What's actually in your file

Most writing about data brokers stays abstract — 'they collect your personal information' — which is true and tells you nothing. It is more useful to look at the shape of an actual listing, because the specifics are what make the problem legible.

A typical people-search result carries your full name and any middle initial, your current age, your current address, a list of previous addresses going back years, phone numbers associated with those addresses, and a list of relatives and known associates. Some add employers, property records, or a link to a paid background report promising more.

Composite example, invented to show the shape — not any one person's record: Dana R. K——, 47. Current: Bakersfield, CA. Previous: Reno NV · Boise ID · Spokane WA · Tucson AZ · Little Rock AR · Erie PA. Relatives: three names. Two phone numbers.

Read that again as someone who wants to find you. It is not just an address. It is a movement history — which state you left, roughly when, and who you are connected to. The address history is often the most sensitive part of the record, and it is the part people are least likely to know exists.

Where each piece comes from

Almost none of it is stolen, and that is the uncomfortable part. The raw material is largely public or commercially available:

• Property and deed records — filed publicly whenever you buy, sell, or refinance.
• Voter registration — public in many states, and it carries name, address, and often date of birth.
• Court and civil filings — public by design, including matters that were resolved in your favour.
• Utility hookups, change-of-address filings, and warranty registrations — routinely resold.
• Marketing datasets bought from companies you did transact with.

A broker's actual product is not any single record. It is the join: the work of deciding that the person on a 2016 deed in one state and the person on a phone contract in another are the same person, and stapling them into one profile. That inference is the value they add, and it is why the file is more revealing than any of its sources.

The errors are informative

Listings are frequently wrong in a specific way: they over-include. Relatives who are not relatives. Addresses you never lived at, because someone with a similar name did. Ages off by a year or two. Phone numbers that were reassigned to you and belonged to someone else first.

This tells you something about how the join is made — probabilistically, at scale, with a bias toward including a maybe rather than dropping it. A profile that is 85% right is still perfectly sufficient for someone trying to find your front door, and the 15% that is wrong is not a defence. It is just noise attached to your name that you did not put there and cannot correct.

Why it comes back

This is the part that surprises people who have opted out before. You submit a removal, the listing disappears, and some months later it is back.

Nothing dishonest necessarily happened. The sources are continuous: county records keep being filed, voter rolls keep being updated, datasets keep being resold. A broker that deleted your record and then rebuilt it from a fresh public-records pull has, from its own point of view, simply found a new person who happens to be you. Some honour a suppression list that prevents this. Many do not, or only for a while.

Which is why a single opt-out is a snapshot, not a state. Anything that treats removal as a one-time task is describing a different problem from the one that actually exists.

What is reasonable to expect

Opting out works, in the narrow sense that listings do come down and stay down for a period. It is not the same as your data ceasing to exist, no service can promise that, and the honest version of this is unglamorous: it is maintenance, not a cure. The right question to ask anyone offering to do it for you is not how many brokers they claim, but whether they can show you what was sent, what came back, and when they will do it again.