Having just returned from several state annual work comp conferences, in speaking with training individuals a frequent question I heard was what do other organizations spend on training? It is a fair question. It is also a trap, because the answer depends entirely on which survey you open.
Symptom versus finding
These trainers have to deal with a CFO who noted a specific article/review says, “We are below the industry average, so we are fine,” or “We are above it, so cut.” Either way, a single benchmark number is being treated as a verdict.
The objective finding is that the published averages are not measuring the same thing, and none of them measures effect on claims performance. A training average describes what organizations spend. It does not describe what they get back.
What the two sources actually say
ATD reported in its 2026 State of the Industry summary that direct learning expenditure was $846 per employee in 2025, down from $1,254 in 2024. Formal learning hours rose from 13.7 to 16.7 per employee over the same period. ATD gave no explanation for the drop in the summary, and the summary does not report spending as a share of payroll. A one-year fall of about 32%, alongside rising hours, is more likely a change in who was surveyed or how costs were counted than a real collapse in spending. That is my inference, and it is flagged as one.
The 2025 Training Industry Report from Training magazine reported $874 per learner, up from $774 the year before. It also split the figure by size: $468 for large organizations, $782 for midsize, and $1,091 for small. Hours per employee were 40, down from 47. The survey covered U.S. organizations with 100 or more employees.
Read those side by side. One source says spending per person fell sharply. The other says it rose. One counts about 17 hours. The other counts about 40. Neither is wrong. They define spending, learners, and hours differently, and they sample different organizations.
Why this matters for a claims operation
Take a 50-adjuster unit. The table shows what the same headcount costs under each published figure.
| Item | Amount | Basis |
| 50 adjusters at $846 | $42,300 | ATD 2026 State of the Industry, 2025 data |
| 50 adjusters at $874 | $43,700 | Training magazine 2025 report, all organizations |
| 50 adjusters at $782 | $39,100 | Training magazine 2025 report, midsize organizations |
| 50 adjusters at $468 | $23,400 | Training magazine 2025 report, large organizations |
| Spread, highest to lowest | $20,300 | $43,700 minus $23,400 |
A $20,300 spread, on a unit this size, comes entirely from picking a different survey. Size matters too. The large-organization figure is lower per head because fixed costs spread across more people, not because large organizations value training less. Compare yourself only against your own size band and say which source you used.
What no survey will tell you
I searched for a published benchmark of training spend specific to claims adjusters or workers’ compensation operations and did not find a reliable one. APQC lists training and development benchmarks for insurance, but the public page names the metrics without giving values. If a vendor quotes you a claims-industry training average, ask for the source and the sample. If they cannot provide either, treat the number as an opinion.
This is the honest limit of benchmarking. You can learn roughly what organizations spend. You cannot learn from a survey whether your spend is producing the claim outcomes you want.
A better way to benchmark
Use the published averages for one purpose only: to show that your request falls inside a defensible range. Then move the argument to the numbers you can verify yourself.
- Pick one source, one size band, and cite them in the request. Do not mix sources.
- Record your own spend per adjuster and hours per adjuster, defined the same way each year. Consistency beats comparison.
- Set the spend beside your scorecard: claim duration, reopen rate, attorney involvement, reserve accuracy, and first-two-year turnover.
- Review the pairing every year. A flat spend line with a worsening scorecard is a finding. So is the reverse.
The sensitivity test
Earlier posts in this series used a break-even frame, and it applies here. At $846 per adjuster, a 50-adjuster unit spends $42,300. WCRI’s October 2024 attorney-representation study found representation raised indemnity payments by $7,700 to $12,400 per claim, so the low end of that range means 5.5 avoided represented claims cover the whole amount ($42,300 divided by $7,700). That is a threshold to test against your own loss run, not a forecast. The same arithmetic shows why chasing a lower benchmark to save $20,000 is a poor trade if it costs even three represented claims.
The takeaway
Benchmark to stay credible. Measure to stay funded. A published average gets you into the conversation. Your own claim outcomes win it.
