Your Training Budget Is Working Harder Than Ever - and Delivering Less
The average organization spends $1,280 per employee per year on direct learning. That figure has barely moved in a decade. But underneath the headline number, something has shifted dramatically: the cost per learning hour used has surged to $103 - a 36% increase from the $76 average that held steady for most of the prior decade.
Organizations aren’t spending more. They’re getting less for what they spend.
This brief examines why - and what a fundamentally different approach to training delivery can change.
$103
Cost per learning hour - up 36%
5.0%
Of payroll spent on learning - an all-time high
33 hrs
Avg. learning hours per employee - declining
Eight Years of Flat Spend, Rising Pressure
Between 2014 and 2021, average direct learning expenditure per employee moved from $1,229 to $1,280 - a cumulative increase of just 4.1% over eight years. During that same period, inflation in advanced economies compounded by over 15%.
In real terms, L&D budgets have been shrinking.
Meanwhile, the demands on those budgets have grown. Managerial and supervisory training surged to 20% of the average learning portfolio - up from 13% just two years prior - driven by the complexities of leading distributed and hybrid teams. This is the most expensive content category to deliver, and it now dominates the training mix.
The math is straightforward: flat budgets, rising costs per hour, and a shift toward the most expensive content types. Something has to give.
The squeeze in one number
Large organizations (10,000+ employees) now spend $1,656 per employee - 73% more than midsize companies ($955). The economies of scale that large employers once enjoyed have reversed. Size no longer guarantees efficiency.
The Delivery Method Trap
Here’s the stat that looks like progress: 67% of learning hours are now delivered via technology-based methods. That number was 56% in 2019.
But look closer. The shift didn’t come from scaling self-paced, low-cost delivery. It came from moving instructor-led classroom training onto Zoom. Virtual instructor-led training (vILT) still requires a live instructor, scheduling coordination, and real-time facilitation - carrying much of the same marginal cost as in-person sessions.
True self-paced e-learning - the kind that scales without adding instructor hours - accounts for only 30% of learning hours used. That figure has been essentially flat since 2020.
The platform shifted. The cost structure didn’t.
This is the core of the $103 problem. Organizations adopted technology to deliver training, but they adopted it as a channel for the same instructor-dependent model. It’s like digitizing a paper form without changing the process behind it - you get the same bottleneck, just on a screen.
The On-the-Job Learning Paradox
Sixty-two percent of organizations say they are committed to on-the-job learning at a “high or very high” level. This makes sense - learning embedded in actual work is more relevant, more timely, and more likely to transfer to performance.
But there’s a problem: almost none of them measure it effectively.
The ATD State of the Industry report has tracked this for six years, and the finding is consistent. Organizations believe in on-the-job learning. They invest in formal training. The gap between those two realities represents one of the largest missed opportunities in talent development.
McKinsey’s 2025 research on the future of L&D reinforces the urgency: while 72% of organizations recognize the need to become skills-based, only 11% report meaningful progress. The gap between aspiration and execution has never been wider.
What would it look like to actually deliver on the promise of learning in the flow of work - with measurement built in?
What’s Inside the Full Report
This research brief goes deeper into each dimension of the cost problem and presents a framework for evaluating AI-native training platforms against traditional delivery models:
- The full $103 breakdown - What’s driving cost per learning hour to an all-time high, broken down by company size and industry
- The vILT illusion - Why “technology-based” delivery hasn’t lowered costs, and what truly scalable delivery looks like
- The managerial training crisis - Why the fastest-growing content category is also the most expensive, and how daily practice changes the equation
- Benchmarks that actually matter - The metrics L&D leaders should track (competency rates, time-to-proficiency, retention after 90 days) vs. the ones that mislead (completion rates, seat time, satisfaction scores)
- The on-the-job learning opportunity - How AI-native platforms close the gap between stated commitment and actual measurement
- Cost modeling: traditional vs. AI-native - A practical framework for estimating what your organization could save by shifting from instructor-dependent to AI-powered delivery
- Building the business case - How to position training infrastructure modernization as a cost reduction initiative, not a new line item