A training feedback survey captures reaction at session end. What it measures, what it cannot, and how to design one that feeds real evaluation — with examples.
Feedback · Kirkpatrick Level 1
A feedback survey is the short questionnaire participants fill out at the end of a session. It tells you how the experience landed. It cannot tell you whether anyone learned, whether anyone applied what was taught, or whether the program produced its result.
This guide gives you the full feedback-survey craft for workforce training, professional development, course delivery, and grantee training series — a ready template, an 18-question bank, the six design rules, and the line where feedback ends and evaluation begins.
“How was today?” captures a feeling tied to no decision. “Which moment in your work this week will use what you learned?” names the moment — and seeds the 30-day follow-up that asks if it happened.
The Reaction Layer in Context
The Kirkpatrick model names four layers of training evaluation. A feedback survey lives at the bottom: it captures reaction. Each layer above answers a different question with a different instrument and cadence.
Level 01 · Reaction
The feedback survey lives here, captured at session end while the impression is sharp.
Instrument: feedback survey · session end
Level 02 · Learning
Measured by paired pre and post checks with identical wording so the delta is real.
Instrument: pre/post assessment · session start and end
Level 03 · Behavior
The trickiest layer because it lives off-platform in the participant's daily work.
Instrument: follow-up survey · 30 to 90 days post-program
Level 04 · Results
Jobs placed, retention, throughput, behavior at scale.
Instrument: outcome data join · 6 to 12 months out
A well-designed feedback survey owns Level 1 completely and points the program at the next layer. It cannot fill in the layers above.
A training feedback survey is a short questionnaire participants fill out at the end of a training session — five to ten questions, a two-minute completion target. It captures reaction: how relevant the content felt, how clear the delivery was, what the participant would change, and whether they intend to apply anything. It is the Kirkpatrick Level 1 instrument, run at session end, not the next day, because reaction is sharpest in the moment. The same instrument is also called a training survey, post-session survey, satisfaction survey, or smile sheet.
Runs at the end of a multi-week course. Adds questions that only make sense over a longer run: which module was most useful, which fell flat, how pacing held across weeks.
Runs at the end of a half-day or one-day workshop — one shot at the end of the day. Can ask about the day's overall arc because the experience is contained.
For internal L&D. The participant returns to a known job context, which lets the survey ask sharper intent-to-apply and manager-handoff questions.
A feedback questionnaire by any of these names shares the same structure; what changes is what it is paired with afterward — a learning check, a behavior follow-up, or nothing. Paired with nothing, it measures reaction and stops. For the layers above, see training evaluation.
Design Principles
Most feedback surveys ship as a list of questions copied from a template. The list is fine; the connections are missing. Six rules cover the connections that turn a list into a survey that improves the next session.
01 · Timing
In-room surveys at session end land 80–95% response. Emailed after the participant leaves, they drop to 20–40%.
A 35% response carries the views of those who chose to respond, rarely the median view of the room.
02 · Length
Drop questions that do not change a decision. Longer surveys produce drop-off and straight-through clicking, not better data.
A six-question survey everyone completes beats a fourteen-question one half the room finishes.
03 · Pairing
A 4.2 average across two cohorts can hide a fixable problem in one and a content gap in the other. The open-ended prompt tells you what produced the rating.
The open-ended response is what tells the next facilitator what to change.
04 · Identity
If reaction can't be linked to the same participant's pre/post delta and 30-day follow-up, the program produces three unconnected reports.
Unconnected instruments produce three reports that disagree, not one that explains.
05 · Intent
Satisfaction says participants did not hate the session. Intent-to-apply says which content has a chance of showing up in the work.
The application question is the only feedback question that predicts Level 3 behavior.
06 · Use
Naming what changed based on the last cohort's feedback raises the next survey's response rate. Use the feedback or stop running it.
A survey that clearly affects the next session raises response by 15–25 points.
Template & Question Bank
A ready training feedback survey: 18 questions across six dimensions. Use 6–10 of them, keep it under two minutes, and pair every rating with its open-ended counterpart. The same set works as a workshop feedback survey, a course feedback questionnaire, or a professional-development form — adjust the time framing.
Relevance to role
Clarity of delivery
Pace & structure
Usefulness
Intent to apply
Overall & identity
Same 18 questions, ready to send. Pick a format, or generate a connected version that links to pre/post and follow-up.
