Target Advertising as a Surveillance Practice

Targeted advertising is now a ubiquitous tool in the online world. Many tech companies use user-provided data to create categories of individuals (based on personality and other factors) to better target advertising banners and ensure they reach specific objectives, ultimately achieving the goal of the sale. The technological innovation of micro-targeting is often underestimated, but it has truly changed, if not revolutionized, the economic system we live in.

Target advertising: the market for future behaviors

Addressing targeted advertising means analyzing, on the one hand, the business models of the richest companies on the planet and, on the other, trying to shed light on the potential implications that these methods of accumulation have for us.

Understanding our behaviors and trends in terms of consumption, purchasing, and lifestyle allows target advertising companies (e.g., Alphabet, Meta) to reduce advertising failure and thus grow financially. Our behaviors are then analyzed and studied. In the online world, all data is collected and managed through data mining. This practice consists of the process of exploring, analyzing, and modeling vast amounts of data to identify relevant patterns, trends, and correlations.

Data mining and algorithms

Using a combination of statistical techniques and advanced algorithms , data mining allows you to extract valuable insights from data , thus improving your chances of achieving strategic and business outcomes. This practice is essential to fully exploiting the potential of available data, enabling organizations to improve their performance, optimize processes, and anticipate market needs. In short, data mining transforms complex data into actionable insights, promoting business progress and competitiveness.

At this stage, algorithms require a huge amount of data to obtain qualitatively better information. Indeed, the basic criterion of this process is: quality is directly proportional to quantity. The goal is to achieve the mathematical success of an advertisement: sending a banner at a given time, in a given weather condition, in a given emotional state, on a given day, with a given battery level, etc. Being able to predict with a minimal margin of error when I will need a given item or service is the new challenge of the contemporary economy. The data demonstrates this; Amazon, for example, is the largest research organization in the world. In fact, in 2018 it invested $22 billion in research and development, with the production of 2000 patents in 2019. To give you an idea, the Massachusetts Institute of Technology (MIT) invests an average of $3 billion per year.

A funnel-shaped target advertising

Many of these research investments are tied to the development of predictive shipping mechanisms, such as delivering a package to your location before you've even purchased it, guaranteeing a delivery time of half an hour or an hour if purchased (Delfanti 2023). Data mining is the essential prerequisite for producing predictive products, that is, analyzing our behavior that allows, through data mining, to infer our future behaviors, thanks to the raw material of this process: our behavior.

In the online world, we are exposed to two realities that Zuboff calls "texts." The first is what we see and create every day online; the second, the shadow text, is what we don't see, where all our digital traces are assembled and where a group of data scientists study ways to predict our future behavior. Once our behaviors are analyzed and studied, they move on to targeted advertising, also known as micro-targeting.

What is microtargeting?

Targeted advertising is, as Shoshanna Zuboff defines it, "the holy grail of advertising" (2019;88) as it allows advertisers to drastically reduce the failure rates of their banner ads. The birth of this type of advertising dates back to Google, which between 2000 and 2001 patented its first software capable of personalising ads thanks to a user tracking system that recorded all past searches and created a UPI (user profile information) for this purpose, which contained all the user's personal traits (Google 2003).

This information was obviously deduced from the data. Therefore, the ads would no longer be linked to the words entered by the user in the search bar, but would find another path: to connect with the individual's personality traits.

The main and richest companies that have pioneered and benefited from this new business model are Alphabet and Meta (which for simplicity I will call Google and Facebook). In fact, in the first four months of 2016, 89% of Google's revenue and 96,6% of Facebook's revenue came from advertisers (Srnicek 2017).

Target advertising: Google in detail

Google was founded with the aim of organizing all the world's information and making it available to everyone. User-generated data was initially reinvested in service optimization, a process defined as the "behavioral value reinvestment cycle" (Zuboff 2019, 79). With the financial crisis of the 2000s that hit the entire technology sector (the so-called dot-com bust), many tech companies closed and Google risked bankruptcy. The only way to survive was to change its business model. Thus, the aforementioned software was created, which led the company to increasingly profitable growth. To better illustrate the true extent of the change Google achieved in terms of profits with targeted advertising, the following table shows earnings from 2000 (the year the software was introduced) to 2004.

Targeted advertising is truly a source of invaluable wealth for digital companies. Data is often compared to oil, and tech companies are like the new oil tankers of the twenty-first century.

Thanks to predictions about our behavior, derived from data mining by tech companies that exploit them, advertisers can influence our behavior given the constant matching between who we are and what we might or might want to buy. Indeed, beyond talking about us, they can be used against us, even unconsciously influencing our choices (Barassi 2021).

A new economic system?

The business model that characterizes digital platforms is for many something completely new, unprecedented in history, and therefore we must approach it with new lenses, without deceiving ourselves with obsolete terminology such as those who called the first automobile a "horseless carriage," because they do not help us understand the phenomenon. Shoshanna Zuboff's essay: Surveillance Capitalism (2019) can be interpreted as a manual that contains new concepts that transcend interpretations tied to the past, thus managing to map this new, still unexplored world.

For example, the terms "extraction imperative"; "behavioral surplus"; "the cycle of expropriation"; "radical indifference"; "machine intelligence"; "instrumental power," "rendering," etc., are just some of the new concepts Zuboff offers us to understand the phenomenon. This new language helps us interpret the "dynamic principles" of this system that "extracts" our data for survival. The action that allows for profit is called "behavioral surplus extraction," which derives from the use of data that is no longer reinvested in improving the service, as it once was, but used for targeted advertising. Once upon a time, we were the ones doing searches on Google; today, the opposite is also true.

Free contract or expropriation?

