When was the last time that you saw an online advertisement that you found creepy? And do you remember what exactly caused this feeling of creepiness? Many of us have experienced some feelings of mistrust while surfing the web or visiting Facebook, and this mistrust may be triggered by a wide variety of causes and suspicions. In this post, we analyze and discuss a number of cases, based on interviews with twelve Facebook users. Unsurprisingly, users do not interact with advertisements that they consider creepy, but they also often ignore non-creepy advertisements. Ad explanations should be credible and perhaps advertisers should try and throw in some humor.

Clowns and creepiness

Before we dive into the results, it might be good to first have a closer look at the concept of creepiness itself. Oxford's Learner's Dictionary defines 'creepy' as "causing an unpleasant feeling of fear or slight horror", "strange in a way that makes you feel nervous".

Psychologist McAndrew explained a couple of factors that contribute to this: many people, including children, are afraid of clowns. Even though clowns are meant to be funny and to make you laugh, their behavior is unpredictable: before you know it, you have a pie in the face. Furthermore, clowns wear masks and make-up that hide their true identities and feelings, and clowns typically wear unusual clothing. All this may trigger people to be on their guard in a potentially threatening or unpleasant situation.

clown

New technologies have always met anxiety and fear, because social norms were still developing. For example, back in 1865 the UK Locomotive Acts required a person to walk in front of a car to ensure that everyone knew that danger was coming. Currently, we have essentially the same type of discussion on self-driving cars - as well as on voice interfaces like Siri and Alexa, which are useful, but also a bit creepy (particularly when there is a real chance that your private conversations are overheard).

Arguably, online advertisements have been around long enough for us to have gotten used to them. However, the ways in which users are targeted are often unknown or secret, new personalization and profiling methods are constantly being explored and exploited, and advertisers often seem not to adhere (sufficiently) to (current) social norms. Not all online advertisements evoke the same responses, though.

red flag act

We interviewed twelve Facebook users, asked them to scroll through their Facebook walls and to react upon advertisements that they encountered. The responses to creepy ads were quite predictable, but among the responses to non-creepy ads and ad explanations there were some surprises.

Responses to creepy ads

19 out of 45 advertisements that we discussed were considered creepy. Unsurprisingly, the overwhelming majority of them were attributed to unacceptable (perceived) data collection practices. For example, one participant said:

"I was telling my parents about it [Radler Beer] on Skype. And all of the sudden it showed up on my news feed on the next day.
I thought that was weird because I had not had typed it anywhere."

Surprisingly, almost all cases did not concern sensitive or embarrassing data, but were about everyday topics such as holidays, beer, shops and tickets. However, the users typically could not explain how Facebook got to know about their interests or activities, and thought that the system is not supposed to know about them; the only way 'Facebook could know' was thought to be by overhearing private conversations (which is not that far-fetched) or other unacceptable and unexpected practices.

The place where the advertisement is shown also plays a role. Placed in between other users' posts, page announcements and group conversations, the ads generally did not cause much anxiety, but when placed in between private messages... :

"[they] make me feel that Facebook is watching me while I'm talking."

facebook ad messenger

This advertisement in a private messaging window was considered creepy

Responses to non-creepy ads

As you might have expected, most advertisements that were considered non-creepy were also considered unrelated to the participants' (current) activities. A typical user response would be to ignore these ads (which could be considered an argument for displaying ads that are at least a bit creepy).

"I just scan and scroll through them. And I do not even pay attention, I just skip right over them."

However, if a specific advertisement would have been displayed often enough, users often got frustrated about this repetition and started to suspect that they were (incorrectly) being stereotyped:

"I searched for vegan things, it doesn't mean that I'm vegan. They are putting me in buckets!"

Only three advertisements were considered both harmless and beneficial, and therefore appreciated.

Responses to ad explanations

When you click top-right of an Facebook ad, you can select the option "Why am I seeing this ad?", which triggers an explanation as displayed below. The goal of these explanations is arguably to convince you that nothing creepy is going on.

why am i seeing this ad

Interestingly, almost none of the users were aware of the existence of these explanations. So we asked them to inspect some of them. A first impression was that most explanations were overly simplified and therefore not credible:

"I'm not sure if they are completely honest. For example, there is an ad about [a chili sauce]. It says people ages between 18 and 40 who lives in the Netherlands.
That's the only explanation and I think there should be more to it [why] they are targeting me."

Ironically, these types of explanations made the ads seem even more creepy than without an explanation. As a participant stated:

"I am also curious about what is not explained."

ad explanation

Implications

An obvious conclusion of the above results is: if advertisers want users to accept or even appreciate their advertisements, then they should ensure that they are not creepy. That is, if users request to know why a personalized ad was shown to them, the given explanation should be (or at least appear) to be credible, socially acceptable and... complete. The current Facebook ad explanations clearly do not fulfill the latter criterion and are therefore (rightfully) frowned upon.

Our finding that non-creepy ads are typically non-relevant - and therefore ignored - may seem an argument in favor of heavily personalized ads and user profiling. However, it is not. The advertising industry has several other means for ensuring that advertisements are relevant. For instance, cooking utensils are best promoted on recipe websites and surrounding cooking programs; hotel advertisements are most effective when users are browsing holiday destinations. Contextualization is the keyword and it is nothing new: the newspaper and television advertisement industry have successfully exploited this for decades.

To come back to the analogy with clowns: if targeted, creepy personalization is here to stay (and I am afraid that that's the case), perhaps the algorithms could try to throw in some humor: after all, a clown that makes you laugh is not considered creepy anymore.

clown street performance

This blog post is based on the full paper Unexpected and Unpredictable: Factors That Make Personalized Advertisements Creepy that will be presented at the ABIS 2019 workshop on 17 September 2019 at ACM Hypertext 2019 in Hof, Germany. The research has been carried out as part of the master project of Boping Zhang.

Eelco Herder

eelcoherder 256px

Privacy Engineering, User Modeling, Personalization, Recommendation, Web Usage Mining, Data Analysis and Visualization, Usability, Evaluation

Dr. Ir. Eelco Herder
Radboud Universiteit Nijmegen
Institute for Computing and Information Sciences
Toernooiveld 212
Mercator 1 - Room 03.01
6525 EC Nijmegen
The Netherlands

Email:
Phone: +31 24 36 52077

linkedin

facebook