You can say how many people came and how highly they rated the day. Neither number tells you whether anybody can do something on Monday that they could not do on Friday. That is the part you were asked to deliver, and it is the part the end-of-course form never sees.
A form handed out at the end measures the last twenty minutes, the biscuits and how much the room liked the trainer. It is collected before anyone has tried to use any of it, so it cannot measure use. It also arrives too late to help you: by the time you read that a section confused people, the room has gone. Questions asked during the session are a different instrument. They tell you what to reteach while there is still time to reteach it, and they leave you a before-and-after pair rather than a mood. That pair is not proof of training ROI, and this page will not pretend otherwise. It is evidence, which is more than a headcount and a score out of five.
Six questions spread across one course, not asked in a block at the end. The worked example below is a session on handling customer data requests. Swap the subject matter and keep the shape, because the shape is the part that measures the learning outcomes you were asked about.
Ask it in the first two minutes, before you teach anything. It is the easiest question in the room to answer honestly, and it is half of the pair of numbers you will show the person who paid for the day.
A baseline, deliberately unscored. It tells you which half of your material this particular room already knows, so you can spend the day on the half they do not.
Asked immediately after the hardest ten minutes, while you can still fix what it finds. Write each wrong option as a mistake you have genuinely heard — a room that picks the breach-notification window here has merged two deadlines, and that is a specific thing to correct rather than a low score.
The quiz shows you that something did not land. This shows you what. Recognising the right option from a list of four is not the same as being able to explain it, and the gap between the two is where most training quietly fails.
Near the close, when everyone has seen all of it. Ranking forces one topic to the top, which is the difference between a follow-up session with a subject and a follow-up email that says 'any questions'.
Last, in a word or two. Read it for concreteness rather than sentiment: a cloud full of 'careful', 'aware' and 'focus' means the room enjoyed the day and took nothing home, while verbs and object names mean it did.
The score tells you almost nothing; which wrong answer wins tells you everything. Make each distractor a real mistake you have heard someone make, and a failed question becomes a named misconception you can teach against next time.
Open with the unscored question, not the quiz. A leaderboard in the first five minutes teaches the room that answering is risky, and you will spend the rest of the day looking at very cautious data.
If a third of the room misses the concept check and you move to the next slide anyway, you have taught them that the questions are decoration and they will stop answering properly. Leave five minutes in the plan for reteaching, and use it in front of them.
The same confidence question, worded identically, in the last two minutes. One number is an opinion; the same number twice is a change, and it costs thirty seconds to collect.
Send the same two or three questions as a two-minute session a month or two later. A same-day answer partly measures the goodwill of a room that has just had lunch; the six-week answer is about the job, and it is the one a sceptical budget holder will actually read.
Nobody admits confusion in front of their manager, and in most training rooms the manager is present. If people believe a wrong answer can be traced back to them, you are measuring their caution rather than their understanding.
Every question above, ready to edit. Put your own subject matter in, keep the order, and leave with a before-and-after pair instead of a rating out of five.
Retro questions the quiet half of the team will actually answer
Make the decision in the room, not in the thread afterwards
Brainstorming that collects ideas from the whole room, not the three people who like talking
Let the question everyone is thinking get asked without anyone having to own it
Training feedback that names the step where the room lost you