What is the Kirkpatrick model?

The Kirkpatrick model is a four-level framework for evaluating training, created by Donald Kirkpatrick. It measures Reaction (how learners felt about the training), Learning (what they gained), Behavior (whether they apply it on the job) and Results (the effect on business outcomes).

Reviewed by Clark Quinn, Chief Learning Strategist Updated October 28, 2026

Example

A bank’s L&D team rolls out a new anti-money-laundering course for branch staff and plans its evaluation at all four levels. At Level 1, a two-question survey at the end asks whether the course was relevant and clear. At Level 2, a scored quiz checks whether staff can spot the warning signs the course covered.

At Level 3, three months later, branch managers review a sample of flagged transactions to see whether staff escalate suspicious activity the way the course taught. At Level 4, the compliance team compares the quality of escalations and internal audit findings with the period before the course. The survey shows that staff liked the course and the quiz shows that they learned the rules. Only Levels 3 and 4 show whether anything changed at the counter.

How it works

Donald Kirkpatrick described the four levels in a series of articles published in 1959. Each level measures something different, and each is harder to measure than the one before it.

Level 1: Reaction

How learners responded to the training: whether they found it relevant, clear and worth their time. Teams measure it with end-of-course surveys (often called smile sheets), ratings and comments. Level 1 data is cheap and fast to collect, but a good rating does not prove that anyone learned anything.

Level 2: Learning

Whether learners gained the knowledge, skills or attitudes the training targeted. Teams measure it with quizzes, knowledge checks, pre- and post-tests, scenario questions and practical demonstrations. Level 2 works best when every question maps to one of the course’s learning objectives.

Level 3: Behavior

Whether people apply what they learned on the job. Teams measure it with manager observation, checklists, peer feedback and data from work systems, collected weeks or months after the training. Level 3 depends on more than the course: managers need to support the new behavior, and people need the chance and the tools to use it.

Level 4: Results

The effect on the business outcomes the training was meant to change, such as fewer safety incidents, lower error rates, faster ramp-up for new hires or higher customer satisfaction. Level 4 is the hardest level to attribute to training alone, because other things change at the same time.

How to use the Kirkpatrick model

Plan evaluation in reverse order. Start with the Level 4 result the business wants, define the Level 3 behaviors that would produce it, then design the training and the Level 2 assessments around those behaviors. Level 1 comes last. Planning this way also sharpens the course itself, because it forces the team to name what people should do differently.

The New World Kirkpatrick Model, an update by James and Wendy Kirkpatrick, keeps the four levels and puts more weight on the workplace conditions that support Level 3, such as manager reinforcement and follow-up.

Not every course needs all four levels. A short policy update may only need Levels 1 and 2. A large sales or leadership program justifies the effort of measuring Levels 3 and 4. In the ADDIE model, this work belongs to the Evaluation phase, but the measures should be agreed during Analysis.

Kirkpatrick vs the Phillips ROI model

Jack Phillips added a fifth level, return on investment (ROI). Level 5 converts Level 4 results into money and compares them with the full cost of the training, including development, delivery and learner time. ROI is useful when leadership asks for financial justification. It also needs reliable cost data and a credible way to separate the effect of training from everything else that changed.

Limits of the Kirkpatrick model

• The levels are often read as a chain, but a positive reaction does not guarantee learning, and learning does not guarantee changed behavior.
• Levels 1 and 2 are easy to collect in an LMS, so evaluation often stops there.
• Isolating the effect of training at Level 4 is difficult without a comparison group or baseline.

Why it matters for L&D teams

• A common language. Stakeholders quickly understand “we measured to Level 3”, which makes evaluation plans easier to agree.
• Better course design. Starting from the Level 4 result forces clear objectives and keeps content focused on behavior.
• Evidence beyond completions. Completion rates show who finished a course. The four levels show what changed.
• Effort where it counts. Teams can reserve Level 3 and 4 measurement for the programs where business impact matters most.