How the scorecard is computed

A card has four slots and they never move. A slot with no figure says which filing field is missing rather than falling back to a different number, because a card that reorders itself around whatever is available is a card that answers a different question on every page.

Dollar figures only compare if they are computed on the same worker, so every page uses one salary: $49,500, the US median annual wage, all occupations, May 2024, published by the Bureau of Labor Statistics, Occupational Employment and Wage Statistics. It is deliberately not $50,000 or $60,000. A round number picked because it looks tidy would make every dollar figure on this site something we chose; a published median makes it the filing restated at a stated salary. The same release puts the median hourly wage at $23.80, which is $49,504 over 2,080 hours, so the two agree.

Every dollar figure scales in a straight line with this number. A reader earning twice the median doubles it, and a reader earning half of it halves it. Percentages of pay do not move at all.

Source: Bureau of Labor Statistics, Occupational Employment and Wage Statistics, US median annual wage, all occupations, May 2024 · Bureau of Labor Statistics ↗

The figure is the employer contribution for one plan year, at the standard salary, for a worker who contributes enough to collect all of it. Filings state that in six different shapes, each with its own arithmetic, and there is a seventh case where the filing does not fix the terms and there is no figure at all. Figures are rounded to the nearest dollar at the last step and nowhere else.

Employer dollars answer what is offered. The vesting score answers how fast it becomes yours. It measures one term, the vesting schedule for the employer match, and it measures it precisely. That is a deliberate narrowing: a number that blends several terms is only available on the pages where the filing happens to state all of them, and it hides which term moved it.

The formula

Read the share of the employer’s money that is yours at hire, then at each of the first 6 anniversaries. That is 7 readings. The score is their average, and nothing else is in it. It is rounded to two decimal places once, at the end. Two decimals rather than none, because a schedule vesting 33% then 66% then all of it at 4 years averages exactly 57, while a 3-year cliff averages 57.14, and rounding to whole numbers would print two different schedules as the same figure.

6 years is the window because it is the longest graded schedule the law allows for matching contributions, so every lawful schedule is finished inside it and none is cut short by it. One measure covers all three shapes a filing states: immediate vesting reads 100 at every point, a graded schedule reads its own staircase, and a cliff reads 0 until it lands. That ordering is not imposed on top of the arithmetic, it falls out of it. A cliff and a graded schedule that both finish in five years are not the same offer, and the cliff scores lower because for four of the seven readings the worker owns none of the money.

A cliff at 3 years and a graded schedule finishing at 5 both score 57.14, which is worth sitting with. They are different offers, and over these 7 readings they hand the worker the same total. The score does not claim they are the same schedule; it claims they move the same amount of money across the window, and the card prints the schedule in words beside the number so the shape is never hidden by it.

When there is no score

Three cases carry no number at all: a filing that states no schedule for the employer match, a cliff whose length is not stated, and a schedule stated in quarters rather than years, which cannot be read at anniversaries without inventing the steps in between. A fourth is a graded schedule the filing never carries to 100%, where the years after the last stated step are unknown.

None of those is filled with a middle value, an average or a default. The page names the gap and moves on. Of the 266 plan records read so far, 253 state a schedule this can read and carry a score.

Two other filed terms decide what a worker actually collects: the wait before the match starts, and whether the plan tops the match up after year end. Both used to sit inside the score. Both are now printed under it as facts, with the filing sentence behind each, wherever the filing states them, and left out entirely where it does not.

The reasons for taking them out are different, and both are worth stating plainly.

The wait: too few filings state it

A wait can only be read where the filing describes who qualifies for the match. 93 of the 266 plan records carry that sentence, and 56 of them state one period clear enough to put in words. Requiring it before showing any number left four pages in five with no number, which is not a measure, it is a gap wearing one. So the wait is printed where it is stated and absent where it is not.

It is still read from the match-eligibility sentence alone. Plan-entry text is not used: it says when you may start contributing, which is a different date. Five kinds of sentence yield no wait at all, because what they describe is not a waiting period:

• terms set by collective-bargaining agreements, which differ by agreement;
• a requirement to still be employed at the end of the year or quarter, which is a different term with a different effect;
• a change to the requirement rather than one period in force;
• different rules for different hire dates;
• two different periods in one sentence, or two alternative conditions, because picking one of them would be a choice rather than a reading.

The true-up: silence is not absence

A true-up is the payment a plan makes after year end to a worker whose per-payroll match fell short of the yearly formula, which happens whenever contributions are uneven or stop early at the annual limit. 40 of the 266 plan records state one.

The old score gave the other 226 zero points for it. That was the site breaking its own rule. An 11-K that never mentions automatic enrollment is labelled “not mentioned” here, not “none”, because a filing is a summary and a summary leaves things out. Scoring an unmentioned true-up as zero says the opposite: that a plan which did not describe one does not have one. It cannot be read from the filing, so it is not scored, and a filing that is silent on a true-up now produces no line and costs nothing.

