Toolkit This is what the Digital Ghost Toolkit fixes. See it.
Case study 001 · 22 August 2026
Digital Ghost Protocol

Case study 001 · Personal data removal

Automated removal vs. a guided run

One person. One day. A two-year paid subscription to an automated data removal service, measured against a single guided removal run done by hand. Every corporate claim below is checked against the California data broker registry, which brokers are required by law to file. The finding is not that the software failed to try. It is that it is watching the wrong layer of the industry.

Observed
22 Aug 2026
Account age
26 months
Brokers on board
250
Guided run
One day
01
The short version

What we found

A common worry sounds like this: my address, my phone number and the names of my relatives are sitting on websites I never signed up for, and I would like them gone. A subscription service promises to handle it. This document looks at what one of those subscriptions, a well-known automated removal service, had actually accomplished after two years of paid service, and compares it to a single guided run performed by hand on the same person on the same day.

We are not accusing anyone of lying. The service does real work, and some of it is genuinely useful. What the record shows is a structural blind spot: the board it monitors is built out of consumer-facing websites, while the companies that supply those websites with data are absent from it entirely.

207
Brokers reported
“no exposures found”
30
In progress
after 26 months
13
Marked
exposures removed
0
Wholesale suppliers
on the board

Counted directly from the account dashboard, 22 August 2026. 207 + 30 + 13 = the 250 brokers the service tracks.

  • Four sites cleared, one uncleared parent. Four people-search sites the service reported clean file with the state from the same address as a wholesale supplier the service itself still lists as unresolved. A live listing was found by hand on three of them the same day.
  • One company counted as five brokers. Five separate entries on the board, all reported clean, are registered domains of a single company. A live listing was found on that company the same day.
  • The suppliers are missing. The large data wholesalers, and eighteen registered entities belonging to the three national credit bureaus, appear nowhere on the board. The guided run reached four wholesalers by hand in an afternoon.
02
Method

How this was checked

Three independent sources, none of them opinion.

Sources used, and what each one is good for
SourceWhat it isWhat it proves here
The service's dashboard A live paid account, opened 26 June 2024, read on 22 August 2026. What the service says it is monitoring, working on, and has finished.
California data broker registry Filings that data brokers are required by California law to submit, naming the legal entity, its filing address, and every website it operates. Registration years 2025 and 2026. Who actually owns which website. This is the part a marketing page cannot dispute.
Scopos run tracker The guided removal run, 22 August 2026, with the status of every submission recorded as it happened. What a person working by hand reached in one day, and what is still pending.
03
Headline finding

One address, four filings

In the 2025 registration year, four companies filed with California from the same street address in Sacramento. One of them is Enformion, a wholesale supplier of people data. The other three operate the consumer websites FastPeopleSearch, FastBackgroundCheck, TruePeopleSearch and FamilyTreeNow.

On the dashboard, Enformion sits in the active queue, unresolved, labelled “Ensuring Broker Compliance.” All four of the consumer sites at that same address are reported as no exposures found. The guided run located a live listing on three of those four sites that same day.

Registry filings, 2025 registration year · one filing address in Sacramento
Enformion, LLC
Wholesale people-data supplier
Service status: unresolved
FastPeopleSearch
Mississippi Tornado Alley, LLC
Reported clean Listing found
FastBackgroundCheck
Mississippi Tornado Alley, LLC
Reported clean Listing found
TruePeopleSearch
Free Data Services, LLC
Reported clean Listing found
FamilyTreeNow
Family Tree Now, LLC
Reported clean Not searched
2026 filings: Enformion and Mississippi Tornado Alley still file from that same Sacramento address. The other two have moved to post office boxes in two different cities, which carry the identical private mailbox number as each other.
Why this matters to a normal person

A clean result on a website that shares an address with a supplier you have not resolved is a clean result about the shop window, not the warehouse.

Stated carefully, because the distinction is the whole point: co-location and shared corporate structure show that the clearance is cosmetic in scope. They do not prove that the supplier feeds those specific sites, and they do not prove the service made a false claim. What they establish is that clearing the front ends while the connected supplier remains unresolved cannot be described as finishing the job.

04
Finding two

Five brokers, one company

The registry lists Spokeo, Inc. as the operator of spokeo.com, thatsthem.com, anywho.com, peoplewin.com, freepeopledirectory.com and one further domain. Five of those appear on the dashboard as five separate brokers. Every one of them is reported as no exposures found.

