An ad follows someone across three different apps after one search on a totally separate website. Nobody at that website sold the search directly to the advertiser. Somewhere in between, a company most people have never heard of bought, packaged, and resold that behavior. That company is a data broker.

What does “data broker” mean?

A data broker is a company that collects personal information from many sources. It then sells or licenses that information to other companies. The information can include browsing habits, purchase history, location data, and income estimates. Most people never interact with a data broker directly. It gathers information about them anyway, often through deals with the sites and apps they actually use. The term applies once collecting and reselling data becomes the business itself, not a side effect of some other product.

Where did the term come from?

Data brokering predates the internet by decades. Mailing-list companies once compiled names, addresses, and purchase histories to sell to catalog retailers. This happened long before anyone used the word “data” this way. The digital version exploded once online tracking made behavioral data cheap to collect at scale. By the 2010s, journalists and regulators had settled on “data broker.” It became the standard term for any company built around aggregating and reselling personal information.

How does a data broker actually work?

Brokers gather information from a wide range of sources. Public records, loyalty programs, app permissions, and partnerships with websites that share visitor data for payment all feed into it. No single source reveals much on its own. Combined across dozens of sources, though, the profile gets remarkably detailed. It often includes inferred traits the person never directly disclosed.

That inference step matters most. A broker doesn’t just report what someone bought. It can estimate income bracket, likely health conditions, or political leanings too. These estimates come from patterns matched against millions of other profiles. Since none of it comes directly from the person, the estimate can be wrong. It still gets sold as if it were reliable.

Most people never consent to this in any meaningful sense. A checkbox buried in some app’s terms of service technically covers it, but almost nobody reads that far. That gap between technical consent and actual awareness is the industry’s most common complaint.

A concrete example

Someone fills out a warranty card for a kitchen appliance. It lists their name, address, and a few lifestyle details. Weeks later, unrelated catalogs start arriving that match those same lifestyle details closely. The original company never advertised anything to the person directly. Instead, it sold the warranty card data to a broker. That broker resold it again to marketers with no connection to the original purchase at all.

What it’s not

A data broker isn’t the same as a single company using its own customer data for its own marketing. That’s a direct relationship, not a resale. It’s also not identical to a data breach, where information leaks without anyone’s authorization. Brokers operate legally, through purchased or licensed data, even though the practice draws heavy criticism. And it isn’t the same as a credit bureau specifically. Credit bureaus function as one type of specialized broker, but the category is much wider than that.

Where you’ll encounter it

The term shows up constantly in privacy legislation debates and opt-out service advertisements. It also appears in news coverage whenever a specific broker gets caught selling sensitive location or health data. And it comes up whenever someone wonders how a random company seemed to know something about them that nobody ever told it directly.

Related terms

• Cancel Culture (L23)
• Shadow Banning (L24)
• Algorithmic Bias (L25)