The Invoice

Epistemic status: The numbers are national averages from the industry’s own reports, current and sourced. The ratio at the center of this note is my arithmetic on those numbers, laid out so you can check it. Section 06 is the strongest case against the whole thing, written as well as I can write it.

Every spring a nurse recruiting firm called NSI puts out a report on what it costs hospitals when nurses quit. This year’s covers 527 hospitals in 40 states, and the number I go to first is the price of one resignation. When a bedside nurse leaves, replacing her costs her hospital 60,090 dollars on average. The average hospital in that report pays it about every four days, which comes to 5.19 million dollars a year.

In April, in the manifesto, I used the number from two editions back. It was 56,300 dollars then. I also wrote that most healthcare leaders can’t tell you what turnover costs them every month, that they don’t have the number. That part hasn’t changed. So this note is the number. One nurse, one line at a time, and then the ratio it adds up to, which is the part that should change how we run our companies.

01

The Bill Nobody Opens

At HealthView we were told that culture and taking care of people were too expensive for a home health company. We were told nurses couldn’t be business people or lead sales.

A board member once asked me to prove that culture mattered financially. So I pulled what turnover had cost us over the past twelve months, how long critical jobs sat open, and our revenue per employee against the benchmarks. When I put those numbers in front of the board, the debate changed. We stopped asking whether culture was worth funding and started asking which managers created the most turnover risk, and what that cost us each quarter.

Turnover is how a care company pays for its culture. You can’t put culture on a balance sheet, and you can’t audit a value. But every month, the way a company actually treats its people gets billed back to it in resignations.

The bill is itemized in places nobody adds up. Recruiting sits in the HR budget. Overtime sits in nursing’s. Agency nurses sit in a contract somebody signed in a hurry. Orientation sits with education. The referrals you turned away because nobody was free to see the patient sit in nobody’s budget at all, because money you never collected never shows up anywhere.

So no one person ever holds the whole bill, and a five million dollar line item gets paid every year without anybody signing for it.

02

1913

The first operator I know of who actually opened the bill ran a car factory.

In 1913 Ford’s plant at Highland Park had just put the moving assembly line to work, and the people on it kept walking out. The company hired 50,448 people that year to keep an average workforce of 13,623. That’s a turnover rate of 370 percent. On a normal day about one man in ten didn’t show up. The pay was 2.34 dollars for a nine-hour day.

On January 5, 1914, Ford announced five dollars for an eight-hour day, starting the next week. More than ten thousand people crowded the plant looking for work. Turnover fell to 54 percent in 1914 and 16 percent in 1915. Absenteeism went from ten percent to two and a half.

Ford Motor Company, annual turnover

1913before the five-dollar day370%

1914the year it started54%

191516%

January 5, 1914: from $2.34 for a nine-hour day to $5 for eight hours. Daily absenteeism went from 10 percent to 2.5.

Ford later called the five-dollar day one of the finest cost-cutting moves the company ever made, and he meant cost-cutting literally. In his arithmetic, hiring fifty thousand people a year to keep thirteen thousand cost more than doubling the pay.

There’s a second half to the story. More than half of the five dollars wasn’t wages. It was a profit-sharing bonus, paid only to workers whose home lives passed inspection. Ford built a Sociological Department, and its investigators went into workers’ homes to check on their drinking, their savings, how clean they kept the house, and whether they took in boarders. The wage stuck, and the inspectors didn’t. By 1921 the department was largely gone.

Keep both halves. Pay is part of what keeps people, and the first half of Ford’s deal shows it. But what made the deal work was a shorter day and enough money to build a life on. The part that tried to buy behavior by watching people didn’t last.

03

One Nurse, Itemized

When a bedside nurse resigns, the bill comes due in a set order. These are national averages from this year’s NSI report. It covers hospitals, which have the best data. My side of healthcare comes in section 05.

First, her shifts. A resignation doesn’t take the shifts with it. Somebody covers them: a colleague on overtime, a float nurse, or an agency nurse.

Then the vacancy. It takes an average of 78 days to recruit an experienced RN. In telemetry it takes 87.

Then the cover. A travel nurse costs 91.23 dollars an hour. A staff nurse, with benefits, costs 59.46 dollars. Covering a 36-hour week with a traveler for those 78 days runs about 12,700 dollars more than the nurse who left, and that’s one line.

Then the replacement. Recruiting, interviewing, credentialing and orientation, including the weeks a new nurse works alongside a preceptor, which means paying two nurses to do one nurse’s work.

