Benefits 303

Act I · about 11 minutes

The Year Has Four Quarters, Not One Deadline

Rider premise: You have already read your own renewal. You know what the dials do and what you are paying. Now comes the part nobody teaches, because nobody gets paid to teach it: the eleven months after you sign. Most owners treat those months as a nap between deadlines, and then wonder why the next renewal feels like an ambush.

I'm Betty. I ride this train with you. My job on this Car is small and clear: give the year a shape you can actually keep. Nothing to buy. Nothing to send me. No form, no upload, nobody to call.

Let's start with the honest picture of how a small company runs its benefits year.

Sixty days before the plan renews, an envelope or an email shows up with a number on it. The owner gets tense. Somebody gathers whatever facts can be gathered in a hurry. There's a meeting, or three phone calls that add up to a meeting. A decision gets made under time pressure. Papers get signed. Everyone exhales.

Then ten months of silence.

During those ten months, people get hired and quit. Somebody's hours change. Somebody has a baby. A bill gets paid every month without anybody checking it against the list of who actually works there. Nobody explains the plan to anyone, because the only time benefits get explained is the week of enrollment, when everybody is distracted.

That is not a badly run company. That is the normal pattern, and it's the pattern this Car is here to break.

Here's the first thing to get straight. Your benefits year has two completely different kinds of dates on it, and owners smear them together.

The first kind is genuinely time-bound. A date set by law, or by your contract, that does not care how busy you are.

Your plan year is the twelve-month period your plan actually runs on. It's written in your documents. It is not automatically January through December, and plenty of owners are wrong about their own. Everything else on this list hangs off that date.

Your open enrollment window is the stretch of days when people can join, drop, or change coverage for the coming plan year without needing a life event. It's set by the plan and the insurance company, it's usually a couple of weeks, and once it closes it's closed.

Then the notices. Say your health plan covers prescriptions. You have to tell your Medicare-eligible people whether that drug coverage counts as good enough. That's the creditable coverage notice. It goes out before Medicare's annual enrollment opens on October 15 [1]. The Children's Health Insurance Program notice goes out once a year to everybody eligible for your plan [2]. There's a handful of others in that same family, and the Department of Labor keeps a plain reference list of what goes to whom and when [2].

Then the filings. There is a yearly report on employee benefit plans — Form 5500, or the shorter 5500-SF — filed electronically with the Department of Labor through a system called EFAST2 [3]. It is generally due the last day of the seventh month after your plan year ends, which is July 31 for a plan year that runs January to December [4]. You can push that out by filing for an extension on or before that date. A timely, complete request is automatically approved to no later than the fifteenth day of the third month after the normal due date, which is October 15 for a plan year that ends in December [5]. Whether your particular plans have to file, and which version, is a question for whoever prepares them. I'm not going to guess at your situation. The due date, though, isn't a guess.

Then there's the 50-employee line Car 01 taught you. If you average at least 50 full-time employees, counting part-timers added together, you owe your full-time people a Form 1095-C. It shows what coverage you offered. For the 2025 year those had to reach employees by March 2, 2026, and reach the government electronically by March 31 [6]. Under 50, that whole paragraph is somebody else's problem.

Now say you change the plan in the middle of the year. If the change makes the summary you handed out wrong, you generally owe people notice sixty days before it takes effect [7]. That one surprises owners badly. They think a mid-year change is just a phone call.

Now the second kind of date, and this is where the money is.

There is a long list of things owners believe are locked to renewal that are not locked to anything. Nobody told them. Nobody had a reason to.

You can restate your contribution rule as a real percentage any time you want to look at it. You can reconcile your monthly bill against your actual employee list in any month of the year. You can fix a beneficiary form in March. You can read your own plan document on a Tuesday. You can find out which hospital is in your network before somebody needs it. You can ask a person why they turned down coverage. You can count how many people are enrolled and how many waived. You can write down what you decided and why. You can explain the plan to your crew in plain words on any day you feel like it.

None of that requires permission, a quote, a renewal, or a meeting with anyone. All of it is free. Almost none of it happens, because it has no deadline attached, and things without deadlines lose to things with deadlines every single time.

So here's the fix, and it's boring on purpose. A cadence beats a deadline. A deadline is a wall you run into once a year. A cadence is four short appointments that nobody has to be talked into.

Map the year onto four quarters, counted from your plan year start — not from January.

The quarter right after renewal is for talking to your people. This is the one almost everybody skips, and it is the most valuable quarter you have. The plan just changed, or just didn't change, and either way your employees are the only people in the building who don't know why. This is the quarter where you say out loud what the company spends per person, what the deductible and the out-of-pocket maximum mean in real dollars, and what to do at two in the morning. It costs a morning. It's the highest-return morning in the year, and it's the first thing that gets dropped when the shop gets busy.

The second quarter is for paperwork and plumbing. Reconcile the bill to the census — the census being your actual list of who works there and who's on the plan. Confirm people who left are off every line, not just medical. Confirm new hires started on the date the document says. Check beneficiary forms. Trace retirement money from a paycheck to the account for two pay periods and prove the timing holds. This is the unglamorous quarter, and it quietly prevents most of what goes wrong.

The third quarter is for facts. Not shopping. Not strategy. Facts, gathered while nobody is under pressure: enrollment counts, waiver counts, what you pay versus what each person pays straight out of payroll, your funding type, the numbers we'll build in the next act.

The fourth quarter is for decisions, with the people who can actually say yes in the room. That's your renewal quarter, and it's the only one most owners currently run.

Notice what that ordering does. By the time the tense quarter arrives, the facts are already sitting on a page, the bill already matches the census, and your people already understand what they have. The renewal stops being a research project performed under a clock.

Now, what actually happens to a plan that only gets touched in the sixty days before renewal? Let me be specific, because "you should stay on top of it" is useless advice.

You pay premium for people who left months ago, and when you catch it, the credit you get back is usually smaller and later than you expect. Somebody who thinks they're covered isn't, and finds out at a pharmacy counter. Your contribution rule slides another year away from what you meant it to be. A person who declined coverage never gets asked why, so you never learn the cheap fix that would have kept them. An annual notice doesn't go out, and the only reason nothing happens is luck. And the whole year's worth of things you figured out gets stored in the only place with no backup, which is your memory.

None of those show up as a line on a bill. They show up as a renewal you can't explain and a crew that doesn't trust the plan.

Founder note: Ask yourself who benefits from an owner who only pays attention during the last sixty days. Not the villains — just look at the structure. The renewal quarter is the one stretch of the year when you have the least time, the least information, and the most pressure to sign. That is precisely when you're asked to make the only decision anybody brings you. I've sat on both sides of that table for thirty years. Not once did I see a calendar handed to an owner that said "here are the four times a year we should talk, and three of them are not about buying anything." Nobody's hiding it. Nobody's offering it either. Silence does the work.

My plan year doesn't even start in January — so what are my four quarters, and which one am I standing in right now?

Meridian move: Open your calendar and your plan document. Find your plan year start date in writing, not from memory. Then set four recurring ninety-minute appointments, one per quarter, counted from that date, and name them: TALK, PLUMBING, FACTS, DECISIONS. Put one person's name on each. That's it — four calendar entries and a date you confirmed. It commits you to nothing about the plan itself, no change, no quote, no purchase, and it's the only scaffolding the rest of this Car needs.

Act II · about 12 minutes

The Few Numbers You Watch All Year

Rider premise: You do not need a dashboard, a platform, or a report somebody builds for you. You need about six numbers, written on one page, updated four times a year, in your own handwriting. The reason you don't have them isn't difficulty. It's that nobody ever told you which six.

Here's the test for whether a number belongs on the page. If it changes what you would do, it stays. If it's just interesting, it goes. Most benefits reporting fails that test on every line.

