Training evaluation methods are the structured approaches organizations use to judge whether training achieved its goals. The most common models are Kirkpatrick’s four levels, the Phillips ROI Methodology, and Brinkerhoff’s Success Case Method, and each answers a different question about a program.
This guide compares the main training evaluation models, the data collection methods that sit underneath them, and how to choose a combination that fits a management or leadership program.
Models versus methods
People use the two words interchangeably, but they are different layers:
- Evaluation models are frameworks that tell you what to evaluate and in what order (for example, reaction, then learning, then behavior, then results).
- Data collection methods are the tools you use to gather evidence at each stage: surveys, tests, observation, 360 feedback, interviews, business data.
A good evaluation plan picks one model as its spine and then chooses the methods that produce credible evidence at each level.
The main training evaluation models
Kirkpatrick’s four levels
The Kirkpatrick model evaluates reaction (did participants find it favorable and relevant), learning (did they gain the intended knowledge and skills), behavior (do they perform the critical behaviors on the job), and results (did the targeted organizational outcomes occur). Its strength is simplicity. Its common misuse is stopping at levels one and two.
Phillips ROI Methodology
The ROI Methodology extends Kirkpatrick with a fifth level, return on investment, and adds a required step to isolate the program’s effect from other factors before attributing results. Use it when a sponsor needs a financial figure and is willing to fund the data work.
Brinkerhoff’s Success Case Method
The Success Case Method identifies the participants who got the most and least value from training, then interviews them to learn what helped or blocked application. It is fast and practical, and it explains why results varied, which the level-based models do not.
CIPP and other program evaluation models
Broader program evaluation frameworks, such as CIPP (context, input, process, product), look at the whole program lifecycle, including whether the program was the right response to the need in the first place. They suit large, long-running programs where design decisions are still open.
Comparison table
| Model | Core question | Best for | Effort | Main limitation |
|---|---|---|---|---|
| Kirkpatrick | Did it work at each level? | Most programs; a shared language with sponsors | Low to medium | Often stops at reaction and learning |
| Phillips ROI | Was it worth the money? | High-cost programs with a finance audience | High | Data and isolation work is demanding |
| Success Case Method | Why did it work for some and not others? | Improving a program; telling a credible story | Low to medium | Not a measure of average effect |
| CIPP | Was this the right program, run the right way? | Large or new programs still being designed | Medium to high | Less focused on individual behavior change |
Data collection methods at each level
- Reaction: short post-session surveys focused on relevance and intent to apply, not only enjoyment.
- Learning: pre and post knowledge checks, skill demonstrations, role plays scored against a rubric.
- Behavior: manager observation checklists, repeated 360 feedback, commitment tracking, work artifacts. A meta-analysis by Smither, London and Reilly (2005) found that improvement after multisource feedback was larger when feedback was repeated.
- Results: business data such as turnover, engagement, quality, and delivery, compared against a baseline and a comparison group.
For a list of specific measures to choose from, see the guide to training metrics.
How to choose a method for management training
For most management and leadership programs, a combination works better than any single model:
- Use Kirkpatrick as the structure, with real effort at behavior and results.
- Add a Success Case round of interviews at 90 days to learn what helped managers apply the training.
- Use Phillips ROI only where the program is expensive enough, and the sponsor interested enough, to justify the extra work.
The step-by-step process for running this is in our guide on how to evaluate training effectiveness. For measuring business outcomes specifically, see training impact assessment.
What every model leaves out
The standard models measure the individual: what one participant learned and does. Much of what makes management training succeed or fail sits between people. DorWay describes organizations as a Network of Agreements™, and a manager’s new skill only holds if the agreements around them change too: who owns what, by when, and what happens when a commitment slips.
A useful addition to any evaluation is a simple agreement check at the team level: are commitments clearer, kept more often, and renegotiated openly? That is the practical meaning of accountability, and the free PACT™ Leadership Assessment gives a quick read on it.
FAQ
What are the methods of training evaluation?
The main models are Kirkpatrick’s four levels, the Phillips ROI Methodology, Brinkerhoff’s Success Case Method, and broader program models such as CIPP. Data collection methods include surveys, tests, observation, 360 feedback, interviews, and business data.
Which training evaluation model is best?
None is best for every case. Kirkpatrick suits most programs, Phillips suits high-cost programs that need a financial return, and the Success Case Method is strongest for learning why a program worked for some people and not others.
What is the difference between Kirkpatrick and Phillips?
Phillips adds a fifth level, ROI, and requires isolating the program’s effect from other factors before attributing business results to training.
How do you evaluate management training?
Define target behaviors, take a baseline, measure behavior with managers and direct reports at 90 days, check business results at six to twelve months, and interview the most and least successful participants.
