The Kirkpatrick Model: Four Levels of Training Evaluation
The Kirkpatrick model is the standard way to ask whether training actually worked. Four levels, from a simple smile survey to real business results.
The Kirkpatrick model, created by Donald Kirkpatrick in the 1950s, evaluates the effectiveness of training across four escalating levels. Its central insight is that the easy things to measure are the weakest signals, and the valuable things to measure are the hardest.
The four levels
- ReactionDid learners find it engaging and relevant? Usually a post-course survey, often called a smile sheet.
- LearningDid they actually gain the knowledge or skill? Usually a quiz or assessment.
- BehaviorDid they change what they do on the job? Measured by observation and performance data over time.
- ResultsDid the business outcome improve? Fewer errors, higher sales, faster onboarding.
Why most teams stop too early
Most organizations only ever measure levels one and two, because a survey and a quiz are easy. Yet the real value lives at levels three and four, where you find out whether behavior changed and whether the business moved. Measuring behavior often needs data from outside the LMS, which is one reason xAPI exists. Pair Kirkpatrick with Bloom's Taxonomy at the design stage so that what you set out to teach is what you later measure.
Key takeaways
- Four levels: Reaction, Learning, Behavior, Results.
- Levels one and two are easy but weak; three and four carry the value.
- Behavior and results usually require data beyond the LMS.
- Design with the end evaluation in mind.
Frequently asked questions
What are the four Kirkpatrick levels?
Why do most companies only measure levels one and two?
How is Kirkpatrick different from Bloom's Taxonomy?
Related terms
Last reviewed 2026-08-10. Recourse publishes these guides to help learning, customer education, and change teams. Figures and product details are drawn from public sources and vendor materials and can change.