The honest answer is that the sources disagree, and knowing why is what wins the meeting.

You open your inbox on a Monday. The CFO has sent your leadership development budget back with one comment.

"Benchmark this against industry. Justify the line items. Cut 30%."

That email lands on a VP of People's desk roughly four times a year. Once at planning season. Twice at mid-cycle reforecasts. And once during whatever crisis the company just absorbed.

So you go looking for the number. Fair enough. Let's find it.

‍

So What Is the Number?

Training Magazine: $874 per learner

Training Magazine's 2025 Industry Report, now in its 44th year, is the most-cited survey in the field. It found companies spent $874 per learner in 2025, up from $774 the year before. Total US training expenditure rose 4.9% to $102.8 billion.

Read the methodology before you quote it. The survey covers only US-based organizations with 100 or more employees. And the figure is per learner, not per employee. It counts the people who actually took training, not your headcount. If 120 of your 200 people touched a course last year, those two numbers differ by nearly 40%.

Quote the per-learner figure as a per-employee benchmark and your CFO reads one footnote and takes the meeting from you.

ATD: $846 per employee

ATD's 2026 State of the Industry, reporting 2025 data, puts direct learning expenditure at $846 per employee. That covers content, facilitators, platform, and materials.

It's a steep drop from $1,254 in 2024. Before anyone reads a collapse into that: formal learning hours actually went up over the same period, from 13.7 to 16.7 per employee. Companies bought more hours for less money. ATD's sample is around 340 organizations, so year-over-year swings are also partly compositional — different companies answered.

That detail matters, because a CFO who finds a 33% industry-wide drop will ask why yours didn't fall too. The answer is that the drop reflects falling unit costs and shifting samples, not a market-wide decision that development stopped mattering.

Tech-specific: roughly $1,200 to $2,000, with an asterisk

You want the tech number. It exists, and it is an estimate.

Independent aggregators put technology and software companies at roughly $1,200 to $2,000 per employee, with consulting and professional services higher. But ATD doesn't publish industry or company-size segmentation outside its paywall, so these ranges are assembled from magazine surveys and vendor pricing, and they carry a stated margin of roughly plus or minus 20 percent.

That's not a reason to ignore them. It's a reason not to say "the tech benchmark is $1,850" in a room with a CFO in it.

The stage question

The thing you most want, a credible per-employee figure by funding stage, doesn't exist in public data at all. Anyone handing you "Series B companies spend $X per employee" is giving you an estimate wearing a lab coat. It might even be a reasonable estimate. But your CFO will ask for the source, and there isn't one.

‍

Why the Number Won't Save You

So you have your range. Now the trap.

You walk into the meeting with $874. Your CFO opens a browser and finds $846. Or notices ATD's figure fell 33% year over year and asks why yours didn't. Or finds a vendor survey with a different denominator entirely. You are now debating survey methodology with someone who does spreadsheets for a living. You will lose. And even if you win, you've won the wrong argument, because the amount was never the real question. The real question is whether the money is going to the layer where it compounds. That question you can answer. And unlike the benchmark fight, the evidence is unambiguous and on your side.

One more number from that same Training Magazine report is worth carrying into the meeting. For thirteen consecutive years, management and supervisory training has been the category the highest share of organizations say will receive more funding next year. Thirty percent named it this year, ahead of AI training. The rest of the market has already worked out where the money goes. You aren't asking for something unusual. You're asking for the thing everyone else has prioritized for over a decade.

‍

The Argument That Works: It's Not How Much, It's Where

Three numbers do the work. None of them are about how much you spend.

Number 1. The manager layer is measurably broken

Gallup's State of the Global Workplace 2025 found manager engagement has fallen to 27%. The lowest level in over a decade, down from 30%.

Sit with that. The people you're asking to drive performance, retain talent, and hold culture together are the most disengaged group in the building. They're flying the plane while losing altitude.

That's not an L&D talking point. It's an operational risk, and it belongs in a CFO conversation framed exactly that way.

Number 2. That layer drives most of the outcome

Gallup's research puts managers at roughly 70% of the variance in team engagement.

Not a contributing factor. The dominant one. So when your CFO asks what actually moves retention and productivity, the honest answer is: the quality of the manager, more than anything else you could spend money on.

Engagement isn't a soft metric in this argument. It's the leading indicator on numbers your CFO already tracks. Gallup finds disengaged employees are 18% less productive than engaged ones. Apply that to a fully-loaded engineering team and the cost stops being abstract.

Number 3. Organizations that get the manager layer right make more money

This is the one to put in front of a CFO.

McKinsey found that the EBITDA of organizations in the top quartile of leadership is almost double that of others. They also found organizations are 1.9 times more likely to have above-median financial performance when the leadership team has a shared, meaningful, and engaging vision.

EBITDA. Not engagement scores, not sentiment, not "culture." The number on your CFO's dashboard.

Your CFO will push back on attribution, and they should. The honest answer is that no single program produces that gap. What produces it is sustained, multi-year investment in the quality of the people who run the company day to day. Show them the research. Show them your own engagement data correlated with manager development participation. Show them the trend, not a single-program ROI claim you can't defend.

‍

The cost of doing nothing

The final piece is the one most L&D leaders leave out.

When a manager leaves, replacement runs 50% to 200% of their annual salary, skewing higher the more senior the role. For a manager at $180K, that's $90K to $360K in recruiting, ramp, lost productivity on an orphaned team, and the direct reports who follow them out the door.