Let me walk the six.

Number one: how many people are on the plan, and how many said no. Three counts, not one. How many employees are eligible. How many are enrolled. How many were eligible and declined — that's your waiver count, and it's the number almost nobody writes down.

Nationally, at companies that offer health benefits, about 80 percent of workers are eligible. About 76 percent of those eligible take it. In the end, 61 percent of all workers at offering companies are enrolled [8]. Those are big averages across every size of company, so don't use them as a grade. Use them as a ruler. If half your eligible crew is waiving, something specific is happening in your building, and it's knowable.

What that number tells you: whether the money you spend is actually reaching your people. What it does not tell you: why. The why is number six.

Number two: cost per covered employee, per month. Take everything the company pays toward the plan in a month. Not the plan's total cost — the part you pay. Divide that by the number of covered employees. One number. One page.

For scale. In 2025 the average total premium for employer health coverage was $9,325 a year for one person and $26,993 for a family [8]. Workers paid an average of $6,850 of that family premium out of their own paychecks [9]. Looking forward, one large employer survey has health benefit cost on pace to rise 6.7 percent in 2026, pushing average cost above $18,500 per employee [10].

Again — ruler, not verdict. What makes your own number powerful is that you can watch it move. An owner who knows their cost per covered employee went from $612 to $661 has a fact. An owner who only knows "we got hit with nine percent" has a feeling.

Number three: your contribution rule, as an actual percentage per person. Not what you believe it is. What payroll says it is.

Most owners will tell you "we pay eighty percent of the employee and half of family." Then you pull one pay period and run it person by person. It's 91 percent for the newest hire on the cheap plan. It's 68 percent for the manager on the rich plan with three kids. It's 44 percent for somebody who added a spouse two years ago, and nobody recalculated anything. That spread wasn't chosen by anyone. It's what Car 01 called drift, and this is the number that catches it.

While you're in there, one arithmetic check worth doing at any size. Take your lowest-cost employee-only plan, look at what an employee pays per month for it, and compare it to that employee's pay. For plan years starting in 2026, the affordability line under the federal rules sits at 9.96 percent of income [11]. Below 50 full-time-equivalent employees, that rule doesn't reach you. The arithmetic still tells you something true. It tells you whether your cheapest plan is a real option for your lowest-paid person, or a decoration.

Number four: turnover among covered employees. Not company turnover. Turnover among the people on the plan, which is a different and more expensive number.

For a ruler. In July 2026 the national quits rate was 1.9 percent of employment in a single month. The total separations rate — quits, layoffs, everything — was 3.2 percent [12]. So in a typical month, roughly three of every hundred workers in the country leave a job. In your building, each one of those is a coverage end date, a final bill, a continuation notice, and a beneficiary form that goes stale.

Here's what makes this a benefits number rather than an HR number. If the people leaving are disproportionately the ones on your plan, you're paying for coverage as an exit ramp rather than a reason to stay. If the people leaving are disproportionately the ones who waived, you may be watching the cost of a plan they couldn't afford to join.

Number five: how many people hit the out-of-pocket maximum. The out-of-pocket maximum is the ceiling. Once a person has spent that much on covered care in a year, the plan pays the rest.

This is the number that tells you where your plan design actually landed. For 2026, the federal ceiling on what a health plan can make someone pay is $10,600 for one person and $21,200 for a family [13]. If your plan is the high-deductible kind that pairs with a health savings account, that ceiling is lower — $8,500 and $17,000 for 2026 [14]. Those are the legal walls, not your plan. Your plan's number is on your own summary.

Why count the people who hit it? Because "we have a five-thousand-dollar deductible" is abstract until you know that four families reached the ceiling last year. Four families reaching the ceiling means four households absorbed the maximum your design allows. That's not a complaint about the design. It's the design working exactly as written, and you should know how often that happens before you turn the dial again.

