Training Survey Questions: What to Ask After Training

Every guide on this topic gives you thirty questions. Some give more than a hundred. If you run a small business and just spent a Thursday afternoon on a training session, you do not need a hundred questions. You need six, sent by tomorrow morning, and you need to know what to do when four people rate it a five and nobody can name anything they learned.

That last part is where most training feedback goes wrong, and it is the part almost nobody writes about. Collecting the answers is easy. Reading them honestly is harder, particularly when you designed the training yourself and the people answering work ten feet away from you.

This guide starts with a six-question survey you can copy and send today. Then it gives you the full question bank organized by what each question actually tells you, so you can swap questions in when you need something more specific.

The guide also covers what the vendor listicles skip: how to read the answers when six responses is your whole dataset, what to do about anonymity when everyone recognizes each other's writing, and the follow-up survey six weeks later that matters more than the one you send tomorrow. I built FirstHR for small businesses where the person running the training is also the person running the company.

What Training Survey Questions Are For

The framing that keeps a survey useful is to write down, before you draft any questions, the decisions the answers should inform. For most small businesses there are exactly two: should we run this again, and what should be different. A question that cannot change either answer is a question that costs you completion rate for nothing.

Satisfaction and effectiveness are also different things, and surveys that measure only satisfaction are the reason training gets a reputation for being a waste. A session everyone enjoyed and nobody applied scores well and produces nothing. Any survey worth sending has to ask at least one question that could come back negative even when people liked the session.

The Six-Question Survey

Copy this. It works for almost any internal training session, takes participants about two minutes, and produces enough signal to make a decision.

Each question is doing specific work. Question 1 catches the training that was fine in the abstract and irrelevant to the actual job, which is a common failure. Question 2 is the only one measuring capability rather than sentiment, and the text box keeps it honest, since it is easy to click yes and hard to invent a specific example.

Question 3 predicts application better than any satisfaction score. A University of Nebraska-Lincoln research summary (2022) of the meta-analytic evidence, which pools the results of many studies, finds that a trainee's confidence in their ability to perform predicts whether the training gets used on the job. Ratings of the trainer, materials, and setting show no relationship to on-the-job use.

Questions 4 and 5 are the pair that produces your actual edits for next time, and question 5 is the more valuable of the two because people are more specific about what wasted their time than about what helped. Question 6 catches the gap between understanding something and being able to use it, which is where most training quietly fails.

Why Shorter Wins at Small Scale

The arithmetic is unforgiving. If you have ten people and a long survey gets you four responses, you have a dataset of four, and one grumpy response swings your average by a quarter. If a short survey gets you eight, you have something you can read.

At enterprise scale a thirty-question survey with a thirty percent response rate still produces hundreds of responses, and the statistics work. At your scale, response rate is not a metric. It is the entire question of whether the exercise means anything. Every question you add is a trade of completeness against completion, and at small scale completion wins nearly every time.

This is also why the question banks of thirty to more than a hundred questions that dominate this topic are misleading rather than generous. They are menus, not surveys. Use them the way you would use a menu: pick what you need for this specific session and leave the rest.

The Four Evaluation Levels

The standard framework for training evaluation is the Kirkpatrick model, first set out by Donald Kirkpatrick in the 1950s and refined over subsequent decades. Its four levels are worth knowing not because you should measure all four, but because they explain why most training surveys tell you so little.

A typical small business training survey is a level 1 survey. Did you like it? Was the instructor good? Was the room comfortable? Those answers are easy to collect, and they do not tell you whether anything changed.

The federal Field Guide to Training Evaluation, which the Office of Personnel Management builds on these same four levels, makes the point that evaluation is planned from the intended result backwards rather than from reaction. That is exactly the inversion most surveys make.

The practical implication for a small business: put one level 2 question in your immediate survey, and send a level 3 survey six weeks later. Those two moves cost you almost nothing and produce most of the value the full framework offers. Federal guidance on planning and evaluating training covers the structured version.

Reaction Questions

These measure how the session landed. Useful for catching something genuinely bad, weak as evidence of anything else. Pick one or two, not all of them.

Relevance is the reaction question worth protecting. According to the SHRM Workplace Learning and Development Trends report (2022), 24 percent of employees name training that is not relevant to their role among their frustrations with employer-provided training, and 25 percent say they quickly forget the material once the session ends.

Note what is missing: questions about the room, the catering, the slides. They generate answers and almost never change a decision. If the venue was genuinely terrible you will hear about it without asking.

Learning Questions

These ask whether capability changed. This is the category most small business surveys omit entirely, and omitting it is why so many surveys report success and produce nothing.

The before-and-after pairing in that last row is worth using when you want something comparable across sessions. It is self-reported, so treat it as directional rather than precise, but a session where the average moves from 2 to 4 is doing something a session that moves from 3.5 to 3.8 is not.

Behavior Questions: The Six-Week Survey

This is the survey that is easiest to skip and the one that answers the question you actually care about. Four to eight weeks after the session, send two to four questions from the list below. That is the whole thing.

The third question in that table is the one to watch. When people say they have not applied the training, forgetting is not usually the reason. More often the old way still works, nobody expects the new way, the tooling makes it harder, or there has been no occasion to use it. None of those are training failures and none are fixed by running the session again.

Put a reminder in the calendar for six weeks out on the day you run the session, because otherwise this is the survey that never gets sent. The harder half is what happens when the replies arrive, since a follow-up that nobody draws a conclusion from is just a second round of unread answers. Fill this in once they are back, before you decide anything.

