Training evaluation · Guide for HR and L&D
The Kirkpatrick model: measuring what training changed
What each of the four levels measures, how to plan the evaluation before the course starts, and how to show what Lean Six Sigma training delivered, with a worked example and a template you can copy.
The Kirkpatrick model evaluates training at four levels: Reaction (did participants find it relevant), Learning (did they gain the knowledge, skills and confidence), Behaviour (do they apply it at work) and Results (did the business outcomes the training targeted improve). Plan the evaluation from Level 4 down, then measure from Level 1 up.
What the Kirkpatrick model is
The Kirkpatrick model is one of the most widely used frameworks for judging whether training worked. Donald L. Kirkpatrick developed it in the 1950s and set it out in a series of journal articles from 1959. James D. Kirkpatrick and Wendy Kayser Kirkpatrick updated it in Kirkpatrick’s Four Levels of Training Evaluation (ATD Press, 2016), and Kirkpatrick Partners maintains it today.
The model asks four questions in order. Did people react well to the training? Did they learn? Do they behave differently at work? Did the organisation get the result it wanted? Each answer needs different evidence, collected at a different time, by different people.
For an HR or L&D manager in Malaysia, the model does one practical job: it turns “the course went well” into evidence a sponsor or finance team will accept. That matters most for longer programmes such as Lean Six Sigma Green Belt or Black Belt, where the investment is large and the results are meant to show up in real processes.
The four levels, one at a time
Kirkpatrick Partners defines each level as follows. The examples show what each level looks like for Lean Six Sigma training.
Reaction
Did participants find the training favourable, engaging and relevant to their work?
Relevance is the part that matters. A course people enjoyed but cannot connect to their job is unlikely to be used. Ask “How useful is this for the project you are working on?” rather than “Did you enjoy the session?”, and check during the course, not only at the end, so the trainer can adjust.
Learning
Did participants gain the intended knowledge, skills, attitude, confidence and commitment?
Use a pre-test and post-test on the same content, practical exercises (for example, building a SIPOC or a control chart from real data) and, for belt programmes, the certification exam. Confidence and commitment to apply the method are early signs of whether Level 3 will follow.
Behaviour
Do participants perform the critical behaviours at work, with the support and accountability to keep doing so?
This is where most training value is won or lost. For a Green Belt, the critical behaviours are concrete: a signed project charter, a data collection plan in use, tollgate reviews held on time. Arrange the support (a sponsor, manager check-ins, coaching) before the course ends, and start tracking in the first weeks rather than waiting three months.
Results
Did the targeted organisational outcomes occur?
Use leading indicators (early signs the result is coming, such as fewer defects in the pilot area) and lagging indicators (the final outcome, such as lower cost of poor quality over a quarter). Much of this data already exists in your production, quality or finance systems.
| Level | Question it answers | Typical evidence | When |
|---|---|---|---|
| 1 Reaction | Was it relevant and engaging? | Short pulse survey with a usefulness question | During and at the end of the course |
| 2 Learning | What did they gain? | Pre and post test, practical exercises, exam, confidence rating | Start and end of the course |
| 3 Behaviour | Are they applying it? | Project milestones, manager observation, tool use | From the first weeks after the course |
| 4 Results | Did the business outcome move? | Process and financial measures, validated | Months after the course, against a baseline |
Sample questions for Levels 1 and 2
Short, specific questions give better data than long forms. These work for any Lean Six Sigma course; use a 1 to 5 scale and keep one open question.
- Level 1, usefulness: “How useful is today’s content for the project or process you are responsible for?”
- Level 1, relevance: “Which tool from today will you use first, and on what?” (open answer; vague answers are an early warning)
- Level 1, support: “Do you know who will support you when you apply this at work?”
- Level 2, confidence: “How confident are you that you could lead the Measure phase of a project next month?”
- Level 2, commitment: “What will you have done with this method within the next 30 days?”
The answers to the support and commitment questions are worth passing to each participant’s manager. They are the first evidence for Level 3, collected before the participant has left the room.
Plan from Level 4 backwards, measure from Level 1 up
The most useful idea in the updated model is the order of planning. Kirkpatrick Partners advises starting with Level 4: agree the business result first, then the behaviours that will produce it, then what people must learn, and only then what the training should feel like. The evaluation runs the other way, from Level 1 during the course to Level 4 months later.
In practice, this means one conversation with the sponsor before anyone is enrolled: what result should change, how it is measured today, who will support participants when they return to work, and when you will look at the numbers.
Worked example: evaluating a Green Belt programme
A manufacturer sends 12 engineers and supervisors through Lean Six Sigma Green Belt training. Each must complete one improvement project within three months of the course. The figures below are invented to show the method; they are not results from a real company.
| Level | What to measure | Method | When | Target | Owner |
|---|---|---|---|---|---|
| 1 | Usefulness for the participant’s project | 3-question pulse survey, 1 to 5 scale | End of each training day | Average 4.2 or higher | Trainer |
| 2 | Knowledge of DMAIC tools; confidence to lead a project | Same 40-question test before and after; confidence rating; certification exam | First and last day; exam after the course | Average gain of 30 points; 10 of 12 pass the exam | Trainer, exam body |
| 3 | Leading a real project with the method | Charter signed; data collection plan in use; tollgate reviews | Weeks 2, 4, 8 and 12 | 10 of 12 projects pass the Measure tollgate by week 8 | Project sponsor, line manager |
| 4 | Each project’s main metric, such as scrap rate or changeover time | Baseline against three months after the Control phase, validated by finance | Months 4 to 6 | Validated savings above the programme cost | Finance, sponsor |
Three things make this plan work. Level 3 is defined as observable project milestones, so nobody has to guess whether behaviour changed. Level 4 uses the measures each project already chose in its charter. And finance validates the savings before anyone reports them, which makes the result credible to management.
