The Unsettling World of Data Brokerage

The owner of a data brokerage business has released a disturbing video showcasing the extent to which his industry can collect and analyze data on the habits of billions of people.

Recent Acquisitions and Integrations

In 2019, the data broker Epsilon was acquired by French advertising conglomerate Publicis Groupe. Earlier this month, Publicis also acquired Lotame, another data and advertising firm, and announced plans to integrate it with Epsilon’s business. At the time, Publicis CEO Arthur Sadoun stated that the new corporate integration would enable his company to deliver "personalized messaging at scale" to approximately 91% of the internet’s adult web users.

The scope of Data Collection

To deliver such "personalized messaging" (i.e., advertising), Publicis must gather an extraordinary amount of information on the people it serves ads to. In a newly released video, Sadoun breaks down the kind of information his company claims to have access to. The video, which concerns a software program called CoreAI, demonstrates the degree to which the company can analyze and predict the habits and behavior of individual consumers.

A Hypothetical Example: "Lola"

To illustrate this, Sadoun introduces the audience to "Lola," a hypothetical young woman who represents the typical web user that Publicis now has data about. "At a base level, we know who she is, what she watches, what she reads, and who she lives with," Sadoun explains. "Through the power of connected identity, we also know who she follows on social media, what she buys online and offline, where she buys, when she buys, and more importantly, why she buys."

The Depth of Data Analysis

It gets worse. "We know that Lola has two children and that her kids drink lots of premium fruit juice. We can see that the price of the SKU she buys has been steadily rising on her local retailer’s shelf. We can also see that Lola’s income has not been keeping pace with inflation. With CoreAI, we can predict that Lola has a high propensity to trade down to private label," Sadoun says, meaning that the algorithm determines whether Lola is likely to start buying a cheaper brand of juice. If the software decides this is the case, the CoreAI algorithm can automatically start showing Lola ads for those reduced-price juice brands, Sadoun claims.

The Broader Implications

Poor Lola. The significant problem is that it’s not just Lola who is subjected to this all-encompassing corporate gaze—it is, apparently, almost everybody on the internet. "Thanks to CoreAI, we can do that with 91% of adults all around the world," the CEO boasts. That amounts to nearly four billion people.

Expert Insights

Lena Cohen, a technologist with the Electronic Frontier Foundation, notes that data brokers like Publicis collect "as much information as they can" about web users. "The data broker industry is under-regulated, opaque, and dangerous, because as you saw in the video, brokers have detailed information on billions of people, but we know relatively little about them," Cohen says. "You don’t know what information a data broker has on you, who they’re selling it to, and what the people who buy your data are doing with it. There’s a real power/knowledge asymmetry."

The Need for Regulation

Meaningful regulation is notably absent from the data broker industry, Cohen emphasizes, allowing companies to operate without much transparency or oversight. "The U.S. needs comprehensive federal privacy law," Cohen asserts, while noting: "I don’t know how likely that is right now." Even when state-level privacy regulations are passed (such as the California Consumer Privacy Law), those cases are often not given enough focus or resources for the laws to be enforced effectively.

A Call for Change

Cohen adds that she feels online behavioral advertising—that is, advertising based on an individual web user’s specific browsing activity—should be illegal. "That’s not data that advertisers should have," she says, noting that contextual advertising (which is based on the content in a particular webpage) should be effective enough for companies. Banning behavioral ads would "fundamentally change the financial incentive for online actors to constantly surveil" web users and share their data with brokers, Cohen argues.