The Kirkpatrick Model is a four-level framework for evaluating training programs, progressing from learner reaction through to measurable business results. It is the most widely used training evaluation model.
Why it matters#
Most training programs are evaluated only at Level 1 — a post-course survey. The Kirkpatrick Model makes explicit that reaction data is the weakest form of evidence. Working up through the levels gives a complete picture of whether the training actually changed behaviour and delivered business value. Starting the design process at Level 4 and working backwards also ensures that training is aligned to real outcomes from the start.
The four levels#
| Level | Name | What it measures | Common methods |
|---|---|---|---|
| 1 | Reaction | How learners felt about the training | Post-course survey |
| 2 | Learning | Knowledge, skills, and attitudes developed | Post-tests, simulations, assignments |
| 3 | Behaviour | Whether learners apply new skills on the job | Manager observation, follow-up surveys |
| 4 | Results | Whether the training’s business goals were met | KPIs, performance data, business metrics |
How to apply it#
Design from Level 4 down. Define the business result you need first (Level 4), then identify the behaviours that drive it (Level 3), then design learning that builds those behaviours (Level 2), and finally consider how to deliver it in a way that earns learner engagement (Level 1).
For evaluation, collect data at each level after delivery. Level 3 and 4 data requires time — behaviour change and business results take weeks or months to appear.
Key facts#
- Level 1 is the easiest to collect and the least meaningful alone. A course that scores well on reaction surveys may have produced no actual learning or behaviour change. Level 1 data is useful for spotting delivery problems, not for proving impact.
- Pre-testing improves Level 2 accuracy. Without a baseline, you cannot tell whether post-training performance gains came from the training or from knowledge learners already had. Use pre-tests to establish the baseline.
- Level 3 is hardest to measure. On-the-job behaviour change requires time, manager involvement, and follow-up. Many programs skip it because it is costly to collect — but it is the most direct measure of transfer.
- The model works best when training outcomes can be isolated. If other factors (process changes, new tools, management shifts) affect performance at the same time as the training, it is difficult to attribute results cleanly to Level 4.
- The Phillips ROI Methodology extends the model. It adds Level 0 (inputs) and Level 5 (ROI), giving the framework an explicit financial layer.
When to use it#
- When a sponsor wants evidence that training achieved business results
- When designing a new program — use it to define success criteria at each level before development begins
- When existing training is underperforming — diagnose which level is breaking down