Soft skills training has a reputation for being unmeasurable. It is not. It is badly measured, almost always because nobody wrote down what should be different before the money was spent.

The honest answer for most organisations is that they do not know. The satisfaction survey came back at 4.6, which measures whether people enjoyed the day. That is worth something — a course people hated will not be applied — but it is not evidence of anything you are actually buying.

Soft skills are measurable. They are just measurable in a way that requires deciding what you meant, in advance, which is the step that usually gets skipped.

Start by naming the observable behaviour

“Better communication” cannot be measured because it is not a behaviour. It is a category. Every useful measurement starts by converting the category into something a third party could observe.

If you cannot write the right-hand column, you are not ready to buy the training — not because the training is bad, but because you have not yet decided what you want to be different.

Four things worth measuring, in order of difficulty

1. Did they do the work?

Completion, but specifically the parts with output. Anyone can sit through a video. The written exercise where someone rewrote their own deck is a signal; time-on-page is not. If your course has no output, this level is unavailable to you.

2. Did the behaviour change?

The one that matters most and gets measured least. It needs a baseline, which means observing before, not only after. The cheapest usable version: before the course, ask each participant’s manager to rate three named behaviours on a five-point scale. Repeat at sixty to ninety days. Not scientific, and directionally honest — particularly if the same manager rates both times.

3. Did anything downstream move?

Occasionally you get a clean line: fewer clarification emails after briefings, shorter approval cycles, a higher proportion of proposals accepted first time. When one of these exists in your context, use it. When it does not, do not invent one — a fabricated business metric is worse than an honest behavioural one, because it will be believed.

4. Would it be missed?

Blunt and surprisingly informative: a year later, would the team notice if this had never happened? Ask the managers, not the participants.

The timing that ruins most measurement

Measuring on the last day measures enthusiasm, which peaks precisely then and correlates poorly with anything. Measuring at twelve months measures your organisation, since so much else has changed.

Sixty to ninety days is the useful window. Long enough that the enthusiasm has gone and only the habit remains; short enough to attribute honestly.

Do not over-instrument it

Three named behaviours, one baseline, one follow-up at ninety days is a genuinely good measurement programme, and most organisations do considerably less. A twelve-metric dashboard is usually a sign that nobody decided what mattered, and it will be quietly abandoned by the second cycle.

The question worth asking a provider

Not “what is your completion rate”. Ask: what will participants be able to do afterwards that they cannot do now, and how would I see it?

A provider who answers in behaviours has thought about transfer. A provider who answers in topics covered has thought about the syllabus. Both can be well-intentioned. Only one is selling you the thing you are trying to buy.