Ask most organisations to prove a training programme worked and you get one of two numbers: how many people completed it, and how many said they enjoyed it. Both feel like evidence. Neither is. Completion tells you people showed up. Satisfaction tells you the coffee was fine. Whether the business is any better off is a different question, and it is the only one that matters when you are the person who signed the invoice.
The framework worth knowing is Donald Kirkpatrick's four levels, later paired with Jack Phillips's fifth for return on investment. Level one is reaction, did they like it. Level two is learning, did they know more at the end. Level three is behaviour, did they do anything differently. Level four is results, did a business number move. Almost everyone measures level one, because it is free and flattering, and almost nobody reaches level four.
The trap in the top two levels
Reaction and learning are worth measuring, but they are leading indicators at best. People can love a course and change nothing. People can pass the quiz and revert by Friday. If you stop at level two, you are measuring the training, not the transfer, and you are back to the problem that the room was the cheap part. The value is at levels three and four, which is exactly why they get skipped: they take a little longer and they might tell you the truth.
A worked example you can copy
Say a team of ten does a half-day on reducing quote errors, because rework on bad quotes is eating time. You do not need a research department to measure this. You need one number you already have.
The solution, as a worked calculation
Training ROI on the back of one metric
| Step | What you measure | Example |
|---|---|---|
| 1. Baseline | The problem number before training | 40 quote errors a month, 30 minutes each to fix |
| 2. The cost of the problem | Time or money the problem burns | 20 hours a month of rework |
| 3. The change | The same number 60 to 90 days later | Errors fall to 16 a month |
| 4. The gain | Difference, valued | 12 hours a month recovered |
| 5. The cost | Trainer, time off the floor, materials | One-off, say 1,500 |
| 6. ROI | (Annual gain minus cost) / cost | Recovered time pays back the cost in weeks |
The point is not decimal-place precision. It is that you tracked one real number from before to after, and can defend it.
This is the Phillips move: take the level-three behaviour change, attach it to a number that was already being counted, and convert it to money or time. You do not have to isolate every other factor perfectly. You have to be honest about what changed, over what period, against what it cost. That is a defensible answer, and it beats "94 per cent completion" in front of anyone who controls a budget.
Why this protects you, not just the numbers
Measuring at level four does something political as well as analytical. It moves the conversation from "training is a cost we tolerate" to "here is what this specific programme returned". It also tells you fast when something is not working, so you can stop paying for it. The organisations that keep a training budget through hard years are the ones that can point to a number that moved. The ones that lose it first are the ones holding a stack of happy-sheets.
Questions to ask before you call a programme a success
- What number that already exists should move if this worked?
- What was it before, and when will we read it again?
- Are we measuring behaviour and results, or just reaction and recall?
- What did the programme actually cost, including time off the job?
- If the number did not move, will we admit it and stop?
You will never measure training with laboratory certainty, and you do not need to. Pick the one metric the training is meant to move, read it before and after, value the difference, and set it against the cost. Do that and you are no longer guessing whether training paid. You are one of the few people in the building who can actually say.
Notes and sources
From the book
Training That SticksA practical method for designing training that moves behaviour, embedding it in the first thirty days, and proving the difference it made, with no training department.