The Kirkpatrick model evaluates training at four levels: Reaction (did learners find it relevant and engaging?), Learning (did they gain the knowledge and skills?), Behaviour (do they apply it on the job?) and Results (did business outcomes improve?). Plan it backwards — start from the Level 4 result the business needs — and measure every level, not just feedback forms.
What is the Kirkpatrick model?
The Kirkpatrick model is a four-level framework for measuring training effectiveness, first proposed by Donald Kirkpatrick in the late 1950s and later refined as the "New World" Kirkpatrick Model. It is the most widely recognised training evaluation model in corporate L&D, and it gives trainers, HR and business leaders a shared vocabulary: "We have great Level 1 scores but no Level 3 data" is a sentence every L&D manager understands.
The model's core idea is a chain of evidence. Learners who react well are more likely to learn; learners who learn are more likely to change behaviour if the workplace supports it; and changed behaviour is what drives business results. A break anywhere in the chain explains why a program didn't deliver.
The 4 levels of training evaluation at a glance
Each level measures something different, uses different tools and happens at a different time. Use this table to plan training effectiveness measurement for any program.
| Level | What it measures | Typical tools | When |
|---|---|---|---|
| 1. Reaction | Engagement, perceived relevance, satisfaction, intent to apply | End-of-session survey, pulse polls, quick verbal check-ins | During and immediately after training |
| 2. Learning | Knowledge, skills, attitude, confidence and commitment gained | Pre/post tests, skill demonstrations, role plays scored against a rubric, case exercises | Before (baseline) and at the end of training |
| 3. Behaviour | Application of skills on the job | Manager observation checklists, 360° feedback, work-sample review, system or CRM data, learner action-plan follow-up | Weeks to a few months after training |
| 4. Results | Business outcomes the training was meant to influence | Operational KPIs: quality, productivity, safety incidents, sales conversion, customer satisfaction, attrition | Months after training, compared with a baseline |
Kirkpatrick levels explained with examples
Level 1: Reaction
Level 1 measures how participants felt about the training — whether it was engaging, relevant to their job and worth their time. It is not a vanity metric when designed well: a low relevance score is an early warning that the program won't transfer to work.
Sample Level 1 questions (rate 1–5 unless stated):
- How relevant was today's content to the work you do every week?
- How confident are you that you can apply what you learned in the next two weeks?
- Which activity was most useful to you, and why? (open)
- What will you do differently at work starting tomorrow? (open)
- What, if anything, might stop you from applying this? (open)
Notice what's missing: questions about the lunch or the room temperature. Focus Level 1 on relevance and intent to apply — those predict what happens next.
Level 2: Learning
Level 2 measures whether participants actually acquired the intended knowledge and skills. The key is to test against the learning objectives set during design, ideally with a baseline before training.
Sample Level 2 items:
- Knowledge (multiple choice): "A customer disputes an invoice amount. According to the escalation policy, what is your first step?"
- Scenario judgement: "Read the case below. Which of the three responses best follows the feedback model, and why?"
- Skill demonstration: "Conduct a five-minute coaching conversation with the observer playing a team member who has missed two deadlines." Scored with a rubric.
- Confidence (1–5): "How confident are you in running this conversation without the job aid?"
Skill demonstrations beat quizzes for anything behavioural. You cannot test whether someone can handle an angry customer with a multiple-choice question.
Level 3: Behaviour
Level 3 measures whether people apply the learning on the job — the level where most training value is won or lost. Behaviour change depends heavily on the environment: manager support, time to practise, and whether systems and incentives reward the new behaviour.
Examples: a manager uses a five-item checklist to observe a supervisor's daily shift huddle; a quality lead reviews ten call recordings per agent for the new greeting script; CRM data shows whether sales staff log discovery notes in the required fields.
Level 4: Results
Level 4 measures the business outcomes the training was designed to influence. These should be the same metrics identified during training needs analysis — first-pass yield, customer complaints, safety incidents, time-to-productivity for new hires, collection cycles or conversion rates.
Results are affected by many factors besides training, so present Level 4 as a chain of evidence (reaction → learning → behaviour → results) rather than claiming training alone caused the change.
How to plan evaluation backwards from Level 4
The most effective way to use the Kirkpatrick model is in reverse order: start with the result, then work down to the training. This keeps the program tied to business needs and makes evaluation possible because baselines and owners are agreed upfront. A simple training needs analysis template is the natural place to capture those baselines.
- Level 4 — agree the result. With the business sponsor, name the one or two metrics the program must move, their current baseline and a realistic target window.
- Level 3 — define critical behaviours. Identify the few on-the-job behaviours that most drive that result, plus the manager support and reinforcement they need.
- Level 2 — set learning objectives. Decide what knowledge and skills people need to perform those behaviours, and how you'll test them.
- Level 1 — design the experience. Build a session that is relevant and practice-heavy enough to earn a strong reaction and intent to apply.
- Deliver, then measure upwards. Collect Levels 1 and 2 in the room, Level 3 after an agreed interval, and Level 4 against the baseline.
