 expressed his antipathy since in that way the ads would be too pushy:
Instagram is fine by now, but when I use search engines, those banner advertisements with bare prices really drive me crazy and I will never click them (Participant 2, interview, 2017).
Participants have different tolerance of ads frequency while not many complaints were heard toward current frequency of ads in Instagram, which is around 10 percent according to the observation. While they do emphasize that high frequency would exert their ads resistance. Given that ten percent is much higher than their expected frequency which is twice to five times per day, it can be assumed that relevance of ads alleviates their ads resistance.
Not much privacy control is given on Instagram. Users can hide or report ads and report them to be not relevant, too frequent or inappropriate while no option is there to stop personal information being used. One way suggested to protect privacy is to trick the system by selecting everything to be irrelevant. But during the observation and interview, participants rarely do any safeguard procedure although everyone claimed their privacy concern quite clearly, which shows some sort of dependence on personalized advertising and supports the conclusion of Zhu and Chang (2016), that perceived advertisement relevance mitigates consumer's privacy concerns.
Conclusion and Implication
To conclude, the perception, attitude and behavior toward personalized advertising are examined through this study with the participants, which reflect the thoughts of population to some extent. Users are aware of the personalized advertising and show positive attitudes toward it. They tend to look into them as a reference when need and otherwise ignore them, while they would not directly buy things via the platform, nor would they totally block them. Generally, personalized ads prevail non-personalized ones and privacy concerns are mitigated with the benefits of relevance. As for the attributes, time, location, preference, interest, brand, promotion, privacy control and frequency are valued by the users. Personal information is boycotted. Price and search history are controversial.
The implications for stakeholders such as media executives and advertisers are that they should keep working on the personalization of advertising within the legal and ethical standards and focus on the attributes valued by customers as suggested above. Meanwhile, they should avoid selling too hard, too frequent exposure or exposure of audience’s personal information.
Reflection
This study is an interesting and fresh trial for me, during which I acquired much help from my instructor and my participants. As I’m interested in social media, marketing and advertising, this project drives me to learn more about the industry and I feel more intelligent now both theoretically and technically. One thing I would like to improve is the scale of the study. As I do find some interesting phenomenon and data worth collecting, if I could involve more participants into the observation and interview, we could learn better about users’ thinking logic and make the conclusion more convincing and representative. And that’s a suggestion for future studies.
References
Baek, T. H., & Morimoto, M. (2012). Stay Away From Me. Journal of Advertising,41(1), 59-76. doi:10.2753/joa0091-3367410105
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. doi:10.1016/j.chb.2015.10.025
Bleier, A., & Eisenbeiss, M. (2015). Personalized Online Advertising Effectiveness: The Interplay of What, When, and Where. Marketing Science,34(5), 669-688. doi:10.1287/mksc.2015.0930
Cash, P., Hicks, B., Culley, S., & Adlam, T. (2015). A foundational observation method for studying design situations. Journal of Engineering Design,26(7-9), 187-219. doi:10.1080/09544828.2015.1020418
Chen, P., & Hsieh, H. (2012). Personalized mobile advertising: Its key attributes, trends, and social impact. Technological Forecasting and Social Change,79(3), 543-557. doi:10.1016/j.techfore.2011.08.011
Tucker, C. (2011). Social Networks, Personalized Advertising, and Privacy Controls. SSRN Electronic Journal. doi:10.2139/ssrn.1694319
Wirtz, B. W., Göttel, V., & Daiser, P. (2017). Social Networks: Usage Intensity And Effects On Personalized Advertising. Journal of Electronic Commerce Research, 18(2), 103-123.
Xu, D. J., Liao, S. S., & Li, Q. (2008). Combining empirical experimentation and modeling techniques: A design research approach for personalized mobile advertising applications. Decision Support Systems,44(3), 710-724. doi:10.1016/j.dss.2007.10.002
Yu, J., & Cude, B. (2009). ‘Hello, Mrs. Sarah Jones! We recommend this product!'Consumers perceptions about personalized advertising: comparisons across advertisements delivered via three different types of media. International Journal of Consumer Studies,33(4), 503-514. doi:10.1111/j.1470-6431.2009.00784.x
Yu, J., & Cude, B. J. (2009). Possible Disparities in Consumers Perceptions Toward Personalized Advertising Caused by Cultural Differences: U.S. and Korea. Journal of International Consumer Marketing,21(4), 251-269. doi:10.1080/08961530802282166
Zhu, Y., & Chang, J. (2016). The key role of relevance in personalized advertisement: Examining its impact on perceptions of privacy invasion, self-awareness, and continuous use intentions. Computers in Human Behavior,65, 442-447. doi:10.1016/j.chb.2016.08.048
Appendix A
Research Plan
Research question:
What attributes of personalized advertisements on social media would affect users’ attitude and how’s users’ attitude towards them?
Topics/theories in literature review:
- Personalized advertising
Comb and conclude the development of personalized advertising and how it works currently. Cite research on personalized advertising to show whether it makes big impact on people’s social media using and consuming. Refer to research to illustrate the potential issues of personalized advertisement (validity, efficiency, users’ favorability…).
- Advertising attributes
Several scholars have discussed the important attributes in designing a mobile advertising message (Chen & Hsieh 2012). For example, research found time, location, promotion, price, brand, relevance, interest, platforms and other attributes of the advertisements affect their response rates and effectiveness. My research would combine these previous findings with my own analyses to build the hypotheses of this project.
- Affective labor
Affective labor is widely used in today’s ICT design in order to resonate with consumers and draw their attention. My research would cover the theories of affective labor to examine what kind of advertising design would be better accepted by social media users.
- The role of relevance
Research show that relevance plays an important role in mitigating users concerns towards privacy invasion (Zhu & Zhang 2016). This research would involve relevance as a major attribute of the advertisements to examine if relevance positively affects users attitude toward personalized advertisements.
Group of people / location / site to study:
Plan to select registered students in Columbia University as an example of social media users to respond to the interviews on their attitudes.
How to access to them:
through e-mails or social media
How to collect / record data
Record the observation as video and interview as audio.
Possible limitations of your exploratory research / literature review:
Participants’ self-report may be slightly different from their actual behavior so I may involve observation into my research.
The sample I draw may hardly be a random one and the representativeness of Columbia University students is limited.
Implications of your exploratory research / literature review:
My exploratory tends to implicate that the attributes of personalized advertisements, such as their relevance to the users, the affective labor elements they use in design and etc. significantly affect users attitudes and reactions toward the advertisements, thus to provide suggestions on the design of such promotional campaign.
Appendix B
Instagram Ads Page
Appendix C
Interview Questions
1 Your time spent on Instagram per day
2 Have you ever noticed the sponsored posts in your time line?
3 How do you react to these ads? (ignore, like, comment, click through, buy, dislike and avoid, worry)
4 What’s your positions towards these ads? (thoughts, advice, worry)
5 Have you ever noticed the personalized features of the ads? (your information, your interest, your search history…)
6 What do you think of such personalization?
7 How do you feel like the personalized ads that use your location/time information to target you? Why?
8 How do you like the personalized ads that represent promotion/price/brand name information in the content? Why?
9 How do you feel like your preference/ interest/ search history being used to target you and being represented in ads to add relevance to you? Why?
10 What do you think the frequency of personalized ads should be like?
11 Do you have privacy concerns? How do you like having more privacy control? If so, would you be more open to personalized ads?
12 What would be ads that are effective to you like?
Comments
Post a Comment