If your entire leadership development investment prevents one regrettable manager departure a year, the arithmetic isn't close.

That's the frame. Not "here's what the industry spends." Instead: "here's what happens if we don't."

‍

What We See in the Data

We run a Team Dynamics Assessment across our client base. The current dataset covers 78 organizations and 900+ respondents, gathered between 2020 and 2025.

The pattern in that data is why this argument matters.

Teams score well on psychological safety and caring leadership. Those are consistently the strongest things we measure. People feel safe. They feel supported by their direct supervisor. That's the floor, and most organizations have built it.

What they haven't built is the ceiling. All-Encompassing Vision and Effective Communication score lowest, every time. People are busy without being oriented. The phrase that surfaces over and over is treading water.

Safe, but stuck. And the layer responsible for turning safety into direction is the one Gallup just told us is 27% engaged.

One client, StellarAlgo, shows the mechanism. As they scaled, their Vision scores didn't hold. They fell from 4.42 to 4.13. Vision is entropic. It decays with headcount unless something actively maintains it, and the thing that maintains it is the manager layer.

That's the case for the budget. Not that you're spending an industry-appropriate amount. That there's a specific, measurable gap in a specific layer, and the money is aimed at it.

‍

Where the Money Should Go

Now the allocation. This is a point of view, not a benchmark, and it should be presented that way.

Apply these percentages to whatever your actual budget is. The percentages are the argument, not the dollars.

Roughly 60% goes directly into manager and leadership development. That concentration is the whole point. It's the slice tied to the three numbers above, and it's the slice you defend hardest.

If more than 25% of your current budget is going to learning platforms and generic content, the conversation with your CFO isn't about whether to cut. It's about whether to reallocate. That's a much better conversation to be in, and you should be the one who opens it.

‍

Why Most L&D Budgets Get Cut Wrong

The cut usually lands on the wrong line items, because the budget was built wrong to begin with. Three patterns, and a CFO can see all of them.

The "we already paid for it" platform trap. A company bought a learning platform three years ago, a fraction of the company uses it, and nobody will kill the contract because "we already paid for it." Sunk cost is not a strategy. Audit utilization annually. Low utilization gets renegotiated or cancelled, and you should be the one proposing it.

The "everyone gets training" democracy trap. A budget spread evenly across every employee produces very little behavioural change anywhere. Concentrate the same money on the managers who shape everyone else's daily experience and you get something you can measure. Equality of opportunity is a principle worth holding. Equality of L&D spend is not the same thing, and confusing the two is expensive.

The annual conference trap. Five managers attending five different conferences is not a development program. It's a perk with a training line-item code. The same money spent on one cohort those managers move through together, over months, with practice and accountability between sessions, is a different category of intervention entirely.

Naming these yourself, before the CFO does, buys you credibility for everything else you're defending.

‍

Where to Cut When You Have To

Sometimes the cut is real and you have to absorb it. The principle for what survives is simple.

Protect spending tied to specific managers and specific behaviours. Cohort-based manager training. Coaching with a named person and a named goal. Assessment that tells you where the gap actually is.

Cut spending tied to exposure and awareness. Conference attendance. Platform licenses nobody opens. Content libraries bought for the catalogue rather than the curriculum. Anything whose success metric is "people saw it."

If you have to lose 20%, lose it from the second list entirely before you touch the first.

‍

The Ten-Minute Version

You'll get about ten minutes before the CFO's position hardens. Use them in this order.

Minutes 1 to 3. Give them the number, then take it away. "Training Magazine says $874 per learner, and that's per learner, not per employee. ATD says $846 per employee, down from $1,254, on a sample of about 340 companies. The tech-specific estimates carry a plus-or-minus 20% margin. I can find you a benchmark that says we're overspending and another that says we're underspending, and so can you. I don't think that's the useful question."

Naming the caveats yourself is what earns you the right to reframe. Skipping the homework looks like dodging it.

Minutes 4 to 6. Present the three numbers. Manager engagement at 27%. Managers driving 70% of engagement variance. Top-quartile leadership organizations running roughly double the EBITDA. Then the cost of one regrettable manager departure, at 50% to 200% of salary.

Minutes 7 to 8. Show your allocation and your own data. Sixty percent into direct manager development. Then whatever you have internally: attrition among development-cohort participants versus non-participants, engagement scores by manager, promotion rates. Internal data beats external benchmarks every time, because it can't be countered with a different survey.

Minutes 9 to 10. Name the lag. Manager development runs on an 18 to 24 month feedback loop. A cut made this quarter shows up in retention numbers the year after next, when the causal chain is hard to trace and the person who made the cut has often moved on. The company still pays it. Say that plainly.

The CFO who walked in to cut 30% frequently leaves having cut far less, and feeling like they got a deal. That's the win.

‍

Conclusion

The L&D budget is the easiest line in the company to cut on a quiet Tuesday and the hardest to rebuild after a year without it.

You can find the benchmark. It's $846 per employee from ATD, $874 per learner from Training Magazine, and a tech-specific estimate somewhere between $1,200 and $2,000 carrying a ±20% asterisk. And it won't save you, because your CFO can find a different one in about nine seconds.

Argue about the layer instead. The manager layer is 27% engaged, it drives 70% of the variance, and organizations that fix it earn roughly double the EBITDA. That case is made entirely of numbers your CFO can check, which is precisely why it works.

Build it once. Use it every cycle.

Choose hard.

‍

‍

Frequently Asked Questions: