Here’s what data brokers know about you

A few years ago I started noticing the ads. Not the obvious kind, the uncannily specific kind. A product I had talked about but never searched for. A local service in a town I had only just moved to. It felt less like being marketed to and more like being watched.

It turns out the watching is real, and it has a name and a business model. Behind most of it sits a quiet industry almost nobody interacts with directly: data brokers. Companies you have never heard of hold detailed files on who you are, where you go and what you are likely to do next. This is a plain look at what they are, what they actually know, how some of them misuse it, and what you can realistically do about it.

What is a data broker

A data broker is a company whose business is collecting personal information about people, packaging it, and selling it to others. The defining feature is that you have no direct relationship with them. You never opened an account, never agreed to their terms, and in most cases will never learn their name.

That is what separates a data broker from a company you actually use. When you sign up for a bank or a social network, there is at least a relationship there, however lopsided. A data broker sits entirely outside that. It buys and assembles information about you from everywhere else, builds a profile, and rents that profile out to whoever will pay: advertisers, insurers, landlords, employers, debt collectors, political campaigns and government agencies.

The scale is not small. California now runs a public registry of data brokers operating in the state, and over 600 companies are listed on it. That is only the ones that register in one state. The real number operating globally is far larger.

How they get your data

Data brokers rarely collect information from you directly. They assemble it from a web of other sources, most of which you touched without thinking about it.

The common pipelines are public records (property filings, voter rolls, court records, licenses), commercial sources (loyalty programs, warranty cards, subscription lists, purchase histories) and the vast online advertising machine. That last one matters more than people realize. Every time an app or website loads an ad, your device can broadcast information to dozens of companies bidding to show it, in a process called real-time bidding. The US Federal Trade Commission found that one broker, Mobilewalla, collected data from those ad auctions and kept it even when it did not win the bid, harvesting location and device information off the back of ads that were never even shown.

Brokers also buy from each other. A profile gets passed, merged and enriched across multiple companies until a single file can contain your name, addresses going back years, family members, income band, shopping habits, health interests and a near-continuous history of where your phone has been.

What they actually know

The raw data is only half of it. The more valuable part is the inferences brokers build on top, the guesses about who you are that get sold as fact.

From location history alone, a broker can infer a startling amount. The FTC's case against the broker Gravy Analytics described files that could reveal visits to places of religious worship, domestic abuse shelters, homeless shelters and medical facilities, along with inferences about medical conditions and religious beliefs. Nobody handed that over directly. It was reconstructed from the trail a phone leaves behind.

Stacked together, the picture a broker holds is often more complete than what your closest friends know: not just where you live and work, but your routines, your worries, your financial pressure points and the categories a marketer or an insurer would quietly sort you into.

How some of it gets abused

Most data brokering is legal, and a lot of it is mundane ad targeting. The problem is that the same profiles enable real harm, and the controls on who buys them have been weak.

Sensitive location data is the sharpest example. Data showing who visited an abortion clinic, an addiction treatment center or a place of worship can be used to target, profile or expose people, which is why the FTC moved to ban several brokers from selling it without genuine consent. Beyond that, the same files feed scams (fraudsters buy lists of likely-vulnerable targets), stalking and harassment (an address history is a gift to someone trying to find a person who does not want to be found), and quiet discrimination in pricing, lending or hiring based on inferences a person never sees and cannot contest.

There is also the breach risk. A broker is a single company holding dossiers on millions of people, which makes it a high-value target. When one is breached, the most sensitive possible dataset, a map of where people go and what they do, spills out at once.

Signals a company handles your data responsibly

Not every company that holds your data is a broker, and not every company is careless with it. When deciding whether to trust a service with your information, these are the signals worth looking for.

• It collects only what it needs. Responsible companies practice data minimization, asking for the minimum required to do the job rather than hoovering up everything they can.
• It is clear about retention. Good practice is to say how long data is kept and to delete it when the purpose is done, not to hold it forever by default.
• It does not sell to third parties. A clear, plain statement that your data is not sold or shared with brokers is one of the strongest signals there is.
• It honors your rights quickly. A real opt-out, a working delete function and a response when you exercise them, rather than a form that goes nowhere.
• It encrypts and limits internal access. Data is protected in transit and at rest, and not everyone inside the company can see it.
• It proves rather than stores. The most forward-looking sign: a company that can confirm a fact about you without keeping a permanent copy of the underlying data.

A company that does the opposite, asking for far more than it needs, keeping it indefinitely and staying vague about who it shares with, is telling you something.

How to stay safe

You cannot fully disappear from the data trade, but you can meaningfully shrink your exposure.

• Opt out at the source. California residents can now use a state-run portal called DROP to send deletion and opt-out requests to every registered broker at once. Other states and services offer their own opt-out routes, and several tools will file them on your behalf.
• Use your deletion rights. Many regions now give you a legal right to request that a company delete the data it holds on you. It takes effort, but it works.
• Starve the ad pipeline. Limit app location permissions to "while using," turn off your advertising ID, and be stingy with loyalty programs and warranty forms that exist mainly to collect data.
• Reduce what is public. Lock down social profiles and remove personal details from places that expose them.

None of these are one-time fixes. They are part of a wider routine for protecting your privacy online, which is less about any single setting and more about consistently giving away less than the default asks for.

The new approaches that actually change the model

Opting out one broker at a time is useful, but it is a losing race against an industry built to collect faster than you can delete. The more durable fixes change the model itself, so that sensitive data is never sitting in a company's database waiting to be sold or breached in the first place.

Two ideas are doing most of the work here.

The first is proving a fact without revealing the data behind it. A method called zero-knowledge proofs lets a system confirm something is true, that you are over 18, say, or a unique real person, without the underlying details ever changing hands. A service gets the yes-or-no answer it needs and learns nothing it could store, leak or sell. The data that a broker would have loved to buy simply never exists in a sellable form.

The second is changing who holds the data at all. The principle of personal custody flips the usual arrangement: instead of handing a copy of your information to every company and trusting each one to guard it, you keep your own data on your own device and hand over only a proof when something genuinely needs checking. There is no central warehouse of your life for a broker to buy from, because the data stays with you.

Neither idea erases the existing data trade overnight, and both depend on companies choosing to build this way. But they point at the real fix. The problem was never only that brokers misbehave. It is that the whole system was built to copy and hoard personal data by default. Designs that prove instead of store, and that leave data in your hands instead of theirs, take away the raw material the trade runs on.

The short version

A data broker is a company that collects your personal information and sells it, with no direct relationship to you and usually no way for you to know it is happening. They assemble profiles from public records, commercial data and the online ad system, then sell both the raw data and the inferences built on top, including sensitive details reconstructed from where your phone has been. Some of this enables real harm, from scams to stalking to quiet discrimination, which is why regulators have started banning the sale of sensitive location data.

You can push back by opting out through registries, using your deletion rights and starving the ad pipeline. The deeper fix is structural: systems that prove a fact without revealing the data, and that keep your information in your own custody, remove the raw material the whole trade depends on.

A company you have never heard of should not know more about you than your friends do. For a long time that was simply how the internet worked. It is starting, slowly, not to be.

How to be anonymous online Everyone wants it, almost nobody achieves it

What is a digital footprint, and how big is yours? By the end of this you'll never look at "just checking my phone" the same way