Every number below is produced by the same calculation the company page runs, so this example cannot drift from the page it explains. The filing is the Form 11-K for the plan year ended December 31, 2025, filed June 24, 2026.

Step 1. What the employer puts in

A worker on $49,500 who contributes 6% of pay collects the full match of 3% of pay.

$49,500 × 3.00% = $1,485 a year.

Step 2. Read the schedule

The filing says: “Participants are always 100% vested in their own contributions and earnings. The Company matching contributions vest 20% annually until the participant is 100% vested at the end of five years.”

That is graded, full at 5 years. Read at hire and at each of the first 6 anniversaries, the share of the employer’s money that belongs to the worker goes:

Step 3. Average them

(0 + 20 + 40 + 60 + 80 + 100 + 100) ÷ 7 = 57.14 out of 100. That is the whole score. Higher than 4% of the 253 schedules we have scored.

Step 4. Print what else the filing states

Two further terms can appear under the score. For this plan the filing states one of them:

This filing says nothing about a true-up, so there is no true-up line and no penalty for its absence. A filing that does not describe one is not a plan without one. Neither term changes the score above.

The fourth slot on the card is not computed here at all. The Form 5500 record for this plan puts it in the 87th percentile for plan-paid administrative cost among plans of its size, and the card prints that figure as filed.

See the same numbers on the Northern Trust page.

Source: Northern Trust Corporation, Form 11-K for the plan year ended December 31, 2025 · Form 11-K on EDGAR ↗

The score is not a rating of a 401(k) plan. It measures vesting speed and nothing else. It says nothing about the size of the match, what the plan charges, or the funds inside it. A plan can vest the match on day one and put in a fraction of what a slower-vesting employer puts in. Six things that matter to what a worker ends up with are outside the number, and a reader who forgets that will over-read it.

The fourth slot on the card covers one part of the cost question: what the plan pays for administration, per participant, against plans of its size. It does not cover fund expense ratios, which participants pay inside the funds themselves. Nothing on the card covers how well the plan is run, how quickly contributions are deposited, or how loans and withdrawals are handled, because none of that is a filed term.

Pages may print one comparative sentence about a score, and only this one. On the Northern Trust page it reads: “Higher than 4% of the 253 schedules we have scored.” It says where a schedule sits among the 253 we have scored, it counts only schedules scoring strictly lower, and it names how many the comparison covers. Immediate vesting is the most common shape in the file, so many employers sit on the same number; a plan at the bottom of the range is therefore told that nothing scores lower, not that it is below every other plan. Pages do not call a plan good, strong, generous or poor. A score is a comparison of filed terms between employers, and it stops there.

Everything above reads a Form 11-K. Around 691 employers file one in a year, because the obligation follows company stock sitting in the plan rather than the size of the plan. All 57,824 large 401(k) plans file a Form 5500, and a Form 5500 states no match formula anywhere in it. For the plans with no 11-K behind them, this site could once show assets, participants, fees and a recordkeeper, which does not answer the question anybody arrives with.

A Form 5500 does state one thing that answers it. Schedule H, part II, line 2a is contributions received during the plan year, split between employer money and employee money. That is not a promise about a formula. It is what actually arrived. The Monitor Score is built from it and from the fee figure that was already on the page, and it works for a plan whose employer files nothing else.

The name

The figure carries the publisher’s name, as 401(k) Monitor Score, because that is what it is: one publication’s reading of two amounts a plan filed, which is attributable and arguable, and not a regulator’s finding or a rating agency’s grade. The scope line under it says what it is built from and what it cannot read, and that line travels with the figure everywhere the figure goes.

It scores two filed amounts against comparable plans and nothing else. It is not a judgement of the funds in the plan, of the recordkeeper, or of the employer, and it is not advice about a job or a contribution rate. It is always printed as a number out of 100 and never as an ordinal. A plan at 91 is at 91 out of 100 on this measure; it is not ranked 91st, and it is not higher than 91% of plans. Those would be different claims and both would be false.

The two factors

A dollar figure does not compare across employers, so neither factor is compared raw. Each one is a rank inside a group of plans chosen to hold the obvious differences still.

Both ranks are computed once over the whole 57,824-plan file and written into each plan record. No page ranks or divides anything of its own, so two pages cannot disagree about where a plan sits.

The formula

Multiply each rank by its weight and add the two. The result is rounded to two decimal places once, at the end, and nowhere else.

score = 0.7 × (rank on employer money per active participant) + 0.3 × (share of plans of its size that cost more per participant)

0.7 times where the plan sits among its peers on employer money per active participant, plus 0.3 times where it sits among plans of its size on plan-paid cost. Both standings come from the filings, and the plan page prints each of them beside the dollar figure that produced it.

