L&D Leaders With Level 3 & 4 Evaluation Systems - Audit Evidence

Donald Kirkpatrick published his four-level training evaluation model in 1959. Sixty-seven years later, most organizations still stop at Level 2. A quick refresher. ▸ Level 1 — Reaction. Did they like the training? ▸ Level 2 — Learning. Did they pass the post-test? ▸ Level 3 — Behavior. Are they doing it on the job 90 days later? ▸ Level 4 — Results. Did business outcomes change? Level 1 is a smile sheet. Level 2 is an exam. Levels 1 and 2 are inexpensive, well-instrumented, and present in essentially every training program I've ever audited — from a $563M Department of Defense human-performance contract at Booz Allen, to a regional hospital system, to a national franchise. The Level 1 and Level 2 evidence is genuinely there. Levels 3 and 4 are different. Level 3 requires structured field observation, calibrated evaluators, and a records system that connects the training event to a behavior weeks or months later. Level 4 requires the additional capability to attribute outcome changes to a specific training intervention — rather than to confounding factors. Most training programs cannot produce Level 3 evidence on demand. Almost none can produce Level 4 evidence at all. That gap matters for three reasons. First, the regulators have moved. Across healthcare, federal contracting, franchise insurance, and accredited fitness, the documentation standard is shifting from "training was delivered" to "competency was validated." The artifacts auditors will expect in 2026 are Level 3 artifacts. Second, the workforce already knows. When nurses describe annual training as "repetitive," when operators describe simulations as "checking the block," when employees describe a course they've completed five times — that is the workforce naming a structural problem. A working Level 3 evaluation architecture would have surfaced the problem internally before the engagement survey did. Third, ROI lives at Level 4. The C-suite question — "what did we get for the training spend?" — is a Level 4 question. Programs that can answer it get funded. Programs that cannot, get cut. The fix is not more training content. It is the records architecture that connects training to behavior to outcome. I'll be posting more on this in the weeks ahead — what Level 3 evidence looks like by industry, where most programs break, and what the current Joint Commission, FAR, and franchise-insurer expectations actually require. If your organization has built a Level 3/4 architecture that works, or if you have watched one fail under audit pressure, I want to hear about it. — Adam J. McLean, PhD Founder, McLean Performance Group [email redacted] #TrainingCompliance #Kirkpatrick #LearningAndDevelopment #HumanPerformance #SDVOSB