Which Genius Designed the Reimbursement Process for Tuition Benefits?

I want to be fair to whoever designed this, because forty years ago they were right.

A college degree was rare then. Completion rates were higher. An employer who helped a long-tenured employee finish a credential could reasonably expect to keep that person for a decade. Reimbursement fit that world. The company protected itself against paying for someone who quit or dropped out, and the employee got something they could not otherwise have afforded. It was a sensible piece of design for the problem in front of it.

That world is gone. The benefit stayed exactly where it was.

For a long time I thought that was the whole story. A good design, left too long in the drawer. I no longer think so, and this is the part I want to put plainly.

The question in the title has an answer, and it is not a flattering one

Nobody set out to build an obstacle course. But somebody, at some point, was handed this benefit and told to keep the number down.

That is the thing worth sitting with. A tuition benefit does not live in the part of a company that thinks about growth. It lives on the expense side of the ledger, next to the other things finance is asked to contain. And when you hand a line item to a person whose job is to control it, you do not get a program designed for maximum use. You get a program designed for defensible use.

Low participation is not the failure of that design. It is the design working.

I do not think anyone was cynical about it. Every gate in this process was added by a reasonable person solving a reasonable problem. Pay first, so we do not fund a dropout. Cap it, so the exposure is known. Require job-relatedness, so we are not subsidizing somebody's exit. Require a grade, so we are paying for effort rather than enrollment. Require them to stay, so we see something back.

Read that list again. Every one of those conditions makes sense as cost control. Not one of them was written by somebody asking how to get more of their people capable of doing something they could not do the year before.

So we have spent forty years perfecting a process that works exactly as well as it was built to work. Hardly anyone ever starts it. That was never the accident.

Ninety-two percent say yes. Five percent ever start.

Ninety-two percent of US organizations offer some form of educational benefit. Fifty-seven percent offer tuition reimbursement specifically, down from sixty-three percent in 2019. On paper, that is an open door.

Now look at who walks through it. Seventy percent of organizations that offer tuition benefits see participation at five percent or below. One in five sees it under one percent.

I have spent more than twenty years on the school side of this, and I have watched the distance between those two sets of numbers play out in a hundred different rooms. An employer says they offer tuition assistance. A school says that is wonderful. Everybody agrees there is an opportunity here. Cards are exchanged. And then nothing happens, quarter after quarter, and nobody can quite say why.

The usual explanation is that employees do not know the benefit exists. I have never found that convincing, because I have met them. They know. A mother working nights on a hospital floor, three classes short of the credential that would change what she earns for the rest of her career, knows exactly what her benefits summary says. What stops her is not awareness. It is that she would have to find four thousand dollars by the fifteenth of the month, upfront, to pay the school before anybody reimburses her a cent.

And the direction of travel is getting worse. SHRM recorded forty-six percent of employers offering tuition assistance in 2024, the lowest prevalence it has ever measured. Employers are quietly backing away from this, and inside the logic of a cost line I understand why. If you funded something for a decade and five percent of your people touched it, you would start to wonder too.

Here is the trap in that, and it is worth naming because almost every employer I talk to is standing in it. The benefit looks like a poor investment because it is built in a way that guarantees hardly anybody uses it. The low usage then becomes the argument for cutting it. The design manufactures the evidence for its own removal.

And notice what a trap like that does not require. It does not have to turn anybody away. Almost nobody is rejected by a tuition benefit. There is no committee, no denial letter, no moment when somebody is told no. The filtering happens earlier and far more quietly than that, and it happens to people who never fill anything in.

Which is the part I would rather show you than assert, because saying that a design discourages people is easy to write and easy to wave away. So let us read it the way she reads it. Not the way a benefits policy is written, but the way it looks at nine at night on a phone, in a hospital car park, shift finished, children asleep, portal open.

She is not weighing one obstacle. She is reading a list, and she is adding up what it would cost her before she gets to the bottom of the page.

It is not one barrier. It is a sequence, and each step is defensible

That is what makes this so hard to argue with, and so easy to miss. Walk it in order, the way she has to.

• Pay first. The employee fronts the tuition, out of pocket or on a loan, before the employer contributes anything. For somebody on median wages, one semester can equal a month of take-home pay.

• Then accept a cap. Seventy-one percent of organizations set a dollar limit. The federal exclusion under Section 127 has sat at $5,250 since 1986. Forty-one years without adjustment. Indexing finally begins in 2027, and the figure for that year has not been published yet.

• Then prove it is job-related. Forty-five percent require the course of study to relate to the employee's current role. A warehouse worker who wants to become a nurse does not qualify. The benefit funds the job you already have, not the one you are trying to reach.

• Then earn a grade. In 2019, sixty-nine percent of organizations required a minimum grade before they would reimburse. By that point the employee has paid, studied and finished, and still has not been made whole.

• Then stay. Forty-one percent impose a service period. In 2019, fifty-seven percent required payback if the employee left.

That is the offer on the page. She pays for the credential, waits two terms to be repaid for it, and is then quietly discouraged from using it to go anywhere.

Any one of those gates is survivable. In sequence they are a filter, and the filter does not select for motivation or ability. It selects for people who can afford to float a semester and absorb the risk of owing it back. Which is to say it excludes, almost perfectly, the person the benefit was written for.

