The Search You Can't Prove You Ran
There is a peculiar asymmetry in the Department of Labor's missing-participant guidance that most compliance programs never quite internalize. The Department tells you, in mandatory language, what searching you must do. It tells you, in equally mandatory language, that you must be able to demonstrate you did it. But the enforcement record shows that plans rarely fail on the first duty. They fail on the second.
A fiduciary who searched diligently and kept nothing is, at an audit, indistinguishable from a fiduciary who did nothing at all. That is the problem this piece is about.
The duty is unambiguous, and it is a duty to search and to show
Field Assistance Bulletin 2014-01 sets out four required search steps for terminated defined contribution plans, and it is careful about the verbs:
"Some search steps involve so little cost and such high potential for success that a fiduciary should always take them before abandoning efforts to find a missing participant, regardless of the size of the participant's account balance. The failure to take such steps would violate the fiduciary obligations of prudence and loyalty, as set forth in section 404(a) of ERISA."
The fourth step is the one worth dwelling on, because it has aged unusually well:
"Plan fiduciaries must make reasonable use of Internet search tools that do not charge a fee to search for a missing participant or beneficiary. Such online services include Internet search engines, public record databases (such as those for licenses, mortgages and real estate taxes), obituaries and social media."
Must make reasonable use of. Not may. The Department's reasoning was that free tools had become good enough to displace the letter-forwarding services it previously required, noting that "in many cases, these tools may now be more effective at locating missing participants than either the IRS or SSA letter-forwarding services."
Then, separately, the evidentiary duty:
"Plan fiduciaries must be able to demonstrate compliance with ERISA's fiduciary standards for all decisions made to locate missing participants and distribute benefits on their behalf. If audited, plan fiduciaries could demonstrate compliance using paper or electronic records."
Read those two sentences precisely, because the difference between them is routinely blurred. The first is mandatory and unqualified. The second is permissive as to medium only — the Department is saying the format is flexible, not that documentation is optional. And note the scope: not just searches, but "all decisions made to locate missing participants and distribute benefits on their behalf." The deliberation is discoverable, not only the execution.
The 2021 guidance made "regularly" the standard, and changed the shape of the artifact
The Department's Missing Participants — Best Practices for Pension Plans lists ten search methods and closes with a bullet most readers skim past:
"Searching regularly using some or all of the above steps."
That single word does more work than the nine bullets above it. A one-time search is not the standard — which means the evidentiary artifact is not a memo. It is a series, with dates, showing the same question asked repeatedly and answered differently over time. The Department reinforces the point elsewhere, describing plans that use best practices "as part of their ongoing culture of fiduciary compliance rather than just as one-time or sporadic 'fixes.'"
The document also carries a disclaimer, and any honest treatment has to quote it:
"The contents of this document do not have the force and effect of law, and are not meant to bind the public in any way."
Which is exactly what practitioners seized on. Groom Law Group's assessment at the time was blunt: the guidance "does not establish the type of clear, bright-line rules many plan sponsors and services providers were asking for," and there was "the risk that the DOL's 'best practices' examples could be used as prescriptive remedies," contrary to ERISA's facts-and-circumstances standard.
That criticism was fair in 2021. It is substantially weaker now, for a reason that has not been widely absorbed.
FAB 2025-01 turned "best practices" into a condition of relief
In January 2025 the Department issued Field Assistance Bulletin 2025-01, creating a temporary enforcement policy for transferring benefits of $1,000 or less from an ongoing plan to a state unclaimed property fund. The relief is conditional, and condition two reads:
"The plan fiduciary has implemented a prudent program to find missing participants consistent with the Department's Best Practices for Pension Plans, and nevertheless has been unable to locate the participant or beneficiary;"
This is the pivot. A document that disclaims legal force on its own face is now a precondition of an enforcement safe harbor. A fiduciary claiming that relief must be able to establish a Best-Practices-consistent search program — and "establish" means produce something. The arc from soft guidance to enforcement condition took four years, and it closed the gap Groom identified.
The same conditionality appears in the PBGC context. FAB 2021-01 is explicit that handing a balance off launders nothing:
"However, this temporary enforcement policy does not preclude the Department from pursuing violations under sections 404 or 406 of ERISA for a failure to diligently search for participants and beneficiaries prior to the transfer of their account balances to the PBGC or from pursuing violations under sections 107, 209 or 404 of ERISA for a failure to maintain plan and employer records."
Two independent theories, named separately: failure to search, and failure to maintain records. The second does not depend on the first.
One clarification, because secondary commentary gets it wrong constantly: FAB 2021-01 says it controls "to the extent this memorandum conflicts with the distribution guidance of Field Assistance Bulletin 2014-01." The limiting phrase is doing real work. FAB 2014-01's search steps are untouched, and the Department cited FAB 2014-01 as live authority as recently as FAB 2025-01.
