How to Measure Training ROI: A Practical Guide for L&D and HR Leaders

Why Training ROI Sits on the L&D and HR Agenda Again

By EuroQuest Editorial Team · Published 2026-06-17

Most training-ROI conversations end without a number. The CFO asks the question, the L&D lead delivers a satisfaction score and a completion rate, and the conversation moves on. The problem is rarely effort; it is that the team measured the wrong level of impact and then tried to convert that data into a money number it was never designed to support. This guide is built for training managers, heads of L&D, and HR business partners who have to defend a training budget in a credible way. It walks through the Kirkpatrick four levels, the Phillips ROI addition, a six-step practical method, and a decision matrix for when full-ROI calculation is worth the cost.

4 levelsOf the Kirkpatrick training-evaluation model: Reaction, Learning, Behavior, Results. [Kirkpatrick]
5th levelAdded by Jack Phillips for financial ROI, with a methodology for isolating training's monetary impact. [ROI Institute]
2.9%Share of revenue organizations invested in learning and development, the highest ratio in five years per ATD's 2025 State of the Industry. [ATD]
39%Share of workers' core skills employers expect to change by 2030, the capability pressure every L&D investment now sits against. [WEF]

Why Training ROI Is on the Board Agenda Again

L&D Spend Has Crossed a Visibility Threshold

Learning and development is no longer a back-office line item. ATD's 2025 State of the Industry put L&D investment at 2.9 per cent of revenue, the highest ratio in five years, and finance teams in regulated industries are pushing hard for a defensible link from that spend to measurable workforce capability. Once a budget crosses a visibility threshold inside the firm, the CFO and the audit committee start asking different questions about it.

The board is not asking for a perfect ROI calculation. It is asking for a credible chain of evidence from money spent to capability delivered.

The Skills Shift Has Raised the Stakes

The WEF Future of Jobs Report 2025 documents that 39 per cent of workers' core skills are expected to change by 2030. That is not a marginal workforce-planning question; it is a structural one. Boards that authorize multi-year reskilling commitments expect L&D to demonstrate impact at a level deeper than satisfaction surveys.

The Old Metrics Were Never Designed to Answer Today's Question

Completion rates, satisfaction scores, and attendance numbers were designed to manage program operations, not to defend program value. They answer "did the training happen?" not "did the training matter?" When the CFO asks the second question, the L&D team has to be working with a different measurement model already in place.

The Kirkpatrick Four Levels: What Each One Actually Measures

Level 1: Reaction

Reaction measures whether learners found the program relevant, engaging, and worth their time. End-of-course surveys, net promoter scores, and qualitative feedback all live here. Reaction is necessary, because a program nobody values cannot move the higher levels, but it is rarely sufficient evidence on its own. Most L&D teams over-collect Level 1 data and under-collect everything above it.

Level 2: Learning

Learning measures whether participants acquired the intended knowledge, skill, or attitude. Pre- and post-tests, scenario-based assessments, and skill demonstrations all sit here. Level 2 is the cleanest objective evidence most programs produce, and it is the level where well-designed online and blended formats often outperform classroom because the assessment is built into the platform.

Level 3: Behavior

Behavior measures whether learners actually apply what they learned on the job. This is where most programs fail to produce evidence, not because the change did not happen, but because nobody designed the program to capture the 30-, 60-, or 90-day post-program data point. Harvard Business Review's talent-management library repeatedly documents that behavior-change measurement requires manager involvement, not just learner self-report.

Level 4: Results

Results measures the impact on organizational outcomes — productivity, quality, safety, customer satisfaction, retention, revenue per employee. Level 4 evidence is what most CFOs actually want to see, and the discipline required to produce it is greater than at any prior level. The Kirkpatrick Partners site emphasizes that Level 4 evidence has to be planned for at program design, not retrofitted after delivery.

Phillips Adds Level 5: The Financial ROI Calculation

From Results to Money

Jack Phillips's contribution was to extend the Kirkpatrick architecture with a fifth level: financial ROI. The ROI Institute's methodology walks through how Level 4 Results are converted to monetary terms, how the cost of the program is captured fully, and how training's contribution is isolated from other factors.

The headline formula is simple: ROI per cent equals net program benefits divided by program cost, multiplied by 100. The complexity is in the inputs, especially the isolation step.

The Isolation Problem

Most productivity, quality, or revenue movements have multiple drivers — a new tool, a new manager, a market tailwind, a process redesign, and the training program. Attributing all of the change to training is dishonest, and the audit committee will eventually catch it. Phillips's methodology offers a menu of isolation techniques, from control groups to trend-line forecasting to expert estimation with adjustment for confidence.