Number six: who declined, and why. One line, asked at enrollment, in plain words. On a spouse's plan. On a parent's plan. Getting help buying coverage on their own. Too expensive. Doesn't see the point.

Those five answers lead to five completely different decisions, and four of them are free. The owner who learns that six people are on a spouse's plan has learned their program is fine. The owner who learns that six people said "too expensive" has learned that their contribution rule is doing something they never voted for.

Now, two rules about how to read any of these.

First: watch a trend, not a month. One month is noise. Somebody had a baby, somebody had surgery, somebody left. A number that moves the same direction three readings in a row is telling you something. A number that jumps once and comes back was weather. The whole reason for quarterly readings instead of monthly panic is that four readings a year make a trend visible and make a single bad month boring.

Second: know what you're not being shown, and why. This is where the sniffer earns its keep.

Car 01 taught that funding decides what you're allowed to see. Here's what that means in practice when you're trying to fill in six numbers.

Nationally, 67 percent of covered workers are in self-funded plans — but only 27 percent at small companies [8]. So most small employers are in a fully insured or level-funded arrangement, and in a fully insured arrangement you generally do not get claims detail. You pay the money. One number comes back at renewal. You cannot check the math, and nobody has to show you the math.

Now the honest part, said both directions, because this is where people either get taken or make fools of themselves.

You are not entitled to individual claims. You should not want them. The moment you can see which employee had which diagnosis, you have a privacy problem that is much worse than any renewal. An owner who demands "my claims data" without knowing what they're asking for is asking for something no responsible party should hand over.

What you can reasonably ask for depends on funding, and it sounds like this: total paid claims by month for the plan as a whole. A count — just a count, no names — of claims over some large threshold. Enrollment by tier, meaning how many single, how many with a spouse, how many with family. A monthly invoice that can be lined up against your own employee list. And loss ratio or claims-versus-premium, if the arrangement produces one.

Ask about the arrangement, not about your rights. It sounds like this. "Which of these reports can this funding type actually produce, and what does it cost me to get them?" That question is polite, specific, and impossible to wave away. If the answer is "your plan doesn't generate that," fine. Now you know a real fact about the arrangement you bought. You can decide whether you want to keep buying it.

And there is one thing you are flatly entitled to, which almost no small employer ever asks for. Say someone provides brokerage or consulting services to your group health plan. If they reasonably expect to be paid $1,000 or more, direct or indirect, they have to put that pay in writing for the plan's responsible fiduciary. And they generally have to do it reasonably in advance of signing or renewing [15]. Direct and indirect. Meaning commissions, and the bonuses you never see.

That's not a gotcha. It's a document that's supposed to exist. Most owners have never seen theirs, and most have never asked.

Founder note: I want you to sit with the gap between those two paragraphs. You are not allowed to see what your own plan paid out — on a fully insured plan, the thing your money bought is simply not shown to you. But you are allowed to see exactly what the people around your plan get paid, and that one's sitting in a file nobody opens. Read that again. The industry spent years teaching owners to feel entitled to the data they can't have and never mentioned the disclosure they can. I'm not accusing anybody of a crime. I'm telling you which door is unlocked, because in thirty years I've watched almost nobody try the handle.

If half of these six numbers have to come out of my own payroll report instead of anybody's system — what have I been waiting for?

Meridian move: One sheet of paper, six lines: enrolled, eligible, waived; cost per covered employee per month; your contribution percentage for three specific people; covered-employee turnover for the last twelve months; how many people hit the out-of-pocket maximum; and the count of declines by reason. Fill in what payroll already knows today and leave the rest blank. The blanks tell you which numbers your current arrangement won't produce, which is itself one of the six answers. This commits you to no change, no quote, and no conversation with anyone outside your building.

Act III · about 12 minutes

When Something Breaks Mid-Year

Rider premise: Almost nothing that happens mid-year is unusual. Hours change, families change, people leave, claims land. The damage almost never comes from the event. It comes from the two weeks after the event, when everybody assumed somebody else was handling it.