The guided run found a live listing on Spokeo itself that same day and confirmed its removal. The same legal entity was therefore reported clean five times over while one of its own front ends was carrying a live record.

Spokeo, Inc. as it appears on the board, against the registry
Entry on the boardRegistered operatorService statusGuided run, same day
SpokeoSpokeo, Inc.No exposures foundListing found, removal confirmed
ThatsThemSpokeo, Inc.No exposures foundOpt-out submitted, pending
AnyWhoSpokeo, Inc.No exposures foundCovered by the Spokeo action
PeopleWinSpokeo, Inc.No exposures foundCovered by the Spokeo action
FreePeopleDirectorySpokeo, Inc.No exposures foundCovered by the Spokeo action

Table scrolls sideways on a narrow screen.

05
Finding three

A big number, counted twice

“250 brokers monitored” sounds like 250 companies. In several families on this board it is not. Brand names are counted as separate brokers when the registry shows them filing as one company, and in the largest family the board covers only a fraction of the brands that company actually runs.

Where the count and the corporate record disagree
FamilyWhat the registry showsHow the board treats it
Enformion Enformion, LLC files Tracers.com as one of its own trading names. Enformion and Tracers appear as two separate entries in the active queue, both unresolved.
Mississippi Tornado Alley One company, ten registered websites, including several of the best known free people-search sites. Three of the ten appear on the board. The other seven are not tracked at all. The guided run submitted opt-outs by hand on two of the untracked seven.
AccuData and Deep Sync Compact Information Systems, LLC filed AccuData Integrated Marketing as its trading name in 2025 and Deep Sync in 2026, from one office suite. Counted as two separate cleared brokers, three records each.
BeenVerified One filing covers BeenVerified together with its sister sites. BeenVerified, PeopleLooker, NeighborWho and Ownerly sit as four separate entries, all unresolved. One opt-out in the guided run covered all four, confirmed the same day.
The plain reading

None of this is hidden. It is on the public record, filed by the brokers themselves. But it means the headline number describes brands watched, not companies handled, and in these families the difference runs to roughly ten to one.

06
Finding four

The layer nobody touched

People-search websites are the retail counter of this industry. Behind them sit the wholesalers: companies that assemble records at scale and license them onward. They are registered California data brokers, they have public opt-out channels, and they are the reason a record that disappears from one website can reappear on another.

Not one of them appears anywhere on the 250-broker board. Neither do the eighteen separately registered entities filed by the three national credit bureaus in the 2026 registration year. The guided run reached four wholesalers by hand in a single afternoon.

Wholesale layer, 22 August 2026
CompanyRegistered CA data brokerOn the 250-broker boardGuided run, same day
LexisNexisYes, both yearsAbsentSuppression submitted, pending
AcxiomYes, both yearsAbsentSuppression submitted, pending
EpsilonYes, both yearsAbsentDeletion request submitted, pending
LiveRampYes, both yearsAbsentNot reached in this run
Cotality (formerly CoreLogic)Not in the filings reviewedAbsentDeletion request emailed, pending
Equifax entities4 registrants, 2026AbsentCovered by the prescreen opt-out
Experian entities3 registrants, 2026AbsentCovered by the prescreen opt-out
TransUnion entities11 registrants, 2026AbsentCovered by the prescreen opt-out
One cleared broker, read closely

Postie appears on the board as a completed removal. Postie publishes the names of the data providers it draws from, and that published list includes Epsilon and Acxiom. Both were still live for this person on the day the removal was marked complete, because neither is on the board to begin with. Clearing the customer while leaving its named suppliers untouched is the pattern in miniature.

The structural argument

A subscription that permanently severed the wholesale layer would end its own recurring problem, and with it the reason to keep subscribing.

This is an observation about incentives, not an accusation of intent. What is documented is the outcome: after 26 months, the layer that supplies the industry is not on the board, and a person working by hand reached four of its members in one afternoon.

07
Finding five

Counting a portal as a broker

OptOutPreScreen is listed on the board as a broker with exposures removed. It is not a data broker. It is the opt-out portal that the credit bureaus run themselves, under federal law, so that consumers can stop pre-screened credit and insurance offers. Using it is a good idea, and the guided run registered there too, for five years.

It does not belong in a broker count. A consumer opt-out portal appears on the board as a defeated adversary, while the bureaus that operate it do not appear at all.