When to Stop at Level 4

Phillips himself notes that not every program justifies a Level 5 calculation. The cost of producing the ROI number can exceed the value of having it. For most programs, credible Level 3 and Level 4 evidence is enough; Level 5 is reserved for the strategic, expensive, board-visible programs where the financial answer is the difference between continuation and cancellation.

A Six-Step Practical Method You Can Apply This Quarter

Step 1: Decide the Highest Level You Need to Defend

Not every program needs Level 5. A compliance refresher might only need Level 2 evidence. A senior leadership program almost certainly needs Level 3 and Level 4. Decide the target level at design time and let it shape the measurement plan.

Step 2: Capture Baseline Data Before the Program Starts

No baseline means no credible comparison. Capture the relevant performance, behavior, and outcome data before delivery, even if the measurement looks rough. The discipline of asking what to measure before the program runs is half the value.

Step 3: Build the Manager Into the Measurement

Behavior change does not show up on a learner self-report; it shows up in what the manager observes 30 and 90 days later. A program design that does not contract the manager into the measurement plan will produce weak Level 3 evidence regardless of the rest of the design.

Step 4: Capture Cost Fully and Honestly

Most L&D ROI calculations understate program cost by leaving out participant time away from job, opportunity cost, internal facilitator and design hours, and ongoing platform fees. The CFO will catch this; better to capture cost fully upfront.

Step 5: Isolate Training's Contribution

Use a control group where possible, a trend-line forecast where not, or a structured expert-estimation method with explicit confidence adjustment. Document the isolation method. Reviewers will accept an imperfect method explained well; they will reject a perfect-looking number with no method behind it.

Step 6: Report at the Level Your Audience Needs

The board may want a single Level 5 percentage. The executive sponsor may want a Level 4 narrative. The L&D community of practice may want all four levels. The same evaluation can produce different reports for different audiences; that is a feature, not a problem.

MIT Sloan Management Review's talent-management library documents how the best programs report ROI as a portfolio across all five levels, with the financial number anchored in the lower-level evidence.

When Full-ROI Calculation Is Worth the Cost

Program characteristic Recommended evaluation depth
Compliance or regulatory refresher with stable contentLevel 1 plus Level 2 assessment; Level 3 spot-checks where audit risk is high.
Operational skills program at scaleLevel 1 through Level 3; Level 4 on a sample if productivity is the goal.
Senior leadership or executive programLevel 3 behavior change with manager input; Level 4 on cohort-level outcomes.
Strategic, expensive, multi-year capability buildFull Level 1 through Level 5, with documented isolation and audit-grade cost capture.
Pilot or proof-of-concept programLevel 1 and Level 2 with a structured decision on whether to extend; Level 5 premature.

Most credible L&D portfolios end up with a mix. Compliance and operational programs sit at Level 2 or 3, senior programs sit at Level 3 or 4, and one or two flagship programs each year carry the full Level 5 calculation. The discipline is in matching evaluation depth to program stakes rather than applying one standard everywhere. The OECD's skills agenda repeatedly emphasizes that adult-learning investment without measurement is the dominant pattern in advanced economies, and that the gap is design, not budget.

The CFO is not asking for a perfect ROI number; the CFO is asking for a credible chain of evidence from money spent to capability delivered. The L&D teams that consistently defend their budget are the ones who chose the right level to measure for each program and then measured that level rigorously.

Frequently Asked Questions

Is the Kirkpatrick model still relevant in 2026?

Yes. The four levels remain the dominant evaluation architecture in corporate L&D, and the New World Kirkpatrick refresh has updated the model around manager involvement and required drivers. The criticism is usually of how the model is applied, not of the model itself.

Should every program have a Phillips Level 5 ROI calculation?

No. Phillips himself recommends Level 5 calculation for around five to ten per cent of programs — the strategic, expensive, board-visible ones. For the rest, credible Level 3 and Level 4 evidence is the right target.

How do I isolate training's impact from other variables?

Use a control group where the design allows, a trend-line forecast where the data exists, or a structured expert-estimation method with explicit confidence adjustment. Document the method openly; reviewers accept imperfect methods explained transparently.

What is the most common training-ROI mistake?

Trying to extract Level 4 or Level 5 evidence from a program that was never designed to produce it. The measurement plan has to live inside the program design, not arrive after delivery.

How does AI change training-ROI measurement?

AI-supported analytics make Level 2 and Level 3 evidence cheaper to collect, especially around in-workflow application and microlearning. The harder problems — outcome isolation and full-cost capture — remain organizational and methodological, not technical.

Build Training Programs the Board Can Defend

EuroQuest International delivers executive and professional programs across leadership, HR, and operations categories with measurement and reinforcement built into the program plan. Match evaluation depth to program stakes, capture cost fully, and report at the level your audience actually needs.

Browse EuroQuest Training Categories