So let's sort the ordinary events into three piles. The ones with a clock on them. The ones that look terrifying and aren't. And the ones that look like nothing and quietly cost the most.

Start with the clock pile. These are the ones to handle the same week they happen.

An employee's hours change. In any business with a busy season, this is the most common mid-year event. It's also the one owners handle worst. The obvious move is to leave them on the plan, because they're a good person. That move has consequences nobody explains.

Two separate rules are in the room. First, your own plan's eligibility rule. It says something like thirty hours a week. If somebody drops under it, the plan says they lose coverage, whatever you personally feel about it.

Second, the federal rules, if you're at or above 50 full-time-equivalent employees. There, a full-time employee is someone averaging at least 30 hours a week, or 130 hours a month. You get two ways to measure. Month by month, or look back over a longer stretch and lock in the answer for a period that follows [16]. Which method you use is a choice you make in advance. It is not something to invent in August about one person.

Here's what to do the same week. Write down the date the hours changed. Find the eligibility rule in your own document and read it. Then either the person keeps coverage under the rule as written or they don't — and if they don't, a clock just started.

That clock is continuation coverage. Car 01 named the law: the Consolidated Omnibus Budget Reconciliation Act, generally at 20 or more employees. The timing is what matters mid-year. The employer has to notify the plan administrator within 30 days of a termination or hours reduction. The administrator then has 14 days to notify the person of their rights. The person gets at least 60 days to elect, counted from the later of the event or the notice. They can't be made to pay for that coverage until 45 days after they elect [17]. If you are your own plan administrator, that whole front end collapses into one 44-day window that belongs entirely to you [18].

Thirty days. That's the number to remember. A termination you got around to reporting in October for a September separation is how a small administrative slip turns into a real exposure.

A baby, a marriage, a divorce, a kid aging off. These are the events employees actually walk into your office about, usually late, usually apologetic.

Two rules again, and they are not the same rule. The first is a mandatory enrollment right. Someone has a baby, gets married, adopts, or loses other coverage — they must be allowed to enroll. They have to ask within 30 days of the event. If the trigger is losing coverage under a state children's health program or Medicaid, the window is 60 days [19].

The second rule is about money coming out of a paycheck. If your plan lets people pay their share before taxes, the election is generally locked for the year. It can only change if the event fits a permitted change, your plan document allows that change, and the new election matches the event [20].

Plain version: the enrollment right and the pre-tax election are two different doors, and people assume opening one opens the other.

Divorce deserves its own sentence, because it's the one that goes wrong most often. The employee has to tell the plan about a divorce, a legal separation, or a child losing dependent status. Not you. They generally have 60 days to do it [18]. When nobody tells anybody, an ex-spouse stays on your plan and uses it. Now you have a claim paid for someone who wasn't eligible, and a continuation notice that never went out. That is a genuinely ugly cleanup, and it started with a conversation nobody had.

A key employee leaves. Same clock as the hours drop, plus two extras. The retirement plan has a balance and a beneficiary form on it, and that form pays the person named on it — not the person in the will. And if that employee ran benefits day to day, you just lost your records. Find out this week whether anything lived only in their head or their email.

Now the second pile: the events that look terrifying and mostly aren't.

A big claim lands. Someone on your plan has a serious year. A surgery, a long hospital stay, a diagnosis with a long tail. Owners panic here, and a lot of the panic is misdirected.

If you're fully insured, the immediate answer is: nothing about this month changes. You pay premium. The insurance company took that risk, which is the entire thing you bought. Yes, a big year can show up later in how they price your renewal. But it is not a bill landing on your desk, and it is not something to solve this week.

If you're level-funded or self-funded, a large claim is exactly what your stop-loss coverage exists for. The correct move is to confirm the claim got reported the way your agreement requires. That's a paperwork step, not a crisis.

