Most evaluations measure a reaction: did people find the session useful, did they enjoy it, would they recommend it. That's worth knowing, but it isn't the same as whether anything changed in how they actually work. Kirkpatrick's Level 3 is the name for that harder, more valuable question — did behaviour change on the job? — and you can design your evaluations to answer it.
The trick is when you ask. A reaction is best captured right after the experience; a change in behaviour only shows once people have had time to try things back at work. So measuring Level 3 is less about a special question type and more about asking later, and asking about practice rather than opinion.
Ask again, once there's been time to apply it
Add a second Evaluation to the learning journey, scheduled weeks after the session rather than right after it. By then participants can tell you what they've actually done differently — not just what they intended to do.
Keep the questions about behaviour in the participant's own work:
an Open-ended question like "What have you done differently since the session?" surfaces concrete examples
a Rating on how often they now use a specific skill or habit gives you something you can compare
Measure the same thing twice to see movement
If you ask the same Rating question early and again later — "how confident are you handling a difficult conversation?" — the gap between the two answers tells you whether the training moved the needle, not just whether people liked it. Asking one scale at two points in time is the simplest way to make a change visible.
Read the two points together
When the follow-up answers come in, read the results alongside the earlier ones. Anonymous responses won't tell you which individual changed, but the pattern across the group — higher self-ratings, real examples of new behaviour — is the evidence that the learning reached the workplace.