And it does the excluding at the top of the page. Most people never reach the second gate, because they can see it from the first. She closes the portal. Nothing is recorded anywhere. Next year somebody reports that participation came in under five percent again, and the budget conversation starts from there.

What it costs to keep calling this an expense

Here is where I would like to change the terms of the conversation, because the expense framing is not only unkind. It is bad arithmetic.

Two employers actually measured this, and the results have been sitting in public for a decade. Cigna studied its own tuition program and found a 129 percent return: every dollar in came back with twenty-nine cents on top, after the cost of running the program, and participants' wages rose 43 percent faster than non-participants'. Discover ran the same exercise and found 144 percent, with $10.9 million in avoided talent costs. Both studies were funded by the Lumina Foundation, both look at a single employer, and both are now more than ten years old. I am not going to pretend they are a market benchmark. But they are the most serious attempt anybody has made to put a number on this, and the number was not near break-even. It was a return.

Now the part that should bother you. Cigna's participation rate, across roughly thirty-seven thousand US employees, was 5.8 percent.

A company measured a 129 percent return on its tuition benefit and still could not get more than six people in a hundred to the starting line. The program was not the problem. The paperwork standing in front of it was.

That is the argument in a single data point. The return is real, and the process is throwing most of it away before it can be earned.

Meanwhile the money is moving the wrong way. Employer training spend fell from $1,254 per employee to $846 in one year, a drop of roughly a third, while the average hours employees logged in learning rose from 13.7 to 16.7. More being asked of people, a third less behind them. That gap does not close on its own, and it does not close with another software license.

The cost of leaving it open shows up in somebody else's data. Work Institute has run exit interviews at scale since 2013, and career has been the single largest category of reasons people give for leaving, every year since they started counting. Their most recent quarterly report puts it at 19.6 percent of the reasons cited. Inside that category, alongside career paths and promotion and development, they name one thing that should stop any HR leader cold.

School. People are walking out of jobs in order to go and study, while holding a benefit that would have paid for it if only they could have floated the first semester.

Think about what that actually costs. You post the role again. You pay to recruit, you pay to onboard, you absorb the months before the replacement is useful, and you lose the person who already knew how your organization works. All of it to replace somebody who left to get an education you had already agreed to fund. There is no honest version of that which belongs on the expense side of a ledger.

We have known the return is real. Proving it has been the problem.

Those two studies are more than ten years old and nobody has replaced them. Not because the question stopped mattering. Because proving it is genuinely hard, for a reason that has nothing to do with effort.

A return has two halves, and employers only hold one of them. The return half they already track. Who stayed, who was promoted, who got filled from the inside instead of hired from outside. That sits in the HR system and it is reasonably good data. The investment half is where it falls apart. A tuition line in a benefits budget is a lump sum and a stack of receipts. It tells a CFO what left the building. It cannot tell them who was funded, at which school, studying toward what kind of work. You cannot run the equation with one side missing, so almost nobody runs it, and the benefit stays an expense by default. Not because anyone decided it was one. Because nothing else was available to call it.

That is the half EFFASkills closes. It takes funded spend and makes it countable. Every employee funded, every program, every school, every dollar settled, and the occupations those programs prepare people for along with the workforce skills those roles call for. A tuition budget stops being a lump sum and becomes a picture of what the money bought, in the same shape as the data the employer already keeps on who stayed and who moved up.

It does not hand anyone a finished return figure, and I would be careful with any vendor whose product claims to, because the outcome side of that equation lives in your HR system and not in ours. What it does is give an employer the half of the equation that has been missing for forty years, in a form they can set beside their own numbers. That is how a benefit stops being a cost nobody can defend and starts being an investment somebody can argue for in a budget meeting.

And then the rest of it exists to get people through

None of which matters if hardly anyone starts. So everything else we built works backward from her, the woman in the car park at nine at night, and the first thing it removes is the first gate.

Direct bill means the employer pays the school. The employee never advances a dollar, never files a claim, never waits on a grade to be made whole, and never signs anything that turns their own credential into a liability if they move. The sequence does not get easier. It stops existing.

This is not an exotic idea. It is what Section 127 actually describes: an employer educational assistance program, in which the employer provides the assistance. The statute never said the employee should buy their own education and then apply for a refund. We read reimbursement into a provision that never required it, and built four decades of process on top of the misreading.

Schools feel the same problem from the other side. It arrives as students who enroll, then stop, and never explain why. It arrives as an employer partnership that looked promising in year one and produced almost nobody by year two.

So underneath both sides there is one process, one language, and a single record the employer and the school can both see as it happens, rather than two systems and an email thread between them. Our fee on settlement is 3.5 percent standard.

The question worth asking

I do not believe anyone meant to build a benefit that almost nobody could use. I think a reasonable design for 1986 was handed to people who were told to manage it as a cost, and every year since, somebody added one more sensible-looking condition to it.

Ninety-two percent of employers already say they will pay for this. The money is not the constraint. The paperwork is.

Reimbursement was the right answer for forty years. It is not the right answer for the next forty. The question is no longer whether employers will fund education. It is whether the money ever reaches the woman standing three classes short.