Measuring the ROI of a training plan: the Kirkpatrick and Phillips methods

What is the ROI of training?

The return on investment (ROI) of training expresses, as a percentage, what the training brought in once its cost has been deducted. The formula used by the ROI Institute, which promotes Jack Phillips's method, is:

• ROI (%) = (training benefits − training costs) ÷ training costs × 100
• Benefit-cost ratio (BCR) = training benefits ÷ training costs

An ROI of 0% means the training paid back its cost and nothing more. A BCR of 1.5 means each euro invested produced €1.50 in benefits.

The calculation looks simple. The difficulty lies upstream: knowing which share of the observed improvement is really due to the training, and putting a monetary value on it.

Why ROI measurement remains rare in companies

Three obstacles come up in every organisation:

• Attribution. When an indicator improves after training, other factors may have played a part: the market, a new tool, a change in the team. Without a comparison group, the effect of the training remains a hypothesis.
• Time lag. Effects on work and on results appear after several weeks or months, often outside the budget cycle.
• The cost of measurement. Tracking business indicators before and after takes time from managers and HR teams.

The OECD, in a study of 100 company cases published in 2021, found that enterprises use only basic methods to assess the outcomes of training, mainly employee surveys and observation at work, even though they spend significant time and resources on delivering it.

The Kirkpatrick model: four levels of evaluation

Donald Kirkpatrick designed this model in the 1950s, as part of his doctoral work, and published it as a series of articles starting in 1959. It remains the most widely used training evaluation framework. Kirkpatrick Partners, the organisation that promotes it today, describes it as follows:

Level	Question	Common tools
1. Reaction	Do participants find the training engaging and relevant to their work?	End-of-session questionnaire
2. Learning	Did they acquire the intended knowledge, skills, confidence and commitment?	Pre- and post-tests, role plays
3. Behaviour	Do they apply the expected behaviours on the job, with the support they need?	Manager observation, structured self-assessment
4. Results	Do the targeted organisational outcomes occur because of the programme?	Business indicators (quality, lead times, sales, safety)

Kirkpatrick Partners recommends designing the evaluation starting from Level 4: define the expected result first, then work back to the behaviours and learning that lead to it. The site also states that Level 3 measurement should begin soon after the training rather than months later.

A meta-analysis published in 2003 in the Journal of Applied Psychology (Arthur, Bennett, Edens and Bell) illustrates how unbalanced practice is: among the studies included, 234 measured learning, 122 measured behaviour and only 26 measured results. The average effects observed were similar across levels (Cohen's d between 0.60 and 0.63).

The Phillips method: from Level 4 to ROI

Jack Phillips added a fifth level to the model: ROI. The method, set out in Return on Investment in Training and Performance Improvement Programs (3rd edition, Routledge, 2024), follows five levels: reaction and planned action, learning, application, business impact, ROI.

It adds two steps that make the difference:

1. Isolating the effect of the training. The ROI Institute lists several techniques: control group, trend-line analysis, forecasting, participant or manager estimates, accounting for other factors.
2. Converting results into monetary value. Standard values (cost of an error, of an hour of work), historical data, internal experts, structured estimates.

Benefits that cannot be credibly converted (engagement, team climate, image) are reported separately as intangible benefits. They are not included in the calculation.

How do you know whether your training plan actually worked?

Before calculating an ROI, you need to know whether the training produced a change. A three-step evaluation is enough in most cases:

• Before the training (T0). Measure the starting level against the learning objectives and record the business indicators that should change. Without this baseline, no later measurement can be interpreted.
• At the end of the training. Measure what was learned (Level 2) rather than satisfaction alone. Immediate feedback mainly flags the modules that were poorly understood.
• In the following weeks and months. Ask participants and their managers which practices have actually changed (Level 3), then track the chosen business indicator (Level 4).

For more on indicators, see The 7 indicators every L&D department should track and Measuring skills acquisition beyond completion rates.

Worked example (fictitious case)

The figures below are invented for illustration and are not a benchmark.

A company trains 40 customer advisers in complaint handling (14 hours each).

Item	Calculation	Amount
Training fees	Provider quote	€24,000
Training time	40 × 14 h × €35 (fully loaded hourly cost)	€19,600
Design and evaluation	Internal time	€6,400
Total cost		€50,000

Over the following twelve months, escalated complaints fall by 120 per month in the trained team and by 40 per month in a comparable untrained team. The isolated effect is therefore 80 escalations avoided per month, or 960 per year. With an internal cost of €75 per escalation, the benefit reaches €72,000.

• Net benefits: 72,000 − 50,000 = €22,000
• ROI = 22,000 ÷ 50,000 × 100 = 44%
• BCR = 72,000 ÷ 50,000 = 1.44

Without a control group, the company would have attributed 120 escalations per month to the training and reported an ROI of 116%. The gap shows why isolating the effect is the decisive step.

Which method for which training?

Type of training	Relevant evaluation level	Note
Awareness, regulatory compliance	Levels 1 and 2	The goal is to prove understanding and traceability
High-stakes job training (sales, customer service, technical)	Levels 3 and 4	Choose a business indicator that is already tracked
Costly or strategic programme	Level 5 (Phillips ROI)	Plan the comparison group from the design stage

Calculating an ROI for every course makes no sense: measurement takes time and should remain proportionate to what is at stake in the programme. To prepare an investment decision, see also How to build a business case for an AI training system.