Every employer plan with 100 or more participants files a Form 5500 each year with the Department of Labor, with a financial schedule (Schedule H) and a schedule of who was paid (Schedule C). The Department publishes them as public data. We load every filing for the last six form years, about 575,000 of them, and build every review from that set. Nothing is typed in by hand.
Data facts from the snapshot manifest: DOL Form 5500 Latest datasets, form years 2019, 2020, 2021, 2022, 2023, 2024, snapshot 2026-09-10-r4, 574,270 Schedule H filings, 1,155,282 plans in the lookup of which 79,967 are within scope.
A plan is identified by its employer identification number and plan number together, never by name alone. You confirm the plan before anything else happens. If we cannot match a unique plan, we do not proceed.
Schedule H is a set of accounts that has to balance. We check that net assets at the end of the year equal net assets at the beginning plus net income, adjusted for transfers. We check that total income minus net income equals total expenses. And we check that total expenses minus benefits paid equals administrative expenses plus the small remaining items. Across the country these hold in roughly 99, 98, and 95 percent of filings, so a failure is a real flag about the filing itself, and a year that fails is reported as a finding rather than used for comparison.
The administrative expense total on Schedule H is compared to the sum of what providers reported receiving directly on Schedule C. We grade the result. Exact means the two agree to the dollar. Schedule C below Schedule H is common and usually explainable, because providers paid under $5,000 are not listed. Schedule C above Schedule H means providers reported receiving more than the plan reported spending, which the filings alone cannot reconcile, and it is reported as a finding. Nationally the tie is exact in fewer than half of all filings.
Administrative expenses are stated two ways: per participant, and as a percent of the average account balance in the plan. A plan with large average balances can be expensive per participant and inexpensive as a share of balances at the same time. Neither number is the right one on its own. Both are reported, each against the plan's own peers.
In the 2024 filings, about five plans in ten pay an investment adviser from plan assets. That cost is counted as part of running the plan.
The peer set is plans with the same six-digit industry code, in the same participant size band, of the same plan type, filing for the same year. If that set has fewer than 30 plans, we widen to the three-digit industry group, and we always show the all-industry set for the same size and type alongside. The set used, its size, and the reason are stated in every review. Where the set is too small to be meaningful, the review shows the plan's facts and says "filed, not yet benchmarked" instead of inventing a percentile.
Quoted from the data snapshot: "Peer pool. A filing is in the peer pool when its first two reconciliation checks pass, its administrative expense is greater than zero, its participant count is greater than zero, its average net assets are greater than zero, its plan year and form year agree, it does not cover a short plan year, and the filer is not a direct filing entity. Peer sets. The primary peer set is the pool filings with the same six-digit business code, the same participant size band, the same plan type, and the same plan year. If that set has fewer than thirty plans, the set widens to the same three-digit industry group with the same band, type and year. If that set is also below thirty, the all-industry set with the same band, type and year is used. The all-industry set is always shown alongside. Every report names the set it used, its size, and why. Percentile. A plan's percentile in a set is the share of the set's plans below it, plus half the share equal to it, times one hundred. A plan in the pool is counted in its own set. A plan that is not in the pool is ranked against the set without joining it."
Every review carries a fee rating from A to E. The letter is the plan's position among comparable plans on administrative cost per participant. A is the lowest-cost quarter, and is issued only when every condition on the report's parameters page holds. B runs to the median, C to the 75th percentile, D to the 90th, and E is the most expensive tenth. The peer set, its size, and the plan's exact percentile are printed with the letter.
Ratings are positions among comparable plans, not verdicts. By construction, one plan in four rates D or E: its cost per participant sits in the most expensive quarter of its peers. One in four sits in the cheapest quarter, and fewer than that clear every condition for an A.
Most plans elect a reporting option that lists providers paid through fund expense ratios and revenue sharing by name without dollar amounts. That is usually the largest part of what participants pay, and it cannot be rebuilt from the public filing. The review says so plainly and names the documents that would allow it: the plan's 408(b)(2) service-provider disclosure and the fund lineup with balances. We do not estimate it.
Before a number is called a finding, the review considers the ordinary reasons a plan can look unusual: a company stock fund, a self-directed brokerage window, a broader service scope, expense-account rebates, or a one-time project year. Each is marked ruled out, or not determinable from public data, with the document that would settle it.
Every review is read by an experienced plan accountant before it is sent. Their initials and the date are on the last page.
It does not use anything about you. It does not evaluate investments or say whether a fee is fair. It does not advise. It reports, checks, compares, and cites.