The Bizarre Accuracy of Online Ads: A Case Study of Trackers on the Internet
A call for transparency and consent
A common theme of dystopian literature is a central panoptic instrument whose gaze never departs from its subjects. An example of this omniscient mechanism might be a technology that keeps a detailed track of one’s behavior, and uses its computational power to predict and influence one’s future actions based on past behavior. Although they may not seem as malevolent, modern web trackers and cookies are a realization of this idea. Our digital footprints are, at least in all meaningful ways, indelible and permanent. Our digital traces are often captured in detail, tracked without our positive consent, and used in an attempt to influence us.
As average Internet users, we have some vague sense understanding of websites tracking our behavior. We accept this reality as a nebulous fact, knowing full well that this disturbing fact will never motivate us enough to forgo the conveniences of this or that web service.Thus, we are seldom told who collects our data and how our data is used. All we are told is that websites are not using our data for malicious purposes. Furthermore, it is said, it is unreasonable for users to expect a free service with no way for the companies to make money.
Cookies and trackers, we are told, are useful tools that we should embrace to ensure users meet the right products in advertising. Even if we care to investigate, most of us are discouraged by the technical details — such as how internet service providers interact with cookies and how websites can steal data that we are seemingly volunteering. We find ourselves trapped in a world that we do not understand, occasionally surprised by how accurate an online ad is, but then moving on with a shrug.
Tracking services are tools that collect data on a user’s browsing activity on the web; any third-party can insert a tracker into a website with the website’s permission. Without any more technical details, it is possible to see what trackers are embedded in a given website. As a part of Harvard University’s Techtopia program, housed at the Berkman Klein Center, we carried out a case study of which companies embedded trackers on two popular websites. We made use of the browser plug-in Privacy Badger, which lists and selectively blocks the trackers embedded in a website. The examples are pulled from the websites of Fox News and Easy Bib.
What follows is a discussion of the sheer number of trackers embedded in each website, the various types of trackers that range from the largest corporations to a long tail of lesser-known companies, and a whole structure of specific ad tracking services that are disturbingly effective.
Case 1: Fox News
It is no surprise that each of the technology giants follow our every move, on pretty much any website we browse. Amazon, Microsoft, Adobe, and Salesforce all have trackers on the home page of foxnews.com, and the same is true for most other news outlet websites as well. These serve a range of purposes, although all related to advertising. For example, Adobe’s and Amazon’s trackers collect general analytics for targeted advertising purposes, whereas Microsoft’s tracker tracks how often the ads they have put up on the website are clicked by users — a concept called “ Ads Conversion Rates. “ This measures the effectiveness of the ads so that companies can optimize their strategies for maximal engagement.
More surprising is the numerous lesser-known services we encounter, such as Optimizely, Nielsen, and Taboola. Although the eventual target of each of these services is efficient ad-targeting, the specificity of their methodologies is unsettling.
Optimizely is a company that does A/B testing on the users that visit a website. A/B testing is systematically altering the content presented to two groups of users in order to establish which presentation serves the service’s purposes better. We may think of A/B testing in the realm of psychological experiments or focus groups that are employed for optimization research, but Internet users are subject to A/B testing in their daily experiences in ways that are not made transparent to them.
Even though we cannot trace what types of A/B testing is done on foxnews.com, we can predict the general structure. For example, by modifying the placement of the ad in the website or the photo used with the ad, the service might try to tap into different types of diversions users are likely to fall into while reading the news. Thus, they can increase the click-ratio of the advertisers they serve, while we, as the user, see nothing more than a number of ads that we try our hardest to keep from distracting us.
Similarly, Nielsen Company collects data to categorize the users of a website into demographics that make it easier for advertisers to target their content. Their home-page lists the hundreds of user groups they have data on:
“Know what’s next in Asian American consumers / LGBT consumers / Hispanic consumers /etc.”
Nielsen Company is not at all new; it was founded almost a century ago as a market research firm. Most of us are familiar with the Nielsen ratings, which measure what demographics groups watch which TV programs. However, in 2015, Nielsen made a strategic move to specialize in the collection and distribution of digital data, including website use. Through the use of web trackers, we can reasonably predict that Nielsen matches each visitor in the website with a demographic group and checks their behavior against the trends in that demographic group. Thus, they can give advertisers suggestions as to what types of ads to display for different user groups.
While we might have accepted the collection of demographic data from television audiences, as Internet users, we are subject to much more detailed inspection. Whereas TV trends could tell advertisers which demographics groups are likely to watch what types of TV shows, web trackers can tell which articles, which visuals, and which fonts affect each user group’s behavior. Furthermore, what constitutes “demographics” is much more specialized. Nielsen’s “solutions” include a detailed analysis of a user’s behavior on a website to extrapolate all kinds of properties; from age group to nationality, from dietary preferences to political affiliations. Although we do not have access to its parameters, it is an undoubtable fact that every time we are online, we are subject to exhaustive and detailed demographic categorization.
Another service is Taboola, which curates the sponsored article sections in content aggregators. Familiar to the average user as the “Recommended for you” or “Suggested articles” columns, native advertising is a strategy to engage readers who have a gut-reaction to avoid advertisements. By concealing themselves as native content, these ads can engage the user who mistakes them for content curated by the website itself. By the time the article reveals itself to be an advertisement, the user might already be engaged enough to show interest in the product.