Content and Delivery Questions

These produce your edit list for the next run. Two of them are usually enough.

For online training, that last question is genuinely all you need. Long technology sections in surveys produce detailed complaints about things you cannot change and crowd out questions that matter.

Pre-Training Questions

Sending three questions a week before the session is unusual at small scale and unusually valuable, because it shows you what the group already knows, so you can cut it. A session feels like a waste of time, in my experience, when people sit through material they have already mastered.

The answers also give you a baseline. Asking the same capability question before and six weeks after is the cleanest before-and-after comparison available without formal testing.

Choosing a Question Format

Pick the simplest format that fits each question: a labeled 1 to 5 scale for attitudes, a yes or no with a text box for capability, and open text for specifics. The table below covers when each format fits and what to watch for.

The 1-to-10 scale is the most common format mistake. Wide scales need volume to separate anything, and with eight respondents a 1-to-5 scale with labeled ends produces more usable variation and takes less thought to answer.

When to Send Each Survey

If you only run two surveys, run the immediate one within 24 hours of the session and the follow-up four to eight weeks later. The pre-training and annual surveys are optional additions.

The immediate survey alone is the standard practice, and it systematically overstates how well training worked, because it measures the moment when people are most positive and least informed about whether they will use it.

Anonymity When Your Team Is Eight People

Most guides recommend anonymous surveys. On a team of eight that advice does not survive contact with reality, and pretending otherwise damages trust more than being straightforward about it.

With eight respondents, writing style identifies people. So does any specific reference: mentioning a particular customer, a specific shift, a piece of software only two people use. You will frequently know who wrote what, and your team knows you will know. Promising full anonymity and then obviously recognizing a response is worse than never promising it.

Federal practice backs this up. The Office of Personnel Management's Employee Survey Playbook (July 2026) tells agencies to suppress results for any group with fewer than ten respondents, and it calls that threshold a minimum standard. A team of eight falls below it.

What you can promise credibly and keep: individual responses will not be shared with anyone else, you will not raise a specific response with the person you think wrote it, and you will act on themes rather than on individuals. That is a limited, honest commitment, and it produces more candor than an anonymity guarantee everyone privately discounts.

For genuinely sensitive topics, a short conversation with each person separately often produces more honest input than a survey, because it removes the pretense of anonymity that both sides know is fictional.

Getting People to Respond

Aim for 70 to 80 percent on a small team, which means about six out of eight people, or seven to eight out of ten. Small teams should hit much higher rates than large organizations, because the request comes from a person rather than a system. If you are under half, the cause is almost always one of three things.

First, the survey is too long. Cutting from twelve questions to six is the fastest fix available to you.

Second, nothing visibly happened last time. If the previous survey produced no observable change, people have correctly concluded that answering is pointless. This compounds, and it takes exactly one visible action to reverse.

Third, it arrived too late. A survey two weeks after the session feels disconnected from it. Twenty-four hours keeps the association intact.

Two things help beyond fixing those causes: ask in person or in the team channel rather than only by email, and say how long the survey takes. Two minutes is a promise people can evaluate. "Please complete this survey" is not.

Incentives do not help much at small scale. They tend to make the exercise feel transactional and produce faster, thinner answers.

Reading the Answers

This is the section the vendor listicles leave out, and on a small team it is most of the work. With six responses you are not doing statistics. You are reading for patterns, and some patterns are easy to misread.

The general rule with a small dataset: ignore the averages and read the text. An average of 4.2 across six people tells you nothing you can act on. One person writing that the second half repeated what they already knew tells you what to cut. On a team this size, the open text fields are the survey and the ratings are context.

The second rule: weight what several people independently say far above what one person says emphatically. Independent convergence is the strongest signal available when your sample is six.

Closing the Loop

Closing the loop is the step that determines whether you ever get useful feedback again, and it takes about five minutes.

Gallup's guide to employee surveys warns that asking for feedback without acting on it can backfire, because employees disengage when they feel their input is ignored. People who see action, by contrast, trust the process and engage more deeply.

After you have read the responses, tell the team what you found and what you are changing. Two informal lines in the team channel do more for your next response rate than anything else available to you: "Most of you said the compliance section was too long, so next time we are cutting it in half and adding practice time."

Include what you are not changing and why, when there is something. If two people asked for something you cannot do, saying so directly is better than silence, which reads as the feedback having gone nowhere. People accept a no with a reason far more easily than they accept being ignored.

Then actually make the change before the next session. The gap between saying you will cut the compliance section and cutting it is where credibility is lost, and on a small team everyone will notice.

Common Mistakes

These recur across small businesses running training feedback for the first time. Most come from borrowing an enterprise survey design without adjusting for a team where six responses is the entire dataset.

The costliest is the third: skipping the six-week follow-up. The immediate survey measures enthusiasm at its peak, and the follow-up measures whether the work changed. Running only the first and concluding the training worked is how a small business ends up convinced that training it should stop doing is working.

How FirstHR Fits

FirstHR handles the layer around this: employee records so training completion lives with everything else about a person, plus document management so the responses from the last session are findable when you plan the next one.

Structured onboarding does the same job for training that new hires need, which then gets scheduled rather than remembered. The whole thing is built for small businesses at flat, predictable pricing.

What software does not do is write good questions or read the answers honestly. A six-question survey in any form tool works fine.

What a system provides is that the feedback does not evaporate: you can compare this session against the last one, you know who attended what, and the six-week follow-up sits on a task list with a due date instead of depending on your memory.

Frequently Asked Questions