Small-plan fee dispersion is the finding, and advisors own it

Survey data has a sampling problem, as Admin316's Form 5500 analysis puts it: the plans that answer benchmarking surveys are not the plans that most need benchmarking, so Russell McNorton went to the filings. His sample is every 2024 plan-year Form 5500 with a Schedule H, filtered to 401(k) plans, pension feature code 2K, at least 25 active participants, positive administrative expenses and positive year-end assets — 48,944 plans in all, no sampling and, in the analysis's framing, no vendor supplying its own numbers. The median administrative cost across the set is $162 per participant, or 27.0 basis points of assets.

The analysis offers that median as the number worth carrying into a client meeting, then qualifies it hard: a median without a participant band says almost nothing, because scale drives nearly everything in this data. A plan of about 60 participants pays roughly $364 per person, against $104 for a plan of about 900; in basis points the two sit much closer, 32.5 against 18.6, and that narrowing is the diagnosis. Small-plan cost is dominated by fixed administrative work — compliance testing, filings, notices, distributions, headcount-driven recordkeeping — that costs what it costs regardless of the asset base. The filter also draws a boundary worth noticing: plans reporting no administrative expense, and plans below 25 active participants, sit outside the frame entirely.

Read down the bands and the between-band gap looks tidy: $364 a head at 25 to 99 participants, $228 at 100 to 249. Read across each band and the tidy part disappears — in the smallest band the 90th percentile is $977, while in the next it is $548 against a median of $228. Two plans of identical size, identical assets and identical demographics can sit two and a half times apart on administration that is functionally the same work, and across plans under 500 participants, 30.6% pay more than twice the national median per participant, roughly 10,500 plans in the 2024 filings alone.

That dispersion changes what a plan review is for. The question that produces a decision is not whether a plan is expensive for its size but why it sits in the top decile of its own band.

Four explanations recur in review work: an advisory or recordkeeping arrangement priced when the plan was half its current size and never revisited; per-participant charges still being paid on terminated participants with small balances; administrative work bought twice, once inside the bundle and once from a specialist; and a fourth item, touching discretionary duties, cut off mid-sentence in the published excerpt. The first three are enough to run against any plan in the sample, and none of them is fixed by another share-class conversation. What they share is a provenance — each is an artifact of what the plan used to be rather than a decision anyone made this year — which is why they survive annual fiduciary review: the review asks whether fees are reasonable for the plan as it exists, and a legacy contract answers a question about a plan that no longer does.

That fixed-cost structure is why the story advisors hear most often from small sponsors — fee relief will arrive with asset growth — rarely holds for a plan under 100 lives. Compliance testing, Form 5500 preparation, participant notices and distributions are functions of lives, not of dollars, so growth compresses the denominator without shrinking the numerator. What remains once that arithmetic does its work is 32.5 basis points against 18.6.

For a plan under 100 lives the lever is service scope and provider structure, a harder sale than a cheaper share class because it asks the sponsor to change what the plan buys rather than what it pays for one line of it. The fee review is among the few recurring, defensible deliverables an advisor serving small plans can own, and its worth depends entirely on whether the comparison holds up. A median pulled from a survey is a talking point; what a committee can act on is the plan's own filing, positioned inside its band.

Of the recurring causes, the terminated-participant charge is the one an advisor can usually find and fix without reopening a single contract: a per-head fee paid on people who no longer participate that inflates the plan's reported cost per active participant and distorts every comparison the sponsor makes against its peer group.

This publication has argued that participant-level advice is moving from education toward diagnosis, and that procurement is the new gate. The data extends that argument past where the thesis usually stops: if procurement is the gate, the sponsors sitting in the top decile of their own band are the ones for whom a competitive process is most overdue, since a stale arrangement or a service bought twice is exactly what a competitive process catches. The filings cannot measure a sponsor's internal bandwidth, but they can count the exposure — about 10,500 plans under 500 participants paying more than double the median for every head. For an advisor, the product is a plan-by-plan answer to the top-decile question, read out of the client's own filing. The next test arrives with the following plan year's filings, and nothing in the structure of small-plan fees suggests the 30.6% share will come down.