With the advancement of technology, marketing has gained new tools for influencing consumers' decision-making processes. While in offline sales, it is easier to establish direct contact with each individual customer, the online environment offers a less intimate connection with the seller, necessitating the identification of pain points through indirect indicators. For example, in digital marketing, it is often necessary to test different ad creatives and combinations to enhance the customer experience and continually monitor metrics. However, for large companies, technology like Big Data and data science is increasingly assisting in this endeavor.
Thanks to modern, online-adapted behavioral psychology, it is now possible to delve into the minds, or rather, the subconscious, of users without their awareness. We will also try doing that by suggesting visiting our Telegram channel with tons of useful information on money-making. The legality and ethics of these practices remain open questions. Currently, psychological profiling of users is an expensive tool, enabling the promotion of political ideas and the influence of millions of minds. Let's delve into how this works.
What Is Psychometrics, and Why Is It Needed?
Psychometrics is a discipline that allows for the measurement of an individual's personality traits and the creation of a structured report based on the collected data. It is a promising field within the science of behavior. There is no exact date of origin for psychometrics, but some consider its beginnings in the 1870s when German physiologist and psychologist Wilhelm Wundt aimed to measure human sensations' intensity. For others, the starting point is 1905 when the first IQ test, the Binet-Simon scale, was introduced. Initially, the primary focus of psychometrics was to measure students' intelligence and knowledge using tests. Over time, this trend developed further in the United States, where tests are still widely used as the gold standard in education and psychology research. Gradually, the practice of testing and measuring results spread worldwide. Today, leading countries setting trends in psychometrics include the United Kingdom, the Netherlands, and Belgium.
With the development of digital technologies and machine learning, psychometrics has evolved into the science of behavior as a whole. Its boundaries have become blurred. Today, psychometric methods are used in sociology, medicine, and marketing. In all these fields, researchers draw conclusions about human behavior based on structured data.
How Psychometrics Is Used in Marketing
The new wave of attempts to manipulate consumer behavior in the digital era began with the emergence of Big Data. However, this approach has shown several significant drawbacks. Algorithms for analyzing large data sets are still imperfect, and predicting user behavior based on them is a challenging task. Only giant corporations like Google and Amazon have been successful in handling such tasks so far, as most companies lack access to extensive data. While data can be purchased, questions arise regarding data quality, completeness, and processing capabilities.
A deeper approach to understanding consumer motives was proposed by Michal Kosinski, a British scientist of Polish descent. He was the first to develop digital psychometrics, allowing the determination of personality traits based on a user's behavior on Facebook. Starting in 2010, Kosinski studied Data Science, initially as a member of the Psychometrics Centre at the University of Cambridge and later as a professor at Stanford Graduate School of Business from 2015 onwards. In 2016, he conducted an experiment on Facebook users involving psychotargeting, giving rise to a new marketing tool that could uncover hidden motivations. This marked an entirely new level of personalized marketing that brands had been aspiring to since the dawn of the digital age.
Michal Kosinski
Based on a user's digital footprint, it is possible to determine their psychotype, gender, sexual orientation, skin color, political views, and other characteristics. Digital footprints include likes on specific posts, shares, comments, browsing history, and search queries. Michal Kosinski claims that it takes as few as 10 likes for a machine learning system to know you better than a colleague at work. With 230-240 likes, the computer can know you better than your spouse. All these data belonging to millions of users can be bought or sold on the black market, a resource actively utilized by companies. There is a project called Acxiom that collects vast amounts of user data worldwide. You can order information, for example, about unmarried men aged 25-40 in Massachusetts or women aged 18-32 in the Indian state of Chandigarh.
Moreover, you can independently gather data on users' psychological profiles to launch a campaign, such as on Facebook. To do this, you need to build a psychological model of users inclined toward a specific type of behavior by paying thousands of users to take a specially designed test. Once the model is ready, you can instruct artificial intelligence to identify, for instance, introverts, who may like 12 specific movies or musicians. Now, you can compile a database of users who like specific posts and target them with personalized advertising.
But that's not all. Digital psychometrics also allows for precise predictions of user behavior, surpassing the effectiveness of target audience surveys, focus groups, and testing.
Psychometrics as a Service: Who and How Analyzes Our Digital Footprints
There are already companies employing psychometrics to determine users' personality traits and selling the acquired data to businesses. Artificial intelligence, based on classical models of psychological research, is utilized to work with such large datasets. The foremost among these models is the OCEAN model, or the "Big Five," which allows the assessment of five key personality traits:
- Openness to experience
- Conscientiousness
- Extroversion
- Agreeableness
- Neuroticism
The second psychological model is Schwartz's value orientation. These two models have gained wide acceptance in the academic world and have stood the test of time.
There are also psychometric solutions for data analysis based on new psychological models specifically developed for marketing applications. However, their effectiveness has not been fully explored as of yet.
Simplified Scheme of Psychometric Profiling Based on Facebook Data
One of the popular services available to marketers was IBM's Watson Personality Insights, which has since been discontinued. This tool allowed the creation of a psychological profile of an individual based on their texts. It employed the OCEAN model to determine personality traits and Schwartz's model to assess motivation levels. Custom models, developed by the corporation, were used to identify personal needs. The service provided case studies on its website illustrating its application for marketing purposes.
Psychographic Profile Created Using IBM's Solution
The Apply Magic Sauce service, created by Cambridge researchers, is another tool that utilizes Michal Kosinski's work. Obtaining a user's psychological profile on this platform is straightforward; you simply input their Facebook, Twitter, or LinkedIn profile. The trusted OCEAN model is applied to determine personality characteristics.