Either way: one claim is not a trend. Do not redesign your plan in July around one family's hard year. Note it, watch it with the numbers from Act II, and let three readings tell you whether anything changed.

One bad month. Claims are lumpy. A month where the plan paid out double doesn't mean much on its own, and month-by-month reaction is how owners talk themselves into changes they later regret.

Now the third pile. Small-looking, expensive.

A bill that doesn't match your employee list. This is the single most common quiet leak in a small company's benefits year, and it fails in both directions. You pay premium for people who left, sometimes for months. When you catch it, what you get back is usually less and later than you think. There are limits on how far back a carrier will unwind coverage. Or the reverse: someone thinks they're covered and isn't, because an enrollment never got transmitted, and they discover it at a pharmacy counter with a child in the car.

This is not a hard problem. It's a fifteen-minute problem, done monthly, that nobody owns. Line up the bill against your actual payroll list. Every name on one should be on the other.

Somebody was let on the plan early as a favor. A new manager starts Monday and you don't want them uncovered for sixty days, so they go on right away. It's a kind thing to do. It also means the written plan and the real plan are now two different plans, which is a pattern Car 01 named.

Here's what to actually do, because "don't do that" is not useful once it's done. Don't hide it, and don't quietly hope. Write down who, when, and what you did. Then pick a direction: either the document changes so the practice is legal, or the practice changes so the document is true. What you cannot do is leave the gap open and undocumented. It gets discovered on the worst possible day: a large claim, an audit, someone's departure, or a buyer's due diligence when you sell.

You buy or lose a location. This one hides a real trap. Employee counts can be combined across companies under common ownership for the 50-full-time-equivalent test [21]. An owner with two companies of 30 people each may be sitting at 60 for federal purposes and not know it. Buying a location isn't just new people and new enrollment dates; it can change which rules apply to you.

And a mid-year plan change is not just a phone call. Reduce what the plan covers mid-year and notice rules attach. That includes the sixty-days-ahead requirement, when the change makes the summary you handed out wrong [7]. Decide in April, effective in July, is a real sentence. Decide in April, effective in May, often isn't.

So how do you tell a real problem from a scary-looking one? Three questions, in order. Does a clock start? Does it change who is or isn't eligible? Or is it just a number that moved? Clock and eligibility go to the top of this week's list. A number that moved goes on the quarterly page and waits.

Last thing, and it's the cheapest habit in this whole Car. Keep a log. One running page, one line per event: the date, who it involved, what happened, what you did, what date the change takes effect, and who you told.

Five pieces of information, thirty seconds, while you still remember. The reason is brutal and simple: you will not remember. Renewal is nine months away, the person who told you is on vacation, and the email is buried under four hundred others. A log is how the whole year arrives at the meeting in the next act as facts instead of impressions.

Founder note: Here's the thing nobody says out loud about mid-year. Every delay in that list costs the owner and costs nobody else. You eat the premium for the guy who left in June. You eat the claim for the ex-spouse who stayed on. You eat the hour of apology to the employee who found out at the pharmacy. Meanwhile, a bill that's wrong in your favor gets corrected fast. A bill that's wrong against you gets corrected when you notice. And "we'll credit you next month" is a sentence I've watched go unkept more times than I can count. Nothing in the structure is set up to catch this for you. The fifteen minutes a month is on you because the savings are on you.

Half of these have happened in my business in the last year and nobody wrote any of them down — so what's actually still open right now?

Meridian move: Start the log today with the last three things you remember: a person whose hours changed, a person who left, or a family change somebody mentioned. For each one write the date, what happened, what you did, and what you're not sure about. Then do one pass of this month's bill against your payroll list. That's a page and fifteen minutes. It does not commit you to fixing anything, changing the plan, or telling anyone outside your building. It just moves the year out of your memory and onto paper, where it can be worked.