Obtaining this data is defined as "expropriation" or "theft" (Sadowski, 2019), precisely because of the asymmetry of power that occurs when we are "forced" to accept the terms and conditions of use, actual contracts, also called "click wrap" because everyone, or most, clicks the "accept" button. But this consent does not stem from a willingness to accept the agreement to transfer our data, but because very often we have no other choice. These contracts are called EULAs (end-user licensing agreements).

They are nearly impossible to read and understand quickly; in fact, one study concluded that it would take 76 days, working 8 hours a day, to read the privacy policies a person typically encounters in a year (Sadowski 2019). Recently, we've noticed more and more websites offering the "accept" option alongside the "do not accept" option , even though this latter button is often red, very small, or requires the user to click several steps to activate. Obviously, not being able to take advantage of tracking represents a clear loss for those who might collect them.

Target advertising towards "I'll show you who you are"

Demonstrating how a careful analysis of our data can reveal who we are, a 2011 study by researchers Jennifer Golbeck, Cristina Robls, and Karen Turner (2011) demonstrates how, thanks to data spontaneously released on Facebook and using Robert R. McCrae and Paul T. Costa's Big Five personality theory, it is possible to deduce our personality, what kind of person we are, and, consequently, what kind of advertising might be best suited to people like us. The researchers conclude that, with the results in hand, they were able to train two machine learning algorithms capable of predicting the subjects' personalities.

In the vast landscape of psychology, one of the most interesting and debated concepts is that of the Five Personalities , commonly known as the "Big Five." This model provides a detailed and nuanced framework for understanding the complexity of the human being, exploring the key dimensions that characterize each individual's personality. Let's explore it together.

What is the Five Personality Theory for Target Advertising?

The Big Five Personality Test, based on the Big Five theory, has established itself as a fundamental tool for psychologists and researchers of human behavior . It aims to assess five fundamental personality traits, providing a detailed overview of an individual's characteristics:

• ExtroversionThis trait refers to the degree to which a person is sociable, energetic, and inclined to interact with others. Extroverted individuals tend to be more open and communicative, while introspective individuals may be more reserved and thoughtful.
• PleasantnessAgreeableness refers to a person's propensity to be altruistic, friendly, and cooperative toward others. Those high in agreeableness tend to be compassionate and helpful, while those low in agreeableness may be more aloof and competitive.
• ConscientiousnessThis trait indicates an individual's level of organization, discipline, and responsibility. Highly conscientious people are generally more meticulous, reliable, and success-oriented, while those who are less conscientious may be more disorganized and irresponsible.
• Emotional stabilityEmotional stability refers to an individual's ability to manage stress, anxiety, and negative emotions. People with high emotional stability are more calm, confident, and resilient, while those with low emotional stability may be more vulnerable and susceptible to mood swings.
• Open-mindednessThis trait reflects an individual's openness to new experiences, ideas, and cultures. People with high openness are more creative, curious, and flexible, while those with low openness may be more conservative and traditionalist.

What can be understood from this test?

Through the Big Five Personality Test, you can obtain an in-depth assessment of each of these traits , allowing for a better understanding of an individual's behavior and preferences. It's important to note that the test is no longer necessary if our data and characteristics are already online; an algorithm trained to infer personality based on available data takes care of this.

Steve Bannon and Alexander Nix already understood this when they founded Cambridge Analytica and decided to initially use this test and then microtargeting in politics, influencing millions of voters around the world, particularly in the United Kingdom for the Brexit elections (serving the LeaveEU) and the 2016 US presidential elections (serving the Republican Party). Alexander Nix said at a summit: "We can predict the personality of every user because personality controls behavior, and behavior influences our vote" (The Power of Big Data and Psychographics, Concordia Annual Summit, 2016).

Research by Jennifer Golbeck, Cristina Robls, and Karen Turner has demonstrated how it is possible to predict a person's personality by analyzing the data we post online every day (in their case study, specifically, Facebook).

Target advertising and sociology

Sociology is shedding light on the dark sides of this new data collection system, charting the course for change. The European Union is taking action. A lengthy negotiation with major target advertising companies like Meta is still ongoing. Not too many months ago, we all witnessed the clash between Meta and the European Union over the new paid model for services like Facebook and Instagram, which were ad-free (the major stumbling block was micro-targeting).

To protect itself and simultaneously demonstrate the value of its users' data, Meta offered a subscription of around €13 per month for ad-free access to its services. As if privacy could be bought. Privacy is a right, and the European Union, from the General Data Protection Regulation (GDPR) to the new Digital Services Act (DSA), is striving to protect its citizens while also maintaining the sustainability of this new market.

The stakes are enormous . For some companies, giving up data means giving up their future survival. The real challenge is understanding, on the other side, what users—that is, us—are giving up.

Andrew Leonardi

References

• Barassi V. 2021. Children of the Algorithm, Luiss University Press, Rome.
• Delfanti A. (2023). The warehouse: work and machines at Amazon. Turin, Codice.
• J. Golbeck, C. Robles, and K. Turner. Predicting personality with social media. In Proceedings of the 2011 Annual Conference Extended Abstracts on Human Factors in Computing Systems, pages 253–262.ACM, 201
• Morozov E. 2016. Silicon Valley: the Lords of Silicon, Codice Edizioni, Turin.
• Sadowski J. 2019. “When data is capital: Datafication, accumulation, and extraction”, Big Data & Society, 6,1.
• Srnicek N. 2017. Digital Capitalism, Luiss University Press, Rome.
• Zuboff S. 2019. Surveillance Capitalism, Luiss University Press, Rome.
• Google Patent (2003) = https://patents.google.com/patent/US20050131762A1/en