Most leadership programs end the same way. A survey goes out on the last day. "How confident do you feel giving difficult feedback?" The numbers come back up. Everyone goes home pleased, and the report to the executive team says the program worked.
Maybe it did. But that survey can't tell you, and it was never built to.
The number everyone reports
Confidence is the easiest thing in the world to measure. It takes one question and thirty seconds. It almost always goes up after training, because people just spent a day thinking hard about the topic and feeling looked after. And it looks great on a slide.
The trouble is what it actually tracks. In 2010, a team led by Traci Sitzmann pulled together 166 studies, covering more than 40,000 learners, where people rated their own knowledge and were also tested on it. How people rated themselves lined up only moderately with what they had actually learned. It lined up much more closely with how much they liked the course and how motivated they felt.
In plain words: when you ask people how confident they are, you mostly learn how they feel about the training. The same review found that nearly a third of the studies it looked at treated those self-ratings as proof of learning anyway.
Feeling smooth is not the same as getting better
There is a sharper version of this. In 2019, a Harvard team split 149 physics students at random between two classes with the same material and the same handouts. One got a polished lecture from a highly rated instructor. The other worked through the problems themselves. The hands-on group scored higher on the test. The lecture group felt they had learned more, and said they wanted all their classes taught that way.
It isn't a one-off. Lab studies going back to 2008 keep finding the same thing, and in a 2016 follow-up, about half the people stayed unconvinced even after being told, in detail, why they were wrong.
Effort felt like failing. Smooth felt like learning. It wasn't.
Leadership training has the same shape. A workshop is smooth by design: a clear model, a tidy role-play with a colleague who is being nice to you, a good lunch. People leave feeling ready. Then Tuesday comes, a direct report goes quiet and defensive, and the model is nowhere to be found.
A confidence score tells you how the training felt. It can't tell you what anyone will say on Tuesday.
Confidence still matters
None of this makes confidence useless. Low confidence is often the reason a hard conversation never happens. A manager who doesn't feel ready to give feedback will put it off for a month, then soften it until it means nothing. So you do want confidence to go up.
You just can't let it stand in for skill. They are two different things, and you need a number for each.
The gap is the finding
Put the two side by side and one of four things happens.
Both go up. The result you wanted. People got better at the conversation, and they know it.
Confidence goes up, skill stays flat. This is the dangerous one. People are now more willing to have conversations they still handle badly. It is how a team ends up giving feedback that is confident and vague at the same time.
Skill goes up, confidence stays flat. People got better and don't believe it. They don't need another workshop. They need more practice and a clear read on their own progress.
Neither moves. Disappointing, but at least now you know.
A confidence survey on its own can't tell the first two apart. Both show up as "confidence went up." That is the whole problem.
Measure what people say, not what they say about themselves
The fix isn't a better survey. It is watching the behaviour. In leadership work, the behaviour is a conversation, so measure the conversation: what the leader actually said when the other person pushed back.
That is how we built Parlare Coaching, and it is why the two numbers are kept apart on purpose. A leader rates their confidence on six realistic moments, like a report who keeps missing deadlines, someone openly resisting a change, or giving their own manager hard feedback. After two weeks they can take it again and see whether it moved. Separately, every practice conversation is scored against a named framework, and the report quotes the leader's own lines to show why.
A team leader never sees anyone's confidence rating, only the team average. What they see is a card that sets how the team feels next to what their practice scores show, framework by framework, and flags where confidence has run ahead of practice. When confidence rises and the scores don't, that isn't a contradiction. It is the most useful line on the page.
The question to ask your next program
The next time someone shows you a confidence chart heading up and to the right, ask one question: compared to what? What did people actually say in the conversation, before and after?
If nobody can answer that, you know how the program felt. You don't know yet whether it worked.