How much should you spend on employee training? 2026 benchmarks and a simpler model
Most companies spend somewhere in the range of $1,200 to $1,500 per employee per year on training, or about 1 to 2 percent of payroll, based on public industry reports. That is a useful anchor, but it answers the wrong question. Spend per seat tells you what you bought. It says nothing about what your people actually finished, learned, or used. This piece walks through the real benchmarks, where they come from, and a simpler model that budgets for outcomes instead of access.
How much do companies spend on training per employee in 2026?
The most cited public figures land in a directional range of roughly $1,200 to $1,500 per employee per year. Two long-running sources shape that number. Training Magazine publishes an annual Training Industry Report, and the Association for Talent Development publishes its State of the Industry study. Both have reported per-employee spend in the low four figures and payroll allocation of about 1 to 2 percent for years, with normal wobble by industry and company size.
Treat these as ranges, not laws. The averages blend very different things: instructor salaries, platform licenses, content subscriptions, travel, and admin overhead. A 200-person company without a formal L&D team spends nothing like a 10,000-person enterprise with a training department, yet both feed the same average. So the honest read is this. The benchmark gives you a sanity check, not a target. If you are wildly above or below it, ask why. If you are inside it, that tells you almost nothing about whether the money worked.
What percentage of payroll should go to learning and development?
Public benchmarks have long clustered around 1 to 2 percent of payroll, and that band is a reasonable planning anchor for most mid-sized companies. It is popular because it scales. A percentage of payroll grows and shrinks with headcount automatically, which makes it easy to defend in a budget meeting.
But a percentage is an input, and inputs are the easiest thing in the world to spend without effect. You can hit 2 percent of payroll and still have a workforce that opened three modules and finished none. The percentage answers "did we allocate enough money to look serious about development." It does not answer "did anyone get better at anything." Smaller companies in particular often over-index on the input, buy a big platform to check the box, then watch usage collapse. The hidden cost of LMS bloat is exactly this: paying for capacity nobody consumes.
Why is spend per seat the wrong way to budget?
Spend per seat is the wrong unit because a seat is permission to learn, not learning itself. When you buy a course library or an LMS, you pay for access across your whole population, then hope engagement follows. It rarely does. Industry completion rates for self-directed corporate content are notoriously low, often in the single digits to low double digits for optional courses. That means the majority of a per-seat budget funds shelfware.
Here is the trap in plain terms. If you pay $200 per person for a library and 10 percent of people finish one meaningful thing, your effective cost per completed outcome is not $200. It is closer to $2,000, because ninety of every hundred seats produced nothing. The sticker price looked cheap. The delivered price was brutal. This is why team training so often fails to stick, and why bigger platform budgets do not reliably buy bigger capability.
What is cost per completed skill, and why does it beat cost per seat?
Cost per completed skill is your total program cost divided by the number of skills people actually mastered, and it is the single most honest number in L&D budgeting. Where cost per seat measures what you bought, cost per completed skill measures what you got.
The math is simple. Take everything you spent: platform, content, admin time, and manager hours. Divide it by real completions, meaning a person moved from not knowing something to demonstrably applying it. That denominator is where most programs quietly fall apart, because they were never designed to be finished. A 40-hour course that 8 percent of people complete has a terrible cost per completed skill even if the license was a bargain. A short, personalized daily habit that most people actually finish can have a far lower cost per completed skill even at a higher per-seat price.
If you want the full formulas for tying spend to results, our guide to the ROI of learning programs lays out the calculations step by step. The headline idea is to move your denominator from seats to finished, applied skills.
How does a $9 or $15 per seat tool change the math versus a big LMS?
A focused tool priced at about $9 to $15 per seat per month changes the math in two ways: it makes cost predictable, and it attacks the completion problem directly, which is where per-seat waste actually hides. Predictable pricing is the smaller benefit. The bigger one is finishing.
A traditional LMS carries layered costs that are easy to underestimate: the platform fee, separate content licenses, implementation, and the admin time to assign, chase, and report. Those costs land whether or not anyone learns. A lean per-seat model collapses most of that into one line. But the decisive difference is design. If a tool is built so people finish, your cost per completed skill drops even when the per-seat price is higher than a bulk library, because the denominator finally moves.
