10 Times Big Business Turned People into Test Subjects

Long before websites could monitor every click, companies were already studying how people behaved at work, in stores, and in front of advertisements. Some experiments were straightforward attempts to improve products or working conditions. Others crossed into private lives, manipulated what people saw, or quietly tested techniques designed to influence their decisions.

The most alarming stories do not require invented torture chambers or imaginary mind-control laboratories. Real businesses have altered employees’ surroundings, analyzed audiences’ faces, inspected workers’ homes, and built digital systems that encourage people to keep working or spending. Sometimes the participants understood that they were being studied. In other cases, millions of people became experimental subjects without realizing it.

These ten documented cases show how readily the laboratory can expand into the factory, movie theater, dating profile, voting booth, and smartphone in your pocket.

Related: Ten World-Famous Businesses with Surprising Backstories

10 Google Tested 41 Shades of Blue

When Google wanted to choose the most profitable color for its advertising links, the company did not settle the matter in a design meeting. It reportedly showed users 41 slightly different shades of blue and measured which versions received the most clicks. Each visitor saw only one variation, turning ordinary searches into a vast behavioral experiment.

The test became famous after designer Douglas Bowman left Google in 2009 and criticized its dependence on data. He recalled a dispute over two shades of blue that led the company to test 41 alternatives instead of trusting its designers. Google executive Marissa Mayer later said that adjusting the color of advertising links added approximately $200 million in annual revenue.

No one was psychologically harmed by encountering a marginally different shade of blue. Nevertheless, the experiment demonstrated the power of testing minute changes across enormous audiences. A color difference too subtle for most people to describe could still alter their behavior often enough to generate a fortune. Google had turned aesthetic preference into a measurable—and highly profitable—human reflex.^[1]

9 Disney Trained a Computer to Read Moviegoers’ Faces

Disney has long used test screenings to learn whether audiences laugh, cry, or become restless at the intended moments. In 2017, Disney Research revealed a system that could perform part of that work automatically. Infrared cameras recorded the facial movements of 3,179 people during 150 screenings of movies including The Jungle Book, Zootopia, and Star Wars: The Force Awakens.

The resulting dataset contained approximately sixteen million individual observations. Researchers used it to train an artificial intelligence system to recognize patterns in the audience’s facial reactions. After observing viewers for part of a film, the program could predict how their faces were likely to respond during later scenes.

Moviegoers were not wired to heart monitors, nor did Disney claim to have a mathematical formula for forcing tears. The real technology was subtler. Instead of asking viewers whether a scene worked, Disney could examine involuntary smiles, frowns, and other movements across an entire auditorium. Test audiences had become live streams of emotional data.^[2]

8 Western Electric Kept Changing Workers’ Conditions

Beginning in 1924, Western Electric turned part of its Hawthorne Works factory near Chicago into one of history’s most influential workplace laboratories. The earliest studies tested whether brighter lighting improved productivity. Researchers altered the illumination for selected workers and compared their output with that of control groups. Productivity sometimes rose when the lights became brighter—but also when they became dimmer.

Later experiments moved small groups of women into special test rooms. Investigators changed rest breaks, working hours, payment systems, supervision, and access to refreshments while recording how quickly the women assembled telephone equipment. Researchers also conducted thousands of employee interviews to understand morale and workplace relationships.

The studies eventually inspired the expression “Hawthorne effect”: the idea that people change their behavior because they know they are being observed. Later scholars disputed that tidy conclusion and questioned both the methods and interpretation of the original data. Even so, the experiments permanently changed management theory. Workers’ attention, relationships, and awareness of observation had become variables employers believed they could measure and manage.^[3]

7 Ford Investigators Inspected Employees’ Private Lives

In 1914, Ford Motor Company introduced its celebrated five-dollar workday. The offer sounded revolutionary, but not every worker automatically received the full amount. Part of the money came through a profit-sharing program, and Ford expected recipients to satisfy the company’s standards of respectable living. Its newly created Sociological Department was assigned to determine who qualified.

Company investigators visited employees’ homes and questioned workers, spouses, and neighbors. They examined cleanliness, savings, debts, drinking habits, marital arrangements, and how families spent their leisure time. Immigrant employees were encouraged to study English and adopt Ford’s preferred version of American domestic life. Those who failed to meet the department’s expectations could lose access to the additional pay.

Ford presented the program as benevolent reform intended to help workers become stable and prosperous. Yet it effectively made private behavior an employment qualification. The company was testing whether higher wages, surveillance, and the threat of financial loss could reshape employees’ lives outside the factory. Ford eventually abandoned the Sociological Department, but its experiment foreshadowed later corporate efforts to monitor and modify behavior far beyond the workplace.^[4]

6 Uber Used Psychological Nudges to Keep Drivers Working

Uber advertised driving as flexible work: drivers could log on and off whenever they wished. Behind the app, however, the company studied how interface design and behavioral psychology could influence when they stopped. The driver app offered earnings goals, congratulatory messages, badges, and alerts that emphasized nearby demand. It could also queue another passenger before the current ride ended, reducing the natural pause when a driver might decide to quit for the day.

