The Form 5500 is an index. The evidence is in the attachments
Return- and schedule-level tables stop before the attached audit documents. The opinion language, the delinquent-contribution disclosure, the forfeiture balance and the named holdings are one layer down, in documents that are just as public.
Ask a return-only Form 5500 dataset a question that informs a professional decision and watch where it stops.
Did plan management elect an ERISA section 103(a)(3)(C) audit, and what opinion did the auditor issue? Not a structured field. Were delinquent participant contributions reported, and how much? The return may flag the condition; detail may appear in supplemental schedules or notes. What balance is disclosed for forfeitures, and how has it changed? That may appear in the notes. What holdings are named? Look to the Schedule of Assets Held at End of Year. What auditor compensation was reported? Look to an applicable Schedule C.
These are ordinary professional questions. When a filing discloses an answer, it may appear in a public attachment rather than in the downloadable return tables.
The filing has layers, and they carry different weight
It helps to be precise about what a “filing” contains, because the layers are not interchangeable.
| Layer | What lives there | What it can support |
|---|---|---|
| The return | Plan identity, sponsor, participant counts, assets, schedule check-boxes | What the administrator reported for that year |
| Schedules | Schedule H financial detail and compliance questions, Schedule C service-provider compensation | Reported detail, but only where the schedule applies to that filer |
| The auditor’s report | Opinion type, scope language, the 103(a)(3)(C) election | Document-level evidence of what the auditor actually said |
| Financial statements and notes | Forfeiture disclosures, related-party detail, subsequent events, plan-specific disclosures | The narrative the numbers sit inside |
| Supplemental schedules | Schedule of Assets Held at End of Year, delinquent contributions schedule | Named positions and specific compliance conditions |
The first two rows are available as structured filing fields. The bottom three are where the questions above get answered, and they arrive as attached documents.
Four things you can only get one layer down
The opinion, and what kind it is
Whether a report expresses an unmodified, qualified or adverse opinion, or includes a disclaimer of opinion, is a fact about language in a document. So is whether plan management elected an ERISA section 103(a)(3)(C) audit — the successor to what many still call a limited-scope audit. That election determines the reporting framework for certified investment information; it is normal scope context, not an audit-quality flag.
Delinquent participant contributions
The compliance question flags the condition. The supplemental schedule and notes may report the amount, affected periods, and management’s description of any correction. A hypothetical disclosure of “$41,000 across four payroll periods, corrected with lost earnings in the following year” illustrates the difference between a flagged box and a source-specific review question; it is not an unidentified filing fact.
Forfeiture balances
Forfeiture balances may be disclosed in the financial-statement notes. A balance reported across multiple years is a source-specific review signal, not a conclusion about whether the plan used forfeitures properly. The governing plan document controls.
Forfeitures in a defined-contribution plan are applied under the plan document — to pay plan expenses, reduce employer contributions, or reallocate to participants. They do not “revert.” Reversion is a different concept: plan assets returning to the employer, which is a defined-benefit and plan-termination matter under ERISA section 4044 and taxed under IRC section 4980. Using the two words interchangeably is a reliable signal that copy was written from a summary rather than from a plan document.
What the plan actually holds
The Schedule of Assets Held at End of Year names positions. For a hypothetical $47 million plan, that can mean seeing which funds hold the assets, in what share classes, and at what reported values instead of seeing only the total. The schedule can inform the qualifying-asset review, but it does not decide waiver eligibility by itself; custody, enhanced bonding where applicable, required Summary Annual Report disclosures, and access to supporting statements remain separate conditions.
Why document extraction is difficult
Not because it is secret. Because it is annoying.
- The attachments are documents, not records — layout varies by preparer, and a page can be text, a scan, or both.
- Opinion language is not standardized at the sentence level. Classifying it is a reading problem, not a lookup.
- Coverage is uneven. Some attachments are missing, some arrive late, some are amended.
- And every extracted fact has to keep its plan year and source document attached, or it is worse than useless — a fee from 2021 presented as current is a wrong answer delivered confidently.
That last one is the constraint that shapes everything. Flattening the layers into one row per plan destroys exactly the information that makes the evidence usable. A fact that cannot be traced back to a filing year and a document should not be shown as a fact.
What the documents cannot do
Being clear about the ceiling matters as much as the floor. Public attachments are the auditor’s report and the plan’s financial statements. They are not workpapers. Nothing in a public filing can tell you whether the underlying audit procedures were adequate.
This is worth stating plainly because the temptation runs the other way. EBSA’s November 2023 assessment of employee benefit plan audit quality found that 30% of the audits it reviewed for the 2020 filing year contained major deficiencies with respect to one or more relevant GAAS requirements — deficiencies that would lead to rejection of the Form 5500 filing. That is an improvement on the 39% in the 2015 study, and the finding that actually matters is the stratification: deficiency rates were concentrated among firms performing very few plan audits a year, while firms with larger EBP practices had materially lower rates.
None of which can be assessed from a filing. What a filing can show is narrower and still useful: that no auditor is named, that the indexed attachment inventory does not include an auditor’s report, that the opinion carries a modification, or that the reported auditor changed. Those are observations about the public record, not judgments about anyone’s work.
The practical test
If you want to know whether a data source reads filings or reads forms, ask it for one plan:
- What type of opinion did the 2024 report carry, and was the 103(a)(3)(C) election made?
- What was the forfeiture balance disclosed in the notes, and did it move from the prior year?
- Name three holdings from the Schedule of Assets, with their share of plan assets.
- Was an auditor’s report attached to the most recent filing at all?
Each item may be publicly available when disclosed in the filed attachments. A return-only dataset cannot answer the document-level questions.
Sources
- EBSA, Assessment of the Quality of Employee Benefit Plan Audits, November 2023. PDF
- AU-C 703, Forming an Opinion and Reporting on Financial Statements of Employee Benefit Plans Subject to ERISA (SAS 136), including the ERISA section 103(a)(3)(C) audit.
- Form 5500 instructions and schedule requirements for the applicable plan year, including Schedule H, Schedule C and required supplemental schedules.
- DOL/EBSA public Form 5500 filing data and filed attachments. Scope and limitations at Coverage.
See it on a record
An illustrative plan showing how each fact keeps its filing year and source document attached, rather than being flattened into a row.