Removed is not deleted: the four layers of personal-information removal

“I removed it” is four different sentences. They are usually said as one.

A Google result can disappear while the page is still live. A people-search listing can be suppressed while the court record it copied is untouched. A California DROP status can read Deleted while a first-party account, a relative's report, and a newspaper story still name you. A public record can be sealed in one office and reprinted, indefinitely, by everyone who already captured it.

The FTC's article on people-search sites already says the quiet part: after you opt out, the site stops selling the matched information it has, but public records can put you back on sale, and your details can remain in a relative's report. Google's Results about you is equally plain in the other direction: it can hide personal contact information from Search when the request meets policy, and it will not remove government, educational, or newspaper pages it considers valuable to the public.

The operational ledger — what to record, when to recheck, how to resubmit — is Why removed personal information reappears. This post is the model underneath that ledger. If a request “failed,” the first question is which layer you actually asked.

The four layers

| Layer | What a successful request changes | What it does not change | The official door | | --- | --- | --- | --- | | 1. Search index | Whether one engine shows a URL, snippet, or image for a query | The live page, other engines, direct links, archives | Results about you, outdated-content refresh, Bing's equivalent | | 2. People-search product | Whether that company shows or sells a matched profile in a defined product | Other brands, other search types, relatives' reports, the sources | That company's current opt-out or suppression form | | 3. Registered-broker file (California) | Covered personal information at brokers who must register, once they match you | First-party data, publicly available data, other exemptions, non-brokers | DROP | | 4. Public record | The government or official record, if a correction, confidentiality, or sealing route exists | Almost everything downstream that already copied it | The record-holder, not a people-search form |

1. 01

   Search index

   One engine hides a URL

   Leaves: The live page, other engines

2. 02

   People-search product

   That brand stops showing a match

   Leaves: Other brands, relatives, sources

3. 03

   Registered brokers

   DROP deletes covered matched data

   Leaves: First-party, public, exempt data

4. 04

   Public record

   Only if the record-holder allows it

   Leaves: Almost everything downstream

A fifth action, account deletion, sits beside these layers rather than inside them. Closing your own social or retailer account changes data you control. It does not delete a people-search profile built from public records, and it does not take a URL out of Google.

The FTC also distinguishes a sale or sharing opt-out from a deletion. Stopping a sale is not the same as erasing a row. California's CCPA treats those as different rights. The rights guide keeps them separate. Mixing them is how a “completed” email becomes a disappointment.

Layer 1: the index is not the source

Google's Results about you lets an adult, in supported markets, enter a name and contact details — including nicknames, maiden names, multiple phones, addresses, emails, and certain government ID numbers — and see Search results that display that information. You can ask to remove a result, or mark it reviewed. You can turn on notifications when new matches appear.

The policy limits are the intelligence, not the button.

Google will not offer removal when it considers the page valuable to the public. The help article names educational institutions, government institutions, and newspapers. A result that shows your home address on a county site, or your name in a local article, can match the monitor and still have no “Remove result” action.

Google also will not treat information you control as its problem. A social profile or personal blog you can edit yourself is your edit, not a Search removal.

Even an approved request is an index change. The help article says there can be a delay between approval and disappearance, and that you should check within a few hours after approval. The page can remain up. Bing can still show it. An archive can still hold it. Someone with the URL does not need Google.

There is a second, narrower index tool: asking an engine to refresh a snippet when the source has already changed or vanished. That is the outdated-content path in How to remove personal information from search results. It is the right tool when the broker listing is gone and the search card is stale. It is the wrong tool when the live page still publishes the address.

If you only use layer 1, you will sometimes feel successful and still be fully public. That is not Google failing its policy. That is the policy working as written.

Layer 2: a product opt-out is a product change

The FTC's people-search article is the cleanest official description of this layer. People-search sites collect public records, public social profiles, and commercial files, compile a report, and sell it. Most have a way to tell them to stop selling your information. They may let you suppress an address or phone number and still publish something else, such as a sex-offender status. After you opt out, changed public records can put information back on sale. Your details can remain in reports about relatives, neighbors, or associates.

That is a complete theory of recurrence in four sentences. It is also why “I opted out of Spokeo” is not a sentence about the internet.

Read the confirmation for scope. PeopleConnect's suppression center, which we cite in the opt-out directory, describes coverage across specified brands for name-search background reports and states limits around other search types and other people's reports. Spokeo's opt-out page describes removing the Spokeo listing, not the original sources or other sites. Those are not fine-print tricks. They are the layer speaking.

A useful test after any product opt-out:

1. Search the same brand by the same identifier you used the first time.
2. Search it by the other identifier that originally found you — phone if you used name, name if you used phone.
3. Open any relative or associate report the FTC warning would predict.
4. Open the source the listing cited, if it cited one.

If step 1 is clean and step 2 is not, the request covered one search mode. If steps 1 and 2 are clean and step 3 is not, you removed your card and remained a field on someone else's. If step 4 is still live, you are now in layer 4, or in a first-party page, and another opt-out at layer 2 will keep failing for a reason that looks like incompetence and is actually jurisdiction.