One technique resembled the automatic start of the next episode on a streaming service. Rather than actively choosing to continue, drivers had to interrupt a process that was already moving them toward another fare. Other messages informed drivers that they were close to reaching a self-selected earnings target, exploiting the reluctance to stop just short of a goal.

None of these features physically prevented drivers from going home, and Uber argued that its notifications supplied useful information and helped drivers reach their own targets. The concern was the imbalance of knowledge. Uber possessed detailed information about demand, earnings, and individual behavior, while drivers saw only the prompts selected for them. The app was simultaneously their workplace, supervisor, scoreboard, and behavioral experiment.^[5]

5 OkCupid Lied About Romantic Compatibility

In 2014, OkCupid co-founder Christian Rudder published a blog post with the startling title “We Experiment on Human Beings!” He described several tests conducted on the dating site’s users. For one experiment, OkCupid temporarily removed profile photographs. In another, it hid profile text to see how much users’ ratings depended on appearance.

The most controversial test manipulated the compatibility percentages shown to potential couples. Some users who were genuinely rated as poor matches were told they were highly compatible. Others who supposedly matched well were shown much lower scores. OkCupid then measured whether the pairs exchanged messages and continued their conversations.

The results suggested that the displayed score could influence how people treated one another. Poor matches who were told they were compatible behaved more like good matches, while highly compatible users communicated less when shown a low score. OkCupid didn’t merely observe attraction; it altered expectations to see whether belief in an algorithm could help create the connection it claimed to predict.^[6]

4 LinkedIn Quietly Altered Users’ Employment Networks

Between 2015 and 2019, LinkedIn conducted a massive experiment through its “People You May Know” recommendation system. The company adjusted the algorithm so that some users received more suggestions involving strong social connections, while others were encouraged to connect with people they knew less well. Approximately twenty million users were included.

Over five years, those users formed around two billion new connections, submitted more than seventy million job applications, and accepted approximately 600,000 jobs identified through LinkedIn. The experiment allowed researchers to test the “strength of weak ties” theory—the idea that acquaintances can provide opportunities unavailable through close friends whose social circles resemble our own.

Moderately weak connections proved especially valuable for finding employment, although the results varied by industry and by how researchers defined the strength of a relationship. The study produced important evidence about professional networks, but it also revealed something unsettling: an adjustment to a recommendation algorithm could influence whom millions of people met and, ultimately, where some of them worked. A seemingly helpful list of suggested contacts had doubled as a five-year employment experiment.^[7]

3 Facebook Tested Whether Friends Could Make People Vote

On the day of the 2010 United States congressional elections, Facebook conducted an experiment involving approximately 61 million users. One group received an informational message encouraging them to vote. Another saw the same message accompanied by photographs of Facebook friends who had clicked an “I Voted” button. A control group received no special voting message.

Researchers then measured whether users sought information about polling places, publicly reported voting, and appeared in official voting records. The social message featuring friends produced stronger effects than the purely informational version. The influence also spread through close relationships, suggesting that political participation could move across an online social network.

The researchers estimated that the Facebook message directly generated tens of thousands of additional votes, with social transmission potentially producing hundreds of thousands more. The experiment did not tell people which candidates to support, but it showed that a platform could slightly alter real-world electoral participation simply by displaying the behavior of selected friends. Facebook was no longer just carrying political messages; it was testing the mechanics of political mobilization on a national scale.^[8]

2 Facebook Secretly Adjusted the Emotional Tone of News Feeds

For one week in January 2012, Facebook altered the News Feeds of 689,003 users. Some participants were shown fewer posts containing positive emotional language, while others saw fewer posts containing negative language. Facebook and academic researchers then examined the words those users wrote in their own subsequent posts.

Users exposed to less positive material produced slightly fewer positive words and slightly more negative ones. The opposite pattern appeared among people shown less negative material. The measurable effects were extremely small, but researchers argued that the results demonstrated emotional contagion: emotions could spread through a social network even without face-to-face interaction or direct communication.

The study caused an uproar when it became public in 2014. Participants had not been specifically informed that their feeds were being manipulated for a psychological experiment, and critics questioned whether accepting Facebook’s general data policy amounted to meaningful consent. The company had changed the emotional content reaching hundreds of thousands of people to determine whether their own emotions—and the language they shared with others—would follow.^[9]

1 Imperial Tobacco Observed Teenagers Behind Closed Doors

In 1977, Imperial Tobacco commissioned a research project examining the smoking attitudes and behavior of 16- and 17-year-olds. Known as “Project 16,” the study placed teenagers in discussion groups held in hotel rooms. Researchers questioned them about cigarettes, smoking, and the social meaning attached to the habit.

The sessions were not ordinary interviews. Company representatives could observe the discussions through closed-circuit television, allowing them to study the teenagers from outside the room. The resulting report examined the language young people used, the pressures surrounding smoking, and the role cigarettes could play in identity and acceptance among peers.

The tobacco industry later insisted that it did not market cigarettes to minors. Yet Project 16 demonstrates that at least one major manufacturer deliberately gathered detailed information about underage consumers and their psychological relationship with smoking. The teenagers were not being studied to help them quit. Their private discussions offered the company insight into how young people understood a product capable of producing lifelong dependence.^[10]