Then the chance of doing it all again. 22.7 percent of newly hired RNs left within their first year. More than one time in five, you pay the whole bill twice.

NSI’s estimate for the whole bill is 60,090 dollars.

INVOICENº 21

One bedside nurse, resigned

---

Vacancy78 days to recruit

Cover+$31.77 an hour for a traveler

Replacementrecruit, credential, orient, precept

Repeat risk22.7% leave in year one

TOTALNSI estimate, national average$60,090

NEVER BILLED

• The next resignation
• The referral you couldn’t take, $2,038 per 30-day period
• The patient who has to start over

Then there are the lines nobody bills. The next resignation is one of them. Turnover spreads. Researchers followed a regional bank’s 45 branches and more than a thousand departments of a national hospitality company, and found that people were more likely to quit when their coworkers were out looking for jobs, even after accounting for their own situation. They named it turnover contagion.

In a care team you can watch the mechanism. The nurse who leaves doesn’t take her patients with her. They get divided among the nurses who stayed, and that’s how one resignation turns into the overtime that drives the next one.

04

The Idiot Index of Care

In The Yield I wrote about the idiot index, a tool from manufacturing. Take what a finished part costs and divide it by what its raw materials cost. A high number doesn’t mean the materials are expensive. It means the process is bad.

Run the same division on a person. On top, what it costs to lose a nurse. On the bottom, what it would have cost to keep her.

The top is 60,090 dollars. The bottom is harder, because most of what keeps a nurse isn’t money. But money is the easiest part to price, so start there. A staff nurse costs her hospital about 123,676 dollars a year with benefits. A five percent raise for her is 6,184 dollars. Sixty thousand over six thousand is about ten.

The idiot index of care, for one nurse

$60,090to lose her: NSI’s average cost of turnover

$6,184to keep her with a five percent raise

≈10

NSI’s estimate 9.7Cut in half 4.9Cut in half again 2.4

On a factory floor a ratio of ten means somebody designed a bad process. Nobody would pay ten dollars to replace a part they could have kept for one.

Raises have a catch that parts don’t, though. You can’t give one only to the nurse who was about to leave. You don’t know who she is, and if you did, you couldn’t pay her more than the nurse next to her doing the same job. So run it for the whole hospital. Work backward from NSI’s own totals and the average hospital has about 490 nurses. A five percent raise for all of them costs about 3 million dollars a year. Each point of turnover is worth 295,000 dollars. To pay for that raise out of turnover savings alone, you’d have to cut turnover by about ten points, from 17.6 percent to under 8.

The raise does more than keep people. It helps you hire, and in a tight market it can keep you off the travel contracts, which this arithmetic leaves out. Where pay is below the market, pay comes first. But as a way to keep the nurses you already have, money by itself is the most expensive tool there is, because you pay it to everyone who was staying anyway.

So the bottom of the ratio has to be built mostly from things that aren’t payroll. Look at why nurses actually leave. NSI’s top five reasons for RN resignations last year were personal issues, relocation, retirement, career advancement and scheduling conflict. AAPACN, the post-acute nursing association, gives a list that cuts closer: a poor relationship with the immediate supervisor, no flexibility, exhaustion from long hours and overtime, feeling undervalued, and demands that outrun the pay.

Most of that list costs hours more than dollars. A manager who knows what’s happening in a nurse’s life before it turns into a resignation letter. A schedule built around the people working it. A caseload that ends before dinner. A company that does what it said it would. That’s what culture looks like itemized, and each of those things costs hours, which is small next to 60,090 dollars.

I run my companies on a plain reading of the idiot index. A high ratio is a process problem, not a people problem. Turnover reads the same way. When a company runs this ratio at ten, the process failed long before the nurse gave notice.

05

Where the Bill Is Biggest

Everything above is hospital data. My side of healthcare pays a bigger bill, with less room to pay it.

Hospital RNs turned over at 17.6 percent last year. Across 1,091 hospices, RN turnover was 25.48 percent and hospice aides were at 29.02. Across 1,111 home health agencies, home care aides were at 34.17 percent. Home health RNs ran about 27 percent the year before. In care at home, a clinician walks out about one and a half to two times as often as a nurse in a hospital.

Annual turnover, by setting

Hospital RNs2025 · NSI17.6%

Hospice RNs2025 · HCS25.48%

Home health RNs2024 · HCSabout 27%

Hospice aides2025 · HCS29.02%

Home care aides2025 · HCS34.17%

And we can’t raise our prices to cover it, because we don’t set them. Medicare does. For 2026 it set the base payment for a 30-day period of home health at 2,038.22 dollars, about one percent less than in 2025. Pay went up last year for the people doing the work, almost five percent for home care aides and three and a half for hospice RNs. The bill grows and the fee doesn’t, so every dollar of turnover comes out of a price that was set in Baltimore.