This mirrors the Analysis and Design phases of the ADDIE model — which is why good instructional designers plan evaluation before they build a single slide.
Worked example: supervisor training at a Chennai electronics plant
Here is a fictional, illustrative example of planning with the four Kirkpatrick levels.
Scenario: An electronics assembly plant near Chennai has promoted several line operators to shift supervisors. The plant head is concerned about rising rework on one assembly line and suspects shift handovers are inconsistent.
| Level | What was planned |
|---|---|
| 4. Results | Rework rate on the line and number of handover-related defects, tracked monthly against the previous quarter's baseline. |
| 3. Behaviour | Supervisors run a structured ten-minute handover using a standard sheet, and hold a daily start-of-shift huddle. The production manager observes two handovers per supervisor each fortnight using a checklist. |
| 2. Learning | Participants demonstrate a complete handover in a simulated shift change, scored against a rubric; a short test on escalation rules; pre/post confidence rating. |
| 1. Reaction | End-of-day survey on relevance and intent to apply; open question on barriers. Instructions available in Tamil and English. |
Because the barrier question revealed that handover time wasn't built into the shift roster, the plant added a ten-minute overlap. Without that Level 1 insight, Level 3 would likely have failed — and the training would have been blamed.
Phillips ROI: the "Level 5" note
The Phillips ROI Methodology, developed by Jack Phillips, extends Kirkpatrick with a fifth level: Return on Investment. It converts Level 4 results into monetary value, isolates the effect of training from other influences (for example, using a comparison group or estimates adjusted for confidence), and calculates ROI as net program benefits divided by fully loaded program costs, expressed as a percentage.
Full ROI studies take effort, so reserve them for expensive, strategic or high-visibility programs. For most sessions, a credible Level 3 and Level 4 story is more useful than a precise-looking ROI figure built on shaky assumptions.
Common mistakes in training effectiveness measurement
- Stopping at happy sheets. A 4.8/5 feedback score says learners enjoyed the day, not that anything changed at work.
- No baseline. Without "before" data, you can't show improvement at Level 2 or Level 4.
- Leaving managers out. Managers are the biggest driver of Level 3. Brief them before training and give them a simple observation checklist.
- Measuring too early. Checking behaviour a week after a skill that's used monthly produces false negatives.
- Claiming sole credit. Present results as a chain of evidence, not "training caused a 20% jump".
- Collecting data nobody uses. Every measure should have an owner and a review date that leads to decisions.
Learning the Kirkpatrick model in the Bodhih Certified TTT
Kirkpatrick's four levels are covered on Day 4 (Assessment & Evaluation) of the Bodhih Certified Train The Trainer (TTT) Program. Participants build an evaluation plan for the module they have redesigned during the program, using Bodhih's feedback-form and design templates, and complete their recorded final certification delivery the same day. The 40-day post-program micro-learning journey then supports the Level 3 part — applying what you've learned back at work. Definitions of related terms are in our glossary, and facilitation skills that protect Level 1 and 2 outcomes are covered in how to handle difficult participants.
Frequently asked questions
What are the 4 levels of the Kirkpatrick model?
The four levels are Level 1 Reaction (did participants find the training engaging and relevant?), Level 2 Learning (did they gain the intended knowledge, skills and confidence?), Level 3 Behaviour (are they applying it on the job?) and Level 4 Results (did business outcomes improve?). Each level builds on the one before it.
Which Kirkpatrick level is most important?
Level 3 (Behaviour) and Level 4 (Results) matter most to the business, because they show whether training changed how people work and what the organisation achieved. But the levels form a chain: without positive reaction and real learning, behaviour change is unlikely. Plan from Level 4 downwards and measure at every level.
What is the difference between Kirkpatrick and Phillips ROI?
The Phillips ROI methodology adds a fifth level to Kirkpatrick: Return on Investment. It converts Level 4 results into money, isolates the effect of training from other factors, and compares net benefits with the full program cost as a percentage. It is best reserved for expensive, high-visibility programs where leaders need a financial case.
How soon after training should Level 3 be measured?
Measure behaviour once people have had a genuine chance to apply the skill — typically somewhere between a few weeks and three months after training, depending on how often the task occurs. A skill used daily can be checked sooner; a quarterly activity such as performance reviews needs a longer window.
Why do most organisations stop at Level 1 and Level 2?
Levels 1 and 2 are easy to collect in the room, while Levels 3 and 4 need manager involvement, baseline data and follow-up weeks later. The fix is to agree Level 3 and 4 measures with stakeholders during the design phase, before the program runs, so the data and owners are in place.
Where can I learn to apply the Kirkpatrick model?
Kirkpatrick's four levels are taught on Day 4 of the Bodhih Certified Train The Trainer (TTT) Program, where participants build an evaluation plan for their own training module alongside their final certification delivery. The 40-day post-program micro-learning journey then helps participants apply the skills back at work, which is exactly what Level 3 measures.