What an examiner actually asks for
The Department published its own investigative methodology in Compliance Assistance Release 2021-01, covering the Terminated Vested Participants Project. The search-specific passage deserves close reading:
"We also seek information to determine whether the plan takes sufficient steps to address missing participant situations when they occur. This includes examining internal procedures and practices for reaching out to unresponsive TVPs or for searching for them (for example, through the plan sponsor's human resources department or in the records of related employee benefit plans), and contracts and experience with third-parties who perform recordkeeping and missing participant search functions for the plan. These documents and information, combined with interviews of the relevant parties, help us to verify what the plan, responsible plan fiduciaries, and relevant service providers do (or do not do)..."
Four things are being requested, and the fourth catches people. The Department wants the procedures, the vendor contract, the experience with that vendor — results, not merely the engagement letter — and then it cross-examines the humans against the documents. A written policy describing searches nobody performed does not survive the interview stage. It makes things worse, because now there is a documented standard the plan can be shown to have missed.
The named failure modes are equally specific:
"Continuing to deliver required communications to a known 'bad address' without taking steps to verify the correct address, failing to use simple methods and resources like the USPS Address Correction Service and/or the National Change of Address (NCOA) database to find replacement addresses, can indicate problems. EBSA's experience is that many plan fiduciaries fail to take advantage of or properly use the missing participant search and location services offered or provided by the plan's record keeper or other service providers. Fiduciaries should carefully monitor the performance of retained, third party search firms, and ensure that they are complying with any applicable contractual commitments regarding missing persons."
Notice what the third sentence assumes. The Department expects that a plan paying for a search service can show the service performed. That is not an abstract governance nicety — it is an evidentiary obligation, and it is unmeetable if the vendor's output is a status count rather than a record of what was actually examined.
Practitioners describe the same thing from the other side. Heather Bader of Faegre Drinker, speaking to PLANSPONSOR earlier this year, noted that although the guidance does not by its terms require a written policy, the Department has "come down heavy" on wanting to see something written — and put it plainly: "If the DOL investigates them, they have to show they've been taking appropriate action. It's better for plan sponsors to start looking now than in the midst of an investigation." Foley & Lardner titled their section on the subject, simply, "Document, Document, Document."
Buying a service does not close the loop
Groom's sharpest observation in 2021 was framed as a question, and nobody has really answered it:
"When faced with these 'best practices,' would plan fiduciaries have a duty to supplement the work of a commercial locator service if the service, for example, does not use social media to locate individuals?"
That is a top-tier ERISA firm pointing out that engaging a paid vendor may leave a residual, provable obligation with the fiduciary. The Best Practices document lists free search tools and commercial locator services as separate, parallel bullets in a list it expressly declines to prioritize — "The examples are not listed by priority or in any other particular order" — while making cost a legitimate input.
For the avoidance of doubt: no Department text requires a fiduciary to purchase a paid search service. FAB 2014-01's additional-search-steps paragraph is careful — the duty to consider further steps is mandatory, the steps themselves are "possible," and fiduciaries are directed to weigh "the size of a participant's account balance and the cost of further search efforts." The closest the guidance comes to requiring paid tools is conditional on both a large enough balance and the exhaustion of other efforts.
So the exposure is not "did you buy the right service." It is "can you show what you looked at, when, and why you concluded what you concluded." That is an evidence problem wearing a search problem's clothes.
The hard case is not finding records. It's deciding which ones are the participant.
Foley & Lardner put the real operational difficulty better than any vendor has:
"Determining which 'Mary Smith' from Chicago is the missing participant you're looking for could prove to be a daunting and time-consuming task."
This is the part of the job that resists automation and produces the most fiduciary risk, because it is where a search stops being a search and becomes a judgment. Finding twenty records for a common name is trivial. Deciding that three are your participant, that seventeen are not, and being able to say why for each — that is the work. And it is precisely the reasoning an examiner will probe, because it is where a plan either has a defensible basis or has a guess.
An honest evidence record for that judgment has a specific shape. It says what was queried and when. It preserves what was returned — the URL, the snippet, the retrieval timestamp — so the finding can be re-examined months later by someone who was not there. It distinguishes between a record affirmatively matched to the participant, a search that returned nothing, and a search that returned something the evidence could not resolve either way. And it is willing to say indeterminate rather than force a match, because an overstated match in a fiduciary file is worse than an acknowledged gap.
That last point deserves emphasis. A search log that never reports uncertainty is not a rigorous log; it is a log that has quietly converted judgment calls into assertions. Under examination, the difference shows.