Act IV · about 12 minutes

The Meeting That Actually Moves It

Rider premise: Everything in the first three acts fails without one thing. A recurring meeting, short enough that you'll actually hold it, structured enough that something gets decided in it. Not a strategy session. Not a presentation. Forty-five minutes, four times a year, with a page.

Let me describe the meeting exactly, because vague advice about "having a process" is how owners end up with a binder and no habit.

Who is in it. Three seats, four people at the most.

The first seat is the person who can say yes. In a company your size that's you. If the owner isn't in the room, it isn't this meeting — it's a status update that will be repeated to the owner later, badly, and decided in a hallway.

The second seat is the money seat. A controller, a bookkeeper, your payroll person. Somebody who knows what the deductions actually are and what payroll can and cannot execute. This seat exists because half of all benefits decisions die on the way to payroll, and nobody finds out for two months.

The third seat is the person who runs it day to day. An office manager, an operations lead, whoever the employees actually walk up to with a question about their card. This is the most undervalued seat in small business benefits. That person knows which plan people complain about, who asked about a baby, and which bill looked strange in March. They are almost never invited to the meeting where it gets discussed.

Four people, maximum. A meeting with nine people in it is a briefing, and briefings do not produce decisions.

How long it is. Forty-five minutes. Quarterly. Same week of the quarter, counted from your plan year — those are the four appointments from Act I, and this is what happens inside them. If it's scheduled for three hours, nobody will hold it a second time.

What's on the page. One page, and it's the same page every quarter. That sameness is the whole trick, because it turns reading into comparing.

Top third: the six numbers from Act II, with last quarter's value sitting next to this quarter's. Not a chart. Two columns of numbers a person can read in ninety seconds and say "that one moved."

Middle third: the log from Act III. Every event since last time. Hours changes, departures, family changes, the bill that didn't match, the favor you did for the new manager.

Bottom third: anything genuinely time-bound in the next ninety days — a notice, a filing, an enrollment window, a plan year date — plus open decisions carried over from last quarter.

That's it. One page, three blocks. Whoever holds the pen prepares it in twenty minutes from payroll and the log. No vendor builds it. No platform is required. Nobody outside the building needs to be in the loop for this to happen.

What a decision looks like when it's actually made. This is the part that separates a real meeting from theater, so let me be precise.

A decision has four parts. A verb — what specifically will be different. A name — the one person doing it. A date — when it's done by. And a number, when there is one.

"We should look at the family contribution" is not a decision. "We're moving the family contribution from 50 percent to 60 percent, effective the first day of the next plan year, I decided it today, and payroll runs the new numbers by the fifteenth" is a decision. Same topic. One of them exists in the world.

And there are only three legal outcomes for any item on that page. Decide it. Defer it to a specific named date. Or kill it, and say why. That's all. A "defer" without a date is not a defer — it's drift wearing a suit, and it will be on the page again next quarter looking exactly the same.

How it gets written down. One page per meeting. What was decided, with the four parts. What was deferred, and until when. What was killed, and the reason. Two paragraphs is plenty. It goes in a folder — a real one or a digital one — with a name on it, in a place that survives the departure of whoever wrote it.

The "why we killed it" line is the one people skip and the one that pays. Next year, somebody will raise the same idea with the same enthusiasm, and you will have no memory of why you passed on it. Three words on a page ends a forty-minute conversation.

There's also a floor under all of this. Records supporting required plan filings generally have to be kept and available for at least six years [22]. Your quarterly page isn't a filing, and I won't pretend it is. But you're keeping a six-year drawer anyway. Putting the decision page in it costs nothing. Being able to answer "why is our eligibility written this way?" three years later is worth a lot.

What this buys you. By the fourth meeting of the year, your renewal prep is already done. The facts are current because you read them in the third quarter. The events are documented because somebody logged them. The open decisions are listed because they've been carried forward with dates on them. Car 202 taught you to read your renewal; this is how you arrive at it holding something instead of nothing.