Design Choices
Most feedback surveys fail at one of six choices made before the first participant fills out a field. The choices compound: a wrong answer at row one makes row six harder to recover.
| The choice | Broken way | Working way |
|---|---|---|
| Where it's deliveredRoom or inbox | Email two hours later · 30% response | QR/in-room link · 85–95% before they stand up |
| How participants are identifiedReal ID or anonymous | Anonymous · three datasets, no joins | Personal link pre-fills the identifier |
| Question count5–10 or 15+ | 15–20 to "be thorough" · straight-through clicking | 6–10, each tied to a decision · 2 minutes |
| Open-ended pairingNumbers alone or with the why | Eight Likerts, no open-ended prompt | Every rating paired with "what produced this" |
| Application questionSatisfaction or intent | "Did you like it" · predicts nothing | Named work moment + barrier · seeds 30-day follow-up |
| What happens to the resultFiled or fed back | Folder nobody opens · next cohort gets the same session | Next session names two changes made · response climbs |
The first row controls every row that follows. A survey emailed two hours late lands at 30%, which makes the identity question moot, which kills the cross-instrument view, which leaves three disconnected reports.
Worked Example
A 240-participant workforce training program runs an eight-week cohort, four sessions per week, ending with a credential. Funder reporting needs end-of-program reaction, learning delta, and 90-day employment outcomes.
“Average satisfaction stayed at 4.3. We could not tell anyone what to change. We rewrote the instrument to cap at seven questions, pair every rating with one short why, and ask which work moment the participant would use the content in. The 30-day follow-up held at 71% because the participant had named the moment themselves.” — Workforce program lead
The feedback survey did not change shape — seven questions, two minutes, in the room. What changed was the layer underneath.
Program Contexts
The wording is similar; the structure underneath differs because the unit of analysis differs.
01
Per-session feedback across a multi-week cohort. A five-question survey at session end with one rotating open prompt keeps total load under ten minutes and makes the reaction trend legible. Pair with one end-of-program instrument and a 30-day employment follow-up.
02
Feedback must be comparable across workshops and grantee organizations. A foundation-owned template with two facilitator-customizable items makes cross-workshop comparison automatic; responses tag both workshop and grantee org.
03
The participant returns to a known job context, so the survey can ask sharper intent-to-apply and manager-handoff questions. The named work moment seeds a 30-day check sent to participant and manager.
Generic form tools collect a feedback survey well. The architectural gap shows up when the same participant answers a pre-survey, a feedback survey, a learning check, and a follow-up — four files with no clean join. A shared participant identifier is what makes the cross-instrument view possible.
A short questionnaire given to participants right after a session to capture reaction — five to ten questions covering how useful the content felt, how clear the delivery was, and what they would change. It measures Kirkpatrick Level 1 only and cannot prove learning, behavior change, or results.
Six items cover most needs: relevance to role, clarity, pace, the most useful content, what to drop, and intent to apply. Mix a five-point rating with one open-ended counterpart, and add one identity question that ties the response to the same person across pre-survey, feedback, and follow-up.
Ask what the participant will use first, what they would drop, what felt unclear, and how confident they feel applying the content. Tie each item to a decision. Avoid generic satisfaction. Cap at six to ten questions to protect the response rate.
Five to ten questions, completable in two minutes. Longer surveys produce drop-off without better data. If you need more depth, split the instrument: feedback at session end, learning check the next day, follow-up at thirty days.
In-room at session end lands 80–95%. Emailed after the participant leaves drops to 20–40% within forty-eight hours. The biggest lever is timing: capture before they leave the room.
Feedback measures reaction; evaluation measures change. A training evaluation tracks the same participants from pre-survey through follow-up to measure learning, behavior, and results. A feedback survey is one component. See training evaluation.
A starter set of six to ten questions covering relevance, clarity, pace, most useful content, what to change, and intent to apply. The actual survey adds a participant identifier, ties each item to the decision it feeds, and pairs ratings with one open-ended counterpart.
No. It captures reaction at session end. To know whether participants learned, applied it, or whether the program produced the funder's result, you need a learning check, a behavior follow-up at 30–90 days, and a connection to the same participant across all three.
A working session
Most teams already have a feedback survey running. The question is what happens to the data after collection. Bring the form you send, the question bank you copy from, or the funder report you fill in. We walk through how the same participant's reaction, learning delta, and follow-up sit in one record.