08
Side by side

Same person, same day

The comparison below is deliberately unflattering to nobody. Where the service acted, it says so. Where the guided run is still waiting on a legal response window, it says that too.

Consumer sites where a live listing was located by hand

SiteService status, 26 months inGuided run, one day
FastPeopleSearchNo exposures foundSearch de-index approved, opt-out submitted
FastBackgroundCheckNo exposures foundSearch de-index approved
TruePeopleSearchNo exposures foundSearch de-index approved, opt-out submitted
SpokeoNo exposures foundRemoval confirmed
WhitepagesNot on the boardVerified opt-out submitted, pending
CyberBackgroundChecksNot on the boardOpt-out submitted, pending
USPhoneBookNot on the boardOpt-out submitted, pending
What de-indexing is, and is not

Approved search de-indexing removes the listing from search results. It does not remove the record from the broker. It is reported here as exactly that, and the direct opt-out was submitted separately wherever the site offered one. The relevant point is narrower and harder to argue with: the listing was live at the broker on the day the dashboard read clean.

Where both acted

RegistryService statusGuided run
Credit prescreen opt-outExposures removed, 2 of 2Five-year opt-out active
Direct mail suppressionNo exposures foundActive through 2031
National Do Not CallNot on the boardPermanently registered

The service's own removal profile, 26 June 2024 to 22 August 2026

999
Address records
removed, all time
908
Family relationship
records removed, all time
218
Name records
removed, all time
132
Phone records
removed, all time
128
Email records
removed, all time
30
Brokers still
in progress

999 + 908 + 218 + 132 + 128 = 2,385, the all-time total the dashboard reports. The in-progress broker count is separate.

09
What we do differently

Work the ownership, not the list

A guided run is not faster because it is clever. It is faster because it starts from the corporate record instead of a list of website names, and because a person can read a filing and decide what it means.

  • Start at the wholesale layer. The suppliers are worked first, because a record removed downstream while the supplier still holds it is a record that can come back.
  • Group by legal entity, not by brand. One opt-out at the parent covered four consumer sites in this run. Ten domains behind one company are treated as one target, not ten, and not three.
  • Verify by looking. Every site was searched by hand. “No exposures found” is a claim; a live listing on the screen is a fact.
  • Use the free federal and industry registries first. Do Not Call, credit prescreen opt-out and direct mail suppression cost nothing, last for years, and were all active by the end of day one.
  • Report status honestly. Pending means pending. Statutory response windows run up to thirty days, and nothing is described as removed until it is confirmed removed.
  • Leave the client a record. Every submission, date and response is tracked, so the work can be re-run, escalated, or handed to someone else without starting over.
Who this is for

Nobody in this document is a public figure, and none of this requires being one. The people most affected by an accurate home address on a free search site are ordinary: someone who left a bad relationship, a nurse or a teacher who was doxxed by a stranger online, a parent who would rather their children's names not be published alongside their street. The industry treats those records as inventory. The work is getting them out of inventory, and knowing who actually holds them.

10
Integrity

What this does not prove

Everything above is checkable. So are the limits, and they belong in the same document as the findings.

Read these before quoting anything above
  • “No exposures found” may be a matching miss, not a false claim. If the automation never matched a listing that a manual search caught, the status is wrong without anyone having lied. The finding is that automation reported clean where a person found a live record, which holds either way.
  • The in-progress queue is a recent batch. The dashboard shows those brokers at day nine of processing. It is not accurate to say they have been stuck for two years, and this document does not say so.
  • The wholesale submissions are pending, not confirmed. Statutory windows run from fourteen to thirty days. The claim being made about the wholesale layer is one of scope: the service does not attempt it, and the guided run did.
  • Shared addresses show structure, not data flow. Co-location and shared ownership establish that clearing the front ends leaves connected entities unresolved. They do not establish that any particular supplier feeds any particular website.
  • No claim of re-seeding is made. This document does not assert that an uncleared parent re-publishes a removed record. That would need evidence of the transfer itself, which is not in hand.
  • One account, one person, one day. This is a documented case, not a controlled study. A second account might produce different matching results, and a longer window would test durability, which this snapshot cannot.
  • The service does real work. Thirteen brokers show completed removals, and the credit prescreen opt-out was genuinely executed. The argument here is about which layer of the industry is being worked, not about whether anything happens at all.