The arithmetic is published here and not beside the figure. A plan page prints both components under the score, each with the dollar figure that produced it and the peer group it stands in, so a reader who wants to redo the sum has both standings there and both weights here.

Two decimals, one whole number

A plan page prints the whole number. A weighted average of two percentiles that are themselves published to one decimal place does not carry two decimals of real precision, and a figure meant to be repeated from memory cannot carry them either. The exact value is never hidden: it is in the citation line at the foot of the plan page and in that page’s structured data. So a plan reading 91 also states 91.06 further down the same page, and neither figure is a correction of the other.

In structured data the score is a PropertyValue on the page’s Dataset, carrying its value, its range and this scope sentence. It is deliberately not a schema.org Rating, AggregateRating or Review. Those types make a page eligible for star markup, which would tell every downstream consumer that this site rates plans the way a review site rates a restaurant, and that is the one claim this figure must never make.

Why 0.7 and 0.3

The two factors are not the same size of question. Across the plans published here the middle plan received $2,173 of employer money per active participant in a year, and paid $122.17 per participant to run itself. Weighting them equally would let a few dollars of cost cancel thousands of dollars of contributions.

The weights are deliberately not that ratio either, which runs to roughly sixteen to one across the file. A weight that steep would make the score a second name for one component, and what the plan charges is a real, separately sourced fact that readers ask about on its own. 0.7 and 0.3 keep the ordering, leave the cost able to move the number, and stay a round pair anyone can recompute. A reader who would weigh them differently has both numbers in front of them.

The peer group, and why it crosses industry

Employer contributions track pay, and pay tracks industry. In the 5,000-participant cohort the middle plan in pharmaceuticals and chemicals reports $8,101 per active participant and the middle plan at hotels and restaurants reports $323. A single national ranking would mostly measure which industries pay well. Size alone barely moves the middle of the distribution; industry moves it by a factor of about twenty-five, so the peer group crosses the two.

A cell holding fewer than 30 plans is too thin to rank inside, so those plans fall back to the size cohort alone, and the page names whichever group it used. The cost rank stays inside the size cohort by itself, for the reason in the table below.

The industry is the one the employer wrote on its own filing, and that is worth knowing before reading a rank. The Form 5500 asks for a business code and the sponsor picks it, so a company whose code does not describe what it does is ranked against the wrong companies. Costco files code 452300, general merchandise retail, and is compared with retail plans whose middle reports $700 per active participant. Amazon.com Services files code 541990, other professional and technical services, and is compared with consultancies and law firms instead, whose middle reports $2,987. The same dollar figure lands in a very different place in those two groups. This is why every page names the group it ranked inside, in words, rather than printing a rank on its own: a reader who knows the employer can see when the comparison is the wrong one.

How far it reaches

Of the 57,824 large 401(k) plans in the form year 2024 file, 49,551 report enough to state employer money per active participant and 49,419 of those sit in a peer group. 54,783 carry a plan-paid cost rank. 48,494 carry both, and only those are scored: that is 84% of every large 401(k) plan in the country. On this site, which publishes 57,824 of them, 48,470 carry a Monitor Score.

A plan missing either half takes no score at all. It keeps whichever figure it does report, with the reason the other one is absent, because a two-factor score computed on one factor is a different measure wearing the same name. Nothing is filled in from an average, from another year or from another plan. Where this plan stands against plans of its size and industry on two amounts its Form 5500 reports: employer money in per active person, and plan-paid cost per person. It reads no match formula, no vesting schedule and no fund lineup, because a Form 5500 states none of them.

Two filed amounts, ranked against similar plans, still carry seven ways of misleading a reader. None of them is solved by the score. Five are handled by choosing the comparison carefully and saying what is left over, and two by refusing to compute at all. They are listed here because a score that hides its confounds is worse than no score.

One more guard sits underneath these. Contributions cover a whole plan year while the active participant count is a headcount at one date, so a plan whose workforce collapsed during the year would divide a full year of money across the few people left. Those filings show the total that went in and no per-person figure at all, and they are kept out of every peer group.

Nothing here is a hidden input. Every value the formulas read is published, so a reader who wants to check a score can rebuild it from the same page it appears on.

The first four sit on the company page the score belongs to and the last two on the plan page, each with the sentence or the filing line it was read from, and the headline figures of both pages download as a CSV from the page itself. Quoting them is welcome: the terms are on the licence page. The rules the two other pages state, on sources and on how we check our own numbers, are on the methodology page.

This is a comparison of filed terms and filed amounts, and not advice. 401(k) Monitor is an independent data publication, no one here is a financial advisor, and a score is not a recommendation to take a job, leave one, or set a contribution rate. What the two scores compare is what employers have filed with the SEC and with the Department of Labor. What they cannot compare is your situation.