Do not submit a request pretending to be the relative. Ask the broker what route applies to an appearance in another person's report.

Layer 3: DROP is a statutory request to registered brokers

California's Delete Request and Opt-out Platform is the only official mechanism that sends one consumer request to the active brokers on the state registry. On 2 June 2026 CalPrivacy said that registry held a record 581 brokers and that more than 300,000 residents had signed up. We wrote the registry reading as its own post because those numbers are easy to turn into a myth.

DROP's own status vocabulary is the model in miniature.

| Status | Layer-accurate reading | | --- | --- | | Deleted | Matched covered data at that broker was deleted and will no longer be sold; exempt data may remain | | Exempted | The broker says everything it kept is exempt — including, in the official examples, public records or certain health information | | Opted-out | No exact match; sale associated with the identifiers you provided must stop, as DROP describes that status | | Record not found | No responsive record from the data you submitted — not a historical guarantee | | Pending | Not yet processed; brokers may take up to 90 days to report, and they process at least every 45 days |

“Deleted” at layer 3 is compatible with a live Google result (layer 1), a listing at a brand that is not the registered entity you are thinking of (layer 2), and an unchanged assessor file (layer 4). It is also compatible with a people-search page that still shows publicly available data the statute lets a broker keep.

Processing began on 1 August 2026. A request submitted in January could still be pending in mid-August without anyone breaking the published rules. If you evaluate DROP as if it were a same-day delete button, you will write a failure story about a clock you did not read.

DROP also does not reach first-party businesses. The retailer that has your account, the platform that has your profile, and the broker that bought a file about you are three companies. One DROP submission is not three privacy requests.

Layer 4: the public record is usually not deletable

The FTC lists the record types people-search sites compile: property; driving; voter registration; criminal, civil, and judgment files; birth, marriage, divorce, and death; professional licences. Each of those is held by an office that is not a data broker. Each has its own correction, amendment, or confidentiality process, if it has one at all.

A people-search opt-out cannot unfile a deed. DROP's official guidance says publicly available data will not necessarily be deleted. Google's Results about you will often refuse a government URL. The three consumer tools agree, independently, that layer 4 is someone else's.

That does not mean nothing can be done. It means the action has to match the record:

• a wrong address on a property record is a correction at the assessor or recorder, if the office allows it
• a safety-driven address is often an address-confidentiality or victim-protection programme, not a deletion
• a court record may have a sealing or redaction motion, which is a legal proceeding
• a professional-licence page is the licensing board's process

Do not assume deletion is available. Do not publish the record in a social thread while you wait; that creates a new layer-1 object. If you are in danger, a survivor advocate or appropriate professional should design the sequence. Generic self-searching can create its own trail.

A decision tree for the next request

Start from the thing you can see, not from the tool you heard about.

1. Is the live URL still up? If yes, you are not done at layer 1. Identify the publisher.
2. Is the publisher a people-search brand? Use that brand's official opt-out, and use DROP as well if you are an eligible California resident. Record the scope.
3. Is the publisher a first-party business you have an account with? Use that business's privacy channel. In California, that is a CCPA request, not DROP.
4. Is the publisher a government office, court, or licensing board? Use the record-holder route. Do not expect Google, DROP, or Spokeo to unfile it.
5. Is the publisher a newspaper, school, or other page Google treats as public-interest? You may get a monitor hit and no removal. Ask the publisher. Do not file the same Search request until the policy changes.
6. Is the live URL gone, but Search still shows it? Now use the outdated-content or Results-about-you path for that exact URL.
7. Did a listing return later? Compare URL, record ID, visible fields, and search path with the confirmation you saved. The reappearance guide is the worksheet. The cause is usually a new upstream row, a second identity cluster, a relative's report, a different search mode, a stale snippet, or an exemption — not a single company “putting it back” for sport.

If you cannot name the layer, do not send more identity data. More data helps a match at layer 3 and can create a new file everywhere else.

What we will not claim

We will not claim that stacking every tool produces deletion from the internet. The FTC, CalPrivacy, and Google have each already listed what they will not do. An honest completion standard is the one in the reappearance guide: the high-risk listing is gone on the paths you tested, snippets have refreshed, exceptions are written down, upstream sources were addressed where a valid path exists, and you can recognize the next copy quickly.

We will not publish measured broker-by-broker removal times. That work needs a protocol, consent, and a sample. It has not started.

We will not tell you to pay an unsolicited “we found your data” removal service. The FTC's article says you can opt out yourself or pay a service, and that you should ask what the service covers and how it handles reappearance. Unsolicited outbound sales are a different object.

If the listing is on DeepSearch

Remove my info is layer 2 for our own pages. It does not delist a source we linked to, a Google snippet, or a court record. How to correct or remove information from DeepSearch covers mixed identities as well as takedowns.

If we are treating a public record as if our removal form could unfile it, or treating a Search approval as if it deleted a broker row, that is a mistake in this model. The contact page is the right place to say so.

The word “removed” will keep being used for all four layers. The work is to hear which one was meant before you decide the system failed.