How to measure the return on investment of a corporate training plan
The return on investment of training is measured using three main approaches. The Kirkpatrick model evaluates four levels: satisfaction, learning, behaviour and results. The Phillips ROI model adds a fifth level, financial ROI expressed as a percentage. The marginal contribution method isolates training's impact on a specific performance indicator by controlling for other variables. None of these methods is universal: the choice depends on training type, available data and audience to convince.
Why measuring training ROI remains difficult
Measuring training ROI is one of the most discussed and least resolved topics in human resources management. The difficulty is not methodological: the tools exist. It is practical.
The first cause of difficulty is attribution. When a sales person's sales rise 15% after negotiation training, what share of that progress is due to training, what share to market trends, a change in client portfolio, accumulated experience? Isolating training's effect requires a control group that most organisations have not set up.
The second cause is timing. Training effects often appear at 6, 12 or 18 months. Corporate budget cycles are not calibrated for this type of delayed measurement.
The third cause is internal politics. Measuring ROI means admitting that training did not work if indicators do not move. Few organisations have the maturity to draw this conclusion publicly.
The Kirkpatrick model
Developed in the 1950s by Donald Kirkpatrick, this model remains the most cited reference in the literature on training evaluation. It structures evaluation into four levels.
Level 1: Reaction. Did participants like the training? Hot satisfaction questionnaires, completed at the end of a session, measure this level. It is the easiest to measure and the least predictive of real effectiveness.
Level 2: Learning. Did participants acquire the targeted knowledge, skills or attitudes? This level is measured by knowledge tests before and after training (pre-test / post-test).
Level 3: Behaviour. Do participants apply what they learned in their work? This level is measured at 30, 60 or 90 days via observation, manager feedback or transfer questionnaires.
Level 4: Results. Did the training produce an effect on operational indicators (sales, quality, retention, safety)?
The Phillips model
Jack Phillips added a fifth level: financial ROI, calculated as (net benefits - costs) / costs × 100. It requires monetising benefits, which is possible for sales or quality but difficult for leadership or engagement.
Practical advice
Define at T0 the operational indicator that should move if training works. Measure at T+90 days, not only at T+0. Use a control group when possible.