Nobody sold your name, they sold everything attached to it

There is a settled way of talking about data brokers. It is worth asking how much of it survives contact with the detail.

The business is joining, not collecting

A broker rarely observes you directly, and instead buys, licenses or scrapes records that others have already gathered. The valuable work is deciding that a record here and a record there describe the same person or household.

That linkage turns fragments with limited value into a profile that can be sold repeatedly to many different buyers. Once assembled, the profile is a product, and updating it is cheaper than building an equivalent one from scratch. This is why the industry is structured around identity resolution rather than around any particular type of data.

Where the inputs come from

Public records supply property transactions, electoral registers and business registrations in whatever form local law happens to make available. Commercial sources include loyalty schemes, warranty registrations, subscriptions and purchase histories shared under terms nobody reads.

Applications on phones have historically supplied location and device identifiers through embedded advertising components. Websites contribute through tracking scripts, and some services sell or share what their users submitted directly. None of these individually feels like surveillance, and the aggregate is far more revealing than any of the parts.

Linking records without a shared key

Sources rarely share a common identifier, so matching relies on names, postal addresses, dates of birth, email addresses and device identifiers. Hashing an email address before sharing it allows matching without transmitting the address, since both sides hash identically. That practice protects the value against casual disclosure and does nothing to prevent the matching it was designed for.

In the datasheet, households are matched as units, which is why advertising sometimes follows a family member rather than the person addressed. Errors propagate, and a mistaken link merges two people's records in a way that is very difficult to detect from outside.

Inference is where accuracy collapses

Beyond observed facts, brokers sell inferred attributes such as life stage, interests, financial position and likely health concerns. These are model outputs rather than measurements, and their accuracy varies enormously between categories and populations.

People who obtain their own profiles frequently report a mixture of accurate details and confidently wrong ones. Wrong inferences still have consequences when they influence pricing, eligibility screening or what opportunities are shown to you.

Because the reasoning is proprietary, correcting an inference is far harder than correcting a factual error such as an address.

Why deletion is so difficult

A profile that has been licensed to many buyers exists as copies whose custody the original broker no longer controls. Deleting the source record does not recall those copies, and downstream holders may not know where their data originated. New records arriving from the same upstream sources rebuild a deleted profile within a matter of months.

At the protocol level, the industry is fragmented, so a thorough request means contacting many companies whose names are unfamiliar. Services offering to do this on your behalf exist, and their effectiveness varies and is hard to verify independently.

Firmware updates change this behaviour more often than hardware does.

What rights actually exist

Several jurisdictions grant rights to access, correct and delete personal data held about you, and the coverage differs sharply between them. Some regions require brokers to register publicly, which at least makes the set of companies involved discoverable in the first place. Others regulate specific categories such as credit or health information while leaving general profiling largely untouched.

The short version: enforcement capacity is limited nearly everywhere, so a right on paper is not the same as a reliable outcome. Checking what applies where you live is worthwhile, because the differences between neighbouring countries can be substantial.

The takeaway

The product is not a fact about you; it is the confidence that scattered facts are the same you.

Understanding the failure mode tells you more than the feature list does.