Native advertising has received a considerable amount of criticism from those who claim that it is seeks to intentionally deceive users who take themselves to be browsing native content. Even when such ads are separated from the content by a disclaimer, their formatting always matches the native content, furthering the possibility of a confusion. Furthermore, upon clicking, the ads are worded similar to the actual content in the website, so users might not notice that they are consuming sponsored content until the product has been shown, which is generally towards the end of the article. Since this advertising method benefits from the confusion, it is difficult to suggest that the “sponsored content” label itself is enough to alert users. A 2016 study shows the weight of this problem: Out of over a thousand participants, more than half reported feeling deceived upon realizing that an article they mistook for native content was sponsored, and 77 percent did not even interpret native ads as advertising.
Case 2: Easy Bib
Once we move away from news outlets, regulation on advertisement services decreases. A tool that helps academicians create bibliography citations for their work, Easy Bib, allows third party trackers that are more uncommon: These trackers include Beachfront Media, which specializes in curating videographic advertisements, and Study Break Media, which describes its focus as “teenager engagement.” Of course, larger analytics tools that collect general purpose data and sell it to a host of customers are not absent as well — Qualtrics has a tracker embedded to collect information on what users click, how much time they spend on the website, and anything else that might capture the users’ “beliefs, emotions and sentiments.”
Many of the trackers we have seen so far function by matching users’ general interests with other products. A method that forces the limits of specificity is called retargeting, employed by third party trackers such as those of Criteo.
Retargeting is a marketing technique in which the user is presented with an advertisement for a product that she previously perused, with the hope of re-engaging the user. For example, if you click on a pair of shoes on Amazon but eventually end up not going through with the transaction, a retargeting application is tasked with finding you and making sure you see the same pair of shoes as an ad in a completely separate website. This is one of the methods behind the feeling of slight paranoia we feel when we come across a product we perused weeks ago advertised on a completely different website. The ad might as well say: “Hey, I know you want me — c’mon now, click on me!”
Similar to native advertising, retargeting is condemned by certain critics on ethical grounds, suggesting that it comes dangerously close to “being stalked” by a persistent salesman who will not leave you alone — A salesman that can come with you wherever you go, and however long it has been since you decided against the purchase. To the average Internet user, it is not common knowledge that the ads served by separate websites are actually interconnected. However, no matter how we feel about a product’s ability to pursue us for weeks regardless of what website we are browsing, we are subject to retargeting on a daily basis.
Finally, perhaps the most futurist technique of advertisement we have come across in this case study belongs to TripleLift’s branded videos. TripleLift promotes this exciting technique to their possible customers as follows:
“Showcase your video in the most optimal placements, directly within the feed of content on the world’s best publishers.”
In other words, this service allows advertisers to place their product inside the video the user is watching. This is not referring to an advertisement that plays before the video; the service can directly alter the content of native content. Thus, it is an that targets individual users by placing a product inside a video that does not originally include such advertising.
For example, using brand insertion, a viewer watching a TV show on Comedy Central might see a Tide ad in a scene, while another might see a Galaxy S8. Furthermore, through product insertion, a scene of Supergirl might include a shot of this or that alcoholic beverage, even though there is no beverage in the actual scene. Even though we cannot confirm that we have seen these uses of TripleLift in Easy Bib, the methods are actively advertised on TripleLift’s website for all to see. At a time when referencing TV dystopias has become commonplace for tech and society articles, we cannot help but join the crowd — This is literally Black Mirror.
A Way Forward
Today, there is an active battle between advertisers and those concerned with the range of data collected, and the methods used for ad targeting through web trackers. Services such as Privacy Badger and Disconnect are able to catch and even block many of these trackers. Ad blockers are generally successful in detecting ads on the web and blocking them from individual users.
Nonetheless, some trackers are more persistent and ingrained in the necessary functions of the website, so disabling them crashes the websites. Similarly, websites are in an effort to get ahead of ad blockers by disallowing users to access the website if they use ad blockers.
Perhaps data collection and targeted ads are necessary for these websites to make a profit while allowing us to use their services for free, but transparency is key. Without it, we are entering an agreement without knowing what it really entails. For the ad-based model to remain sustainable, we need the industry to develop a culture in which they are open and transparent — transparent about which third parties are with us as we browse their website, how they use the data they collect about us, and what options are available to us if we do not want to be tracked by a given tracker.
Any other solution, such as the technical war waged against advertisers, is bound to be only a temporary measure. As tools to block trackers advance, so will the ability of digital marketing firms to bypass these obstacles. While this fight is ultimately futile, it is completely within the power of these websites to discern between ad services, and public perception can force them to consider protecting the privacy of their users.
On the other hand, unless we aggressively start questioning these practices — questioning when are we being followed, by whom, and for what purpose — then, we are in a de facto agreement with these advertisers to continue this status quo. We are kept out of the loop more and more in a decision that concerns us, the users of the Internet, most intimately. And so we let websites and advertisers have negotiations concerning our data, make bids for it, analyze it and exploit it to whatever end they see fit — Without our active consent, or even knowledge of it.
Acknowledgements: This article is written as part of Techtopia’s Know More project. This article would not have been possible without the contributions of the members of Know More: Seowoo Kim, Lavanya Singh, Chelsea Han, Josh Feldman and Anissa Abdel-Jelil. Furthermore, we would like to thank Prof. Jonathan Zittrain for his mentorship, and Hannah Hilligoss for her constant support. Finally, we are evermore grateful for the countless exceptional opportunities provided to us by the Techtopia Community and the Berkman Klein Center.
Originally published at https://medium.com on May 15, 2019.