Evaluation plan template. Copy the plan above as a table and paste it into Excel or Google Sheets, then replace the measures and targets with your own.
Where HRD Corp’s HRD-TEE fits
If your training is funded through HRD Corp, your trainees already take part in one evaluation. HRD Corp’s Training Effectiveness Evaluation (HRD-TEE), in place since November 2019, uses two standard questionnaires:
- Output assessment: trainees complete it on the final day of every HRD Corp approved course. The training provider collects the forms and submits a summary to HRD Corp within 21 working days.
- Outcome assessment: HRD Corp’s Evaluation Unit sends it to trainees by email after the course, at a time that depends on the programme and up to five years later.
What this means for your own evaluation. HRD-TEE is HRD Corp’s check on the courses it funds. It does not name the Kirkpatrick model and it asks the trainee, not the manager or finance. The Output form overlaps with Level 1; the Outcome form gives a trainee’s view of Levels 3 and 4. Your own plan adds what HRD-TEE cannot see: manager observation of behaviour and validated business results.
Going further: return on investment
Kirkpatrick’s model stops at results. Jack Phillips and the ROI Institute added a fifth level, return on investment, in their ROI Methodology. It asks two extra things: how much of the improvement was caused by the training rather than other changes (isolating the effect, for example with a comparison group or trend analysis), and what that improvement is worth in money.
The Green Belt programme above costs RM 120,000, including fees, participants’ time and coaching. After isolating other factors, finance validates RM 300,000 a year in savings from the 12 projects. ROI = (RM 300,000 minus RM 120,000) divided by RM 120,000, times 100 = 150%.
Calculate ROI only when the sponsor needs it and the data is solid. For most programmes, well-evidenced Level 3 and Level 4 results are more convincing than an ROI figure built on estimates.
Common mistakes
- Stopping at Level 1.An end-of-course survey says how the day felt, not whether anything changed.
- Asking about enjoyment instead of usefulness.In a meta-analysis of training studies, ratings of how useful a course was predicted on-the-job use better than ratings of how much people liked it (correlations of 0.18 and 0.07).
- Choosing Level 4 measures after the course.Without a baseline taken before training, there is nothing to compare the result against.
- Measuring behaviour without arranging support.If no sponsor or manager follows up, a low Level 3 score reflects the workplace as much as the training.
- Claiming every improvement for the training.Other changes happen at the same time. Note them, and isolate the training’s share before quoting money.
Limitations of the model
The levels are not a chain where success at one guarantees the next. The meta-analysis by Alliger and colleagues found weak links between levels, which is why each level needs its own evidence. Reid Bates (2004) argued that the model is incomplete because it leaves out the workplace factors that decide whether training is used, such as manager support and time to practise. The current Kirkpatrick Partners guidance addresses this by building that support into Level 3. Use the model as a structure for evidence, not as proof on its own.
FAQ
What are the four levels of the Kirkpatrick model?
Reaction, Learning, Behaviour and Results. They measure, in order, whether participants found the training relevant, what they gained, whether they apply it at work, and whether the targeted business outcomes improved.
What is Level 5 of training evaluation?
Level 5 is return on investment. It is not part of Kirkpatrick’s original model; it comes from the ROI Methodology developed by Jack Phillips and the ROI Institute, which compares the money value of the results with the cost of the programme.
How soon after training should you measure behaviour?
Start within the first weeks, so problems can be fixed while they are small, and keep measuring until the behaviour is routine. For Lean Six Sigma belts, project tollgates are natural checkpoints.
Does HRD Corp require the Kirkpatrick model?
No. HRD Corp runs its own Training Effectiveness Evaluation (HRD-TEE), with an Output assessment on the last day and an Outcome assessment later. How you evaluate training inside your own company is your choice; the Kirkpatrick model is a sound way to do it.
Key takeaway
Agree the business result before the course, define the behaviours that will produce it, and collect evidence at all four levels. A training programme proves its value at Levels 3 and 4, not on the feedback form.
Lean Six Sigma training for your team
Training that is measured on your own projects
MBizM runs in-house Lean Six Sigma training anywhere in Malaysia, using your own processes as the case material, so the results at Levels 3 and 4 show up in work your team already owns.
Need help designing the whole deployment and how it is measured? See Lean Six Sigma consulting at MBizM.
Related reading
Sources
- Kirkpatrick Partners. The Kirkpatrick Model. Accessed 9 October 2026.
- Kirkpatrick, J. D., and Kirkpatrick, W. K. (2016). Kirkpatrick’s Four Levels of Training Evaluation. Association for Talent Development.
- Kirkpatrick, D. L. Techniques for Evaluating Training Programs (reprint of the original article series). Association for Talent Development.
- Alliger, G. M., Tannenbaum, S. I., Bennett, W., Traver, H., and Shotland, A. (1997). A meta-analysis of the relations among training criteria. Personnel Psychology, 50(2), 341 to 358. Report version: US Air Force Research Laboratory, AFRL-HE-BR-TR-1998-0130.
- Bates, R. (2004). A critical analysis of evaluation practice: the Kirkpatrick model and the principle of beneficence. Evaluation and Program Planning, 27(3), 341 to 347.
- HRD Corp. FAQ: HRD Training Effectiveness Evaluation (HRD-TEE), version 2.0, August 2022; and HRD Corp Support Centre: HRD-TEE. Checked 9 October 2026.
- ROI Institute. ROI Methodology. Accessed 9 October 2026.
