Published on 2nd April 2026

The Kirkpatrick Evaluation Model: How To Use It Well

 

The Kirkpatrick Evaluation Model is one of the most widely used frameworks in Learning and Development, yet it’s also among the most misunderstood.

I’ve been working in leadership development for nearly two decades, and I still see the same pattern. Organisations say they use the Kirkpatrick evaluation model, L&D teams reference it in programme design, but in reality… they’re only really using one or two levels. Usually, the easiest ones. Which means we miss the point entirely.

 

Contents

What is the Kirkpatrick Evaluation Model

The Biggest Mistake People Make With the Model

How to Use the Kirkpatrick Evaluation Model Properly

Designing for Learning, Not Results

When the Data Tells the Wrong Story

Why Measuring ROI in L&D Is Still Difficult

 

What Is the Kirkpatrick Evaluation Model?

The Kirkpatrick evaluation model is a four-level framework used to assess the effectiveness of training programmes:

Reaction – How participants feel about the training

Learning – What knowledge or skills they gain

Behaviour – What they do differently afterwards

Results – The impact on organisational outcomes

On paper, it’s simple.

In practice, it’s where things start to go wrong, because most organisations approach it in the wrong order.

 

The Biggest Mistake: Starting at Level 1 Instead of Level 4

In many cases, the Kirkpatrick evaluation model is presented from Level 1 through to Level 4.

So naturally, that’s how many people approach it.

First, we measure reaction, then learning, maybe behaviour and occasionally…results.

But this creates a fundamental problem.

We focus on what’s easiest to measure, rather than what actually matters.

It’s worth revisiting a variation of a quote often attributed to Albert Einstein:

“Not everything that matters can be measured, and not everything that can be measured matters.”

Nowhere is that truer than in leadership development.

 

How to Use the Kirkpatrick Evaluation Model Properly

 

 

If you want the Kirkpatrick evaluation model to actually work, you need to approach it in the same manner that Kirkpatrick presents it himself.

Start at Level 4, and work down from there.

Level 4: Results (Start Here)

What organisational outcomes are you trying to influence?

This might include:

• Employee engagement scores
• Sales performance
• Retention rates
• Internal promotions
• Reduction in grievances

In many cases, evaluation at this level doesn’t have to involve creating new metrics. It’s more about using existing metrics or returning to what highlighted the need for some sort of training intervention.

Lead and Lag Measures

At level four, it’s often useful to consider both lead and lag measures which we can think of like this:

Lagging measures – outcomes that tell you what has already happened (e.g. sales or deals closed)

Leading measures – the behaviours or activities that drive those outcomes (e.g. number of sales meetings per week)

In simple terms:

Lagging = the result

Leading = what drives the result

For example, if your goal is to improve retention (lag measure), a relevant lead measure might be how often managers are holding meaningful 1:1 conversations or giving regular feedback.

Because if you act on the lead measures, the results usually follow.

 

Level 3: Behaviour

Once you’re clear on the results, you can start thinking about what people need to do differently to achieve those outcomes.

This is where most leadership programmes either succeed… or fail. Examples might include:

• Holding regular 1:1s
• Giving clearer, more structured feedback
• Delegating more effectively

Effective evaluation at this level requires:

• Observation
• Follow-up conversations
• Manager involvement (which is critical)
• Real-world application

 

Level 2: Learning

It’s only at this stage that we start asking questions such as what do people need to learn, or experience, to enable those behaviours?

Unfortunately, this is where most evaluation ends, and that’s the problem.

 

Level 1: Reaction

Finally, we get to Level 1 and the standard ‘happy sheet’.

This is about understanding how participants feel about the training. Level 1 evaluation is absolutely useful, and it’s also limited.

“Great session.”
“Really engaging.”

Comments like this give you some helpful insights about the experience, but they tell you very little about impact. We all know that high reaction scores don’t equal behaviour change.

 

 

The Real Problem: Designing for Learning, Not Results

One of the biggest issues I see with the Kirkpatrick evaluation model is with programmes being designed to hit Level 2 (learning), not Level 4 (results).

You’ll see it in the way learning objectives are written:

“Participants will understand the importance of regular feedback and the BIRD framework.”

It sounds good, but it’s fundamentally flawed because understanding something doesn’t mean we’ll act on it.

We’ve all had experiences where we learn something useful, agree with it and leave with great intentions. And then… nothing changes.

 

How to Design Using the Kirkpatrick Evaluation Model

If you want to apply the Kirkpatrick evaluation model properly, you need to design for outcomes, not objectives.

This starts with being clear about what we want people to do differently and the outcome we’re trying to affect.

If we go back to our learning objective from earlier:

“Participants will understand the importance of regular feedback and the BIRD framework.”

Written as an outcome, this could be:

“Leaders consistently hold regular, structured feedback conversations using the BIRD framework, resulting in clearer expectations, improved performance, and stronger working relationships within their teams.”

Now we’re aligned with Level 3 (Behaviour – what we want leaders to do), and Level 4 (Results – the impact on performance and relationships that we’re trying to affect).

That’s how the Kirkpatrick evaluation model is meant to be used.

 

When the Data Tells the Wrong Story

Even when organisations use the Kirkpatrick evaluation model, interpretation can be tricky.

For example, I often use pre- and post-programme surveys to assess confidence and capability.

Sometimes the data shows no improvement, and at first glance, that looks like failure.

But when you speak to participants, a different picture emerges:

“Before the training, I thought I was good at this. Now I realise there’s a lot more to it.”

This is a classic example of the Dunning–Kruger effect. Increased awareness can lead to more accurate self-assessment – without a change in scores.

So, the data alone doesn’t tell the full story.

 

A Word of Caution: AI and Evaluation Data

Another emerging issue with the Kirkpatrick evaluation model is how we interpret feedback, especially when using AI.

You’ll often see summaries like, “Some participants wanted more time for reflection.”

But what does “some” actually mean?

• 1 out of 15?
• 50% of the group?

Without context, it’s easy to make the wrong decision. We’re already seeing this on platforms like Amazon, where AI summarises reviews with generalised statements.

Things like, “Some users mentioned build quality issues.”

But what does that actually mean?

The same risk applies in L&D. AI can help surface patterns – but it should never replace critical thinking and human judgement.

 

 

Why Measuring ROI in L&D Is Still Difficult

Even when used correctly, the Kirkpatrick evaluation model has its limitations, mostly because L&D never operates in isolation.

Programme outcomes are influenced by factors such as organisational culture, leadership from senior stakeholders, and external market conditions.

Which brings us back to a key point: there is rarely a direct cause-and-effect relationship.

A Simple Test for Your Next Programme

If you’re using the Kirkpatrick evaluation model, ask yourself these questions:

1. Are we starting with results or reaction?
2. Are we measuring behaviour, or just learning?
3. Are our objectives focused on knowledge, or real-world change?

If everything sits at Level 1 and 2, you’re not really evaluating impact.

 

A Final Thought

The Kirkpatrick evaluation model is not the problem.

The way we use it is.

When applied properly, it’s a powerful framework.

But only if we:

• Start with results
• Focus on behaviour
• Design for outcomes
• And interpret data with care

Because ultimately, the value of L&D isn’t found in feedback forms or spreadsheets, it’s found in what people do differently.

And the results that follow.

Your coach,

Ben Morton