Confidence Surveys Are Not Perfect. That Is Exactly Why We Should Use Them Better.

The Idea People Love to Shut Down

There is a certain kind of idea in learning and development that people love to shoot down before it ever has a chance to become useful. Confidence surveys are one of those ideas.

The second you mention using a confidence survey, someone will usually say, “That will not work because people will not answer honestly.” Someone else will say, “Nobody is going to complete that.” And honestly, they may be partially right. Some people may not answer honestly. Some people may rush through it. Some people may treat it like another checkbox. But I have also heard people say Storyline is dead because AI exists now, and that does not make it true. It just means people are really good at declaring things useless before they have taken the time to design them well.

Why I Tried It Anyway

I recently built a confidence survey into a Supervisor eLearning program, and I did it very intentionally. This was not a generic “how confident are you?” survey slapped onto the end of a course so we could pretend we measured something. It was carefully worded around the actual themes of the program. The goal was to gather data that did not really exist before, identify areas where supervisors may need more support, and create a better picture of where development opportunities might be hiding.

But the survey was not just built for data collection. That is where I think a lot of these efforts go wrong. We ask people to give us information, then we do nothing meaningful with it. In this case, I wanted the survey to also serve the learner. Based on their responses, the experience helped them develop an action plan for future growth. They could print the action plan, use it as a reflection tool, and bring it into a coaching conversation with their manager or HR business partner. Then, 90 days later, they take the survey again and have the option to use those results to support another coaching and development conversation if they want one.

That is the difference between a survey that collects opinions and a survey that creates movement.

Confidence Is Not Competence

Confidence surveys are not the same thing as measuring performance. We need to be honest about that. A person saying they feel confident having a difficult conversation does not prove they can actually have one well. A supervisor saying they feel prepared to coach an employee does not mean they will coach effectively under pressure. Confidence is not competence. It is not behavior change. It is not business impact. If we pretend it is, we are fooling ourselves.

But confidence does matter. It matters because people are more likely to attempt a skill when they believe they can do it. They are more likely to practice, persist, ask better questions, and step into situations they might otherwise avoid. In learning language, this connects closely to self-efficacy, which is a person’s belief in their ability to organize and carry out the actions needed to perform a task. That belief does not guarantee performance, but it can influence whether someone even tries.

Better Than Another Smile Sheet

That is why confidence surveys can be useful when they are designed well. They can help reveal where people feel strong, where they feel unsure, and where they may need support after the course is over. In the workplace, that matters because so much of learning transfer happens after the training. It happens when the employee tries to use the skill in the real world, with real pressure, real personalities, real constraints, and real consequences.

This is also why confidence questions can sometimes be more useful than traditional smile-sheet questions. Asking “Did you like this course?” gives you one kind of information. Asking “How confident are you in applying this skill with your team?” gives you something closer to readiness. It is still self-reported, and it still has limitations, but it points toward application instead of entertainment. I am not saying reaction data is useless. I am saying we need to stop acting like satisfaction is the same thing as preparation.

The Wording Is Everything

The wording of the questions is everything. Poorly written confidence questions are almost worthless. “Are you confident as a supervisor?” is too broad. Confident in what? Coaching? Documentation? Conflict? Accountability? Giving feedback? Supporting employees? Partnering with HR? Every supervisor could interpret that question differently, which means the data becomes muddy before it even reaches you.

Better confidence questions are specific, behavior-based, and connected to the actual outcomes of the program. Instead of asking, “Do you feel confident in communication?” ask something more like, “How confident are you in preparing for a coaching conversation using specific examples and clear expectations?” That is a much stronger question because it names the behavior. It also gives the learner a clearer mirror. They are not rating their personality. They are rating their readiness to do something.

The scale matters too. If everything is vague, people will guess. If the survey uses labels like “low, medium, high” without defining what those levels mean, you may get data, but you may not get insight. A good confidence scale should make people pause and think. It should help them distinguish between “I understand the concept,” “I could try this with support,” and “I can apply this on my own in a real situation.” That distinction matters.