Where Sirveil fits, and what we are not
Sirveil operates the Sirveil Exposure Verification API. It answers a narrow question — whether a named person is currently indexed on a named public website — and returns the matched evidence with a retrieval timestamp, in one of three states: indexed, not indexed, or indeterminate. It is metered per call: ten cents for a single named-site check, thirty-five cents for a sweep across a pinned set of people-search domains, with a coverage block reporting what was queried, what came back empty, and what was not reached. Failed calls are not billed. It is available in AWS Marketplace.
What matters for this audience is not the price. It is that every answer is a dated, replayable record, and that the API declines to guess — it reports what it examined and did not attribute, alongside what it did.
We are not a consumer reporting agency, a background screening service, a commercial locator service, or a credit reporting agency, and the API's terms prohibit FCRA-governed uses. We say this explicitly because the Department's own lists of search methods name credit reporting agencies and commercial locator services, and we do not want to be read as offering to fill that bullet. We do not. The bullet we speak to is the free-electronic-search-tools duty and, more importantly, the documentation duty underneath every one of them.
Nor do we claim any plan is required to use us. The guidance requires reasonable use of fee-free search tools and the ability to demonstrate what was done. How a plan satisfies the second half is its own fiduciary judgment.
Read more about the Exposure Verification API →
The short version
The Department requires the search and requires that you be able to prove it. Its 2021 guidance made "regularly" the standard, which makes the artifact a dated series rather than a memo. Its 2025 bulletin made a Best-Practices-consistent search program a condition of enforcement relief, which gave documentation teeth it did not previously have. Its own published methodology asks for procedures, vendor contracts, vendor results, and then tests all of it against interviews. And the hardest judgment in the whole process — which of several same-named records is actually your participant — is the one least likely to be written down and most likely to be asked about.
If your plan's answer to "show me what you looked at" is a status count from a vendor dashboard, that is worth examining before someone else examines it.
Frequently asked questions
Does ERISA require plan fiduciaries to document their missing-participant searches?
Yes. Field Assistance Bulletin 2014-01 states that "Plan fiduciaries must be able to demonstrate compliance with ERISA's fiduciary standards for all decisions made to locate missing participants and distribute benefits on their behalf," adding that if audited, fiduciaries "could demonstrate compliance using paper or electronic records." The first sentence is mandatory; the second is permissive only as to the medium. The scope covers decisions, not just searches.
Are plan fiduciaries required to pay for a commercial search service?
No Department of Labor text requires purchasing a paid search service. FAB 2014-01 makes the duty to consider additional search steps mandatory while describing the steps themselves as "possible," and directs fiduciaries to weigh the participant's account balance against the cost of further searching. The 2021 Best Practices document lists free and paid tools as separate, unranked options and states that the examples "are not listed by priority or in any other particular order."
Did FAB 2021-01 supersede FAB 2014-01?
Only in part, and the limiting language matters. FAB 2021-01 says it controls "to the extent this memorandum conflicts with the distribution guidance of Field Assistance Bulletin 2014-01." It displaces the distribution-option hierarchy for terminating plans using the PBGC Missing Participants Program. It does not touch FAB 2014-01's required search steps, which remain the Department's only required-search-step guidance. FAB 2025-01 cites FAB 2014-01 as live authority.
What does FAB 2025-01 change about missing-participant searches?
It makes a documented search a condition of enforcement relief. FAB 2025-01 provides a temporary enforcement policy for transferring benefits of $1,000 or less from an ongoing plan to a state unclaimed property fund, conditioned on the fiduciary having "implemented a prudent program to find missing participants consistent with the Department's Best Practices for Pension Plans." That converts a document which disclaims the force of law into a precondition a fiduciary must be able to establish.
What does a DOL examiner ask for in a missing-participant investigation?
Compliance Assistance Release 2021-01 describes the Department examining "internal procedures and practices" for searching, and "contracts and experience with third-parties who perform recordkeeping and missing participant search functions," then combining those documents with "interviews of the relevant parties" to verify what the plan and its providers actually do. The Department also states that fiduciaries "should carefully monitor the performance of retained, third party search firms."
Is a record not appearing in a search the same as proof it was removed?
No. A search result establishes that a named person was not indexed on a named surface at a specific timestamp. It does not establish that a record was deleted from any underlying system, and it does not establish compliance by the operator of that surface. Evidence records should distinguish clearly between indexed, not indexed, and indeterminate outcomes rather than collapsing them.
Sources: DOL EBSA, Field Assistance Bulletin 2014-01 · FAB 2021-01 · FAB 2025-01 · Missing Participants — Best Practices for Pension Plans · Compliance Assistance Release 2021-01. Commentary: Groom Law Group · Foley & Lardner · PLANSPONSOR.
This article describes federal guidance as published and is not legal advice. Fiduciary obligations depend on the facts of a particular plan; consult ERISA counsel.