We also carry a line hospitals don’t. When a home health agency is short a nurse, it can’t take the referral. The patient waits in a hospital bed or goes to whoever said yes, and the 30-day period goes with her, at a base rate of 2,038 dollars. A short-staffed hospital still has the patient in the bed. A short-staffed home health agency never meets her.

Then there’s the line I’d most like to delete. A family that learned to trust one hospice nurse gets a new one near the end, and has to start over. No report puts a price on that, and I’m not going to try.

06

The Strongest Case Against This

Four objections, as strong as I can make them.

One. Some turnover is healthy. A team where nobody ever leaves can be a team where nobody ever gets let go, and a company that chases zero will keep people it shouldn’t. Agreed. The bill I care about is the voluntary one, and in NSI’s data 94.9 percent of hospital separations were people choosing to leave.

Two. Life happens. Read NSI’s reasons again: personal issues, relocation, retirement. No culture keeps a nurse whose family is moving across the country. Also agreed, and that’s the floor. But look at the first year. First-year departures were 29 percent of all RN separations, and retirement doesn’t explain those. Whatever drove them happened inside our hiring and our onboarding, which makes them ours to look at. Scheduling conflict made the top five too, and a schedule is a management decision.

This year we added ride-alongs to our interviews. Before anyone talks about an offer, the candidate spends real time on visits with one of our nurses. Some finish the ride and decide this isn’t for them, and that’s the process working. Better to find out on a Tuesday with us than three months in with a caseload.

Three. Look who’s counting. NSI places nurses for a living, and a firm that sells help with staffing has every reason to price turnover high. The 60,090 dollars is a modeled average, not an audit of anybody’s books. Fair. So cut it in half. A ratio of five is still a process no factory would run. Cut it in half again and it’s still past two.

Four. Maybe it’s just money. Ford fixed his turnover with a raise. Home care aide turnover fell two points the year aide pay rose almost five percent. Maybe culture is a soft word for wages. This is the one I take most seriously. Pay is real, and nothing here argues for paying people less. But the arithmetic in section 04 says money alone is the most expensive way to keep someone, and Ford’s own story says the money worked while the inspectors failed. Pay gets you the chance to keep someone, and after that it comes down to what her ordinary Tuesday looks like.

07

What to Do With It

If you run a care company, the move is simple to say. Compute your own top number: what one resignation costs you, including the referrals you couldn’t take. Put it next to cash, where you’ll see it every week. Then price the bottom, one item on the exit list at a time. Most of those prices turn out to be a manager’s hours.

The invoice says the opposite of what we were told. What we can’t afford is losing them.

My rule is short. I don’t quit on you, and you don’t quit on me. That’s a promise, and keeping a promise is about the cheapest retention there is. Trust is the greatest economic force in the world. With it you need fewer rules. Without it you need more, and every rule is a delay. A lot of my industry is built to be sold in three to five years, and you can’t build trust on that clock.

The only public scoreboard I have on any of this is what our own people say. Our Great Place to Work score went from 95 to 98 percent, and Fortune moved us from 23rd to 16th on its Best Small Workplaces list. That isn’t proof the ratio works. It’s what the people who would pay the bill with their own time say about where they work.

One more thing, because I build software too. There’s a market now for flight-risk scores, models that read your people’s activity and flag who’s about to quit. It’s Ford’s Sociological Department again, with better math. By the time a model can see a nurse leaving, the cheap fix was months ago, and it was probably a schedule.

08

The Ledger

Every field note that makes a prediction puts a dated line on the public ledger. This is the ninth.

09

Ways I’m Wrong

The hospital numbers carry most of this note, and my side of healthcare isn’t a hospital. NSI’s 60,090 dollars is a hospital average. Nobody publishes a comparable invoice for a home health nurse or a hospice aide, and until somebody does, I’m borrowing the closest thing that exists.

The ratio is my arithmetic on public averages. The five percent raise is a round number I picked, the 36-hour week is an assumption, and the 490 nurses are backed out of the report’s own totals. Change any of them and the ratio moves. I’d be surprised to see it fall below three.

And I could be wrong about what’s cheap. If someone shows me careful evidence that schedules, managers and caseloads don’t move turnover and only pay does, then culture really is a soft word for wages, and this note should have been one sentence long.

Next spring NSI will publish a new number, and I’ll put it next to this one.