The Limitations Are Real

There are real drawbacks. People can overestimate themselves. People can underestimate themselves. Some may answer based on how they want to be seen instead of how they actually feel. Others may click through because they are busy, distracted, or tired of surveys. And if the organization has low trust, people may not believe their answers will be used in a helpful way. They may think the survey is a trap, a scorecard, or another corporate exercise pretending to be development.

That is why the purpose has to be clear. If the survey is for development, say that. If the results are meant to help people build an action plan, say that. If the intent is to identify themes and improve support, say that. People are more likely to engage honestly when they understand why the questions exist and how the information will be used. They are also more likely to take it seriously when the survey gives something back to them.

The Survey Cannot Be the Finish Line

The biggest mistake is treating the survey as the finish line. A confidence survey at the end of a course can be a useful pulse check, but it should not be the entire strategy. The real value comes when you connect it to reflection, action planning, manager conversations, HR support, follow-up, and eventually better performance data. That is the part I care about. Not the survey itself, but what the survey makes possible.

In my Supervisor eLearning example, the confidence survey became part of a larger development loop. The learner reflects. The learner identifies growth areas. The learner builds an action plan. The learner can print it. The learner can talk through it with a manager or HR business partner. Then the learner revisits it 90 days later. That gives the experience a before, during, and after. It creates continuity. It gives people a reason to come back to the work.

Perfect Data Is Not Coming

Will everyone take it seriously? No. Will everyone complete it perfectly? No. Will the data be flawless? Absolutely not.

But here is the question I keep coming back to. What is the alternative? In many organizations, the current state is no data, no follow-up, no structured reflection, no coaching conversation, and no practical way to identify where people need help. So yes, a confidence survey has limitations. But a thoughtfully designed confidence survey connected to action planning is still a massive step forward from shrugging our shoulders and saying, “We do not know.”

I think that is where learning professionals need to be more willing to experiment. Not recklessly, but creatively. We should know the limitations. We should design around them. We should avoid overclaiming. We should not pretend self-reported confidence is the same as verified skill. But we also cannot let perfect measurement become the excuse for doing nothing.

I Am Still Going to Try

That is where I get frustrated with the status quo. People will sit in meetings and explain all the reasons something might not work, while offering no better solution. They will point out every weakness in a new idea while defending a current process that produces almost nothing. They will say people will not be honest, people will not complete it, the data will not be perfect, the managers will not follow up, the system will not support it. Maybe. But I am still going to try.

Because in a world where meaningful learning data often does not exist, trying to build something useful is not naïve. It is necessary.

Confidence surveys are not magic. They are not proof. They are not a replacement for observation, coaching, performance metrics, or real behavior change. But when they are carefully written, tied to specific behaviors, connected to development planning, and revisited over time, they can become a practical bridge between learning and action.

And sometimes, that bridge is exactly what people need.

The Signal Is the Starting Point

The goal is not to measure confidence so we can admire the numbers. The goal is to use confidence as a starting signal. A signal that says, “Here is where I feel ready. Here is where I need support. Here is the conversation I need to have next.”

That is the kind of data I want more of. Not because it is perfect, but because it gives us something to build from.

So no, confidence surveys will not solve every learning measurement problem. But they can help us stop guessing. They can help us start better conversations. They can help learners own their development instead of passively completing another course. And if we are willing to design them with care, honesty, and a little bit of courage, they can turn a simple reflection question into the beginning of real growth.

My take is simple: do not worship the data, and do not fear the imperfections. Build the best mechanism you can, connect it to action, and keep improving it.

Because the future of learning will not belong to the people who waited for perfect data. It will belong to the people who were brave enough to create better signals where silence used to be.

Mark Livelsberger, M.A.

Founder | Live Learning & Media LLC