This is the logic behind how AI personalization changes the L&D budget conversation. When the system serves each person the next right thing instead of a catalog to browse, completion stops being a hope and becomes the default. That is the whole ballgame for budget efficiency.
What is a simpler model for setting a training budget?
The simpler model is three steps: anchor on the benchmark, then rebudget around finishing and proof instead of access. It replaces "how many seats can we afford" with "how much capability can we produce per dollar."
Step one, anchor. Use the 1 to 2 percent of payroll band, or the $1,200 to $1,500 per-employee range, as a rough ceiling to plan against. This keeps you honest about scale.
Step two, budget for finishing. Ask of every line item, "does this make completion more or less likely." A 40-hour course you assign and forget lowers completion. A 10-minute daily habit that fits into the workday raises it. Weight your spend toward formats that get finished. The science of microlearning explains why short, spaced, personalized lessons outperform long ones on retention and completion.
Step three, budget for proof. Set aside room to actually measure movement, not activity. The Kirkpatrick model gives you four levels, from reaction to results, and you can reach the first two without a heavy analytics function. Proof is what turns next year's budget conversation from a defense into a request.
Where does Omie fit in this model?
Omie is built for the finishing and proof problem, which is exactly where per-seat budgets leak. It turns development into a 10-minute daily habit, personalizing one lesson a day to each person's role, goals, and behavior, then showing managers and HR whether skills actually moved. The library holds over 100,000 micro-lessons, a recommender picks the next best one for each person, and mastery tracking plus Kirkpatrick Level 1 to Level 4 rollups give you the proof layer without a dedicated analytics team.
On price, Omie is Premium at $9 per month for individuals and Business at $15 per seat per month for teams, with unlimited scans, a manager dashboard, and SSO on Business. You can see the full breakdown on our pricing page. The point is not that a lower sticker price is automatically better. The point is that a tool designed to be finished changes your cost per completed skill, which is the number that actually decides whether a training budget was money well spent.
The takeaway
Budget benchmarks are a floor for the conversation, not the conversation itself. Anchor on the public ranges of roughly $1,200 to $1,500 per employee or 1 to 2 percent of payroll, then throw out spend per seat as your working metric. Budget for finishing, measure with something honest like cost per completed skill, and reserve room for proof. Do that and the size of the number matters far less than what it produces.
If you want to pressure-test your own model, see how Omie works for teams or browse the library to get a feel for the format. No fabricated case studies, no invented numbers, just a product built around the two things a training budget usually fails to buy: completion and evidence.
Frequently asked questions
How much should you spend on training per employee? Public industry reports point to a directional range of roughly $1,200 to $1,500 per employee per year, or about 1 to 2 percent of payroll. Treat that as a starting anchor, not a target. The better question is how much you spend per skill people actually finish and apply.
What percentage of payroll should go to learning and development? Public benchmarks from sources like ATD and the Training Industry Report have long clustered around 1 to 2 percent of payroll. Smaller companies without a large L&D team often sit lower and lean on tools rather than headcount. The percentage matters less than whether the spend produces measurable capability.
What is a good cost per employee for corporate training? There is no single correct number. Reported averages sit in the low four figures per employee per year, but that blends salaries, systems, content, and instructor time. A cleaner metric is cost per completed skill: total cost divided by skills people actually master, not seats you bought.
Is a per-seat learning tool cheaper than a full LMS? Often yes, especially for teams without a dedicated L&D function. A focused tool at around $9 to $15 per seat per month has a predictable cost, while a traditional LMS carries platform fees, content licenses, and admin time. The real saving comes from higher completion, because spend on courses nobody finishes is pure waste.
How do you measure training ROI without a big analytics team? Start with completion and behavior, not hours logged. Track whether skills moved with a simple before and after signal, then tie it to one business metric a manager already watches. Frameworks like Kirkpatrick give you four levels to climb, and you can reach the first two without heavy tooling.