In the realm of digital psychometrics, there are also partially domestic developments. For instance, the Russian-British service Datasine, active since 2015, has assisted in personalizing marketing campaigns. At various times, its clients included Aegon Insurance, Tinkoff, Raiffeisen, and BNP Paribas. In 2021, Shutterstock acquired the service.
Psychometrics and Politics: Why You Don't Make the Choices
Surprisingly, psychometrics plays a pivotal role in shaping the destinies of entire nations, as it is actively employed in politics. It all began with the company Cambridge Analytica, whose former director, Alexander Nix, claimed in March 2017 that they had managed to obtain a psychometric profile of every adult resident of the United States. This is hardly surprising, considering that Cambridge Analytica had assisted Donald Trump in the 2016 elections by collecting data from Facebook, analyzing it, and using the insights to create targeted advertisements. The sum received from the then-future president was reported to be $15 million. Similarly, in 2016, Cambridge Analytica worked in the United Kingdom, supporting Nigel Farage, a staunch Brexit advocate. These efforts were instrumental in manipulating millions of British voters, as reported by the Swiss magazine Das Magazin in December 2016.
Interestingly, Das Magazin also suggested that Cambridge Analytica had used developments closely resembling Kosinski's research. Allegedly, one of Kosinski's colleagues had leaked his research findings to the company. Consequently, Kosinski gained international fame and faced allegations of creating mass psychological weaponry. Cambridge Analytica later entered contracts with the French National Front and operated in elections in at least 20 countries. It has since gone bankrupt.
To influence user minds effectively, it is crucial not only to gather and structure data but also to apply the results correctly. For instance, Cambridge Analytica specialists, while working on Trump's campaign, identified subtle connections between personality traits and user behavior. They discovered that individuals who preferred American-made cars tended to support their clients. In one interview, Alexander Nix explained that if a group of users exhibited high conscientiousness and neuroticism on the OCEAN scale, they would be shown an ad designed to evoke emotions by playing on fear of non-compliance and rationality. This approach infused the advertisements with emotional nuances that would resonate with the subconscious motivations of a specific user group.
Header from Experian's website, a service that helps influence voters' minds using psychographics. The company's clients included both the Labour and Conservative parties of the British Parliament.
Application Cases of Psychometrics in Marketing
GutCheck, a company specializing in big data research, provides an example of working with psychometrics on their website. The task was to create effective advertisements for new yoga pants. After collecting and analyzing data, two groups of women were identified: those with high extroversion levels and those with low extroversion but high conscientiousness. Below are examples of advertisements for the first category (on the right) and the second category (on the left):
Michal Kosinski and his colleagues conducted a marketing study demonstrating the impact of advertisements crafted based on users' psychological profiles on their behavior.
The team divided their work into three Facebook studies, displaying ads to a total of 3.7 million users. Studies 1 and 2 aimed to assess people's reactions based on their extroversion and openness to experience levels according to the OCEAN scale. To study users, researchers extracted data on Facebook* likes from several million individuals in the http://mypersonality.org/ database. Scores from users after completing the International Personality Item Pool (IPIP) questionnaire, consisting of 100 items, were also obtained from this database. Based on this data, scientists calculated individuals' personality traits and divided them into groups. Study 3 was based on the results of studies 1 and 2.
In Study 1, the results of Facebook advertising campaigns targeting women with high and low levels of extroversion were presented. The campaign promoted cosmetics, and the ads ran for 7 days. The campaign reached 3 129 993 users, attracted 10 346 clicks, and led to 390 purchases on the cosmetics store's website. Here are the ads used:
On the left are ads for women with high extroversion, and on the right are ads for women with low extroversion.
Here is the campaign statistics:
CPConv = cost per conversion, CR = conversion rate (installs/reach × 100), CTR = click-through rate (clicks/reach × 100), ROI = return on investment (profit/expenditure × 100).
In Study 2, campaigns were created to promote a crossword puzzle application. The impact on people with high openness to experience and low openness to experience was compared. The campaign ran on Facebook, Instagram, and Audience Networks for 12 days. It reached 84 176 users, attracted 1 130 clicks, and resulted in 500 app installations.
Here are the creatives used for people with different levels of openness:
On the left is high openness to experience, and on the right is low openness.
Here are the results after completing the campaign:
In Study 3, researchers promoted a bubble shooter game on Facebook for 7 days. The target audience for the campaign was compared to the user base of competitors' apps (Farmville or Bubble Popp). Analysis of likes from their users revealed that they were highly introverted individuals. This time, researchers selected a larger group of such users and divided them into two parts. One group was shown an ad with standard text for extroverts, and the other group was shown text tailored to introverts. In cases where the ad was adapted to the psychological profile, CTR and conversion rates were 1.3 and 1.2 times higher, respectively.
Here is the campaign's statistics:
In total, the campaign reached 534 250 users, attracted 3 173 clicks, and resulted in 1 837 app installations.
Conclusion
Like all significant inventions, psychometric profiling carries both great benefits and risks. Michal Kosinski claimed in one of his interviews that his methodology can help people find jobs that match their inclinations and psychotypes more accurately. Moreover, these developments can be utilized in psychological assistance, and employee selection for government agencies and private companies. According to Kosinski, psychometric ad targeting in marketing can double campaign effectiveness.
However, psychometrics has already become a tool for mass manipulation. It can deter people who are uncertain about the fairness of elections from participating in politics. This market is currently not fully developed, and it is likely to undergo government regulation in the future, similar to the regulation of the Internet. Time will tell.