And next year doesn't start from zero. It starts from four pages.

Now the sniffer bite, because there's a meeting that looks exactly like this one and produces nothing, and you have almost certainly sat in it.

You know the signs. No page, or a page somebody else prepared that you've never seen before. A presentation. Nice slides. Thirty-five minutes of context and a recommendation arriving in the last five, when there's no time to check anything. Lots of nodding. Everyone leaves feeling productive. Nothing has a name or a date on it. Six weeks later, nobody can say what was decided.

Look at who's around that table, using the three kinds of people Car 01 named.

The Predator runs the clock on purpose. The information comes late, the decision comes at the end, and the urgency is manufactured. They know that an owner who hasn't seen the numbers in advance will defer to the person who has.

The Pretender says "don't worry, we handle all that," and calls your specific question about compensation a trust issue. Watch for the tell: they answer a question you didn't ask, warmly, and you leave the room liking them and knowing nothing new.

And the Professional — again, the biggest group by far — is competent, well-intentioned, and paid on a structure where a transaction at renewal is the only event that produces income. Nothing in their pay says "call the owner in month four about a bill that doesn't match the census." That is not a character flaw. It's a paycheck. And it's exactly why the quarterly cadence has to belong to you rather than to anybody whose income depends on a signature.

So here's the money question to ask about your own meeting, and it isn't rude. Who in this room gets paid when we make a decision, and who gets paid the same either way if we don't? Then, once a year, ask for the written compensation disclosure. That's the one anyone providing brokerage or consulting to your health plan owes the plan's responsible fiduciary when they expect $1,000 or more [15]. Put it on the page as a line item in the same quarter every year. You are not accusing anyone. You are reading a document that is supposed to be handed to you.

Now let me be honest about what this Car doesn't do.

Car 303 does not look at your plan. It doesn't price anything, recommend anything, or tell you what to do at your next renewal. It teaches the rhythm — four quarters, six numbers, a log, and a meeting. That's the whole inventory. Run those four things for a year. You will know more about the second-biggest cost in your company than you learned in the last five years of renewals. And you won't have bought a single thing to do it.

And you're done with this Car when you can answer five questions without notes. What's my plan year start date, in writing? What are my four quarterly appointments and who owns each one? What are my six numbers and which ones can my current arrangement actually produce? What are the three mid-year events that start a clock? And what does a decision have on it before it counts as made?

Five answers. No math, no upload, nothing to buy.

Here's where the ride points next. Car 404 is meant to be the whole season — benefits as one system running alongside the rest of your money, instead of a subject you visit. I'm not going to tell you it's written, because I don't know that it is on the day you're reading this. That's the deal on this train: nobody promises you a Car that isn't there. What I can tell you is that 404 is built on the assumption that you already run a year. If you haven't run one, it has nothing to sit on.

Which means the honest next step isn't another Car. It's one quarter.

Founder note: I've watched owners buy software, hire consultants, and sit through more presentations than any human should. All of it to avoid one forty-five minute meeting, four times a year, with three people who already work there. I understand the impulse. Buying something feels like progress and costs money you've already budgeted. The meeting feels like nothing and costs the one thing you actually can't get more of. But here's what thirty years taught me: the owners who stopped getting surprised weren't the ones who bought the most. They were the ones who wrote one page a quarter and kept the pages. That's it. That's the whole difference, and nobody sells it because there's nothing in it to sell.

Four short meetings, one page each, all year — is that really the thing I've been outsourcing to a panic every October?

Meridian move: Take the four appointments you made in Act I and add three things to each one: the names of the three people in the room, a forty-five minute length, and the words "one page." Then build the page for the next one — six numbers on top, your log in the middle, the next ninety days at the bottom — even if half of it is blank. Hold that one meeting. It commits you to no plan change, no quote, no purchase, and no conversation with anybody outside your business. It only commits you to writing down what you decided, which is the one thing nobody else will ever do for you.