Seminar Feedback Form Template
A 14-question seminar feedback form you can take right now, for a talk with one or two speakers and an audience that came to listen. The two judgements of the speaker are held back to the end on purpose. Read the seminar feedback questions below, copy them, or open the template in the editor.
- 14questions
- About 4 minto complete
- 5-point agreeone scale throughout
- No name askedno identity field
A seminar score is mostly a score for one person
A seminar hands an audience one thing to judge: somebody talking. That makes it the easiest kind of feedback to collect and the easiest to misread, because a single impression of a speaker will fill in every other box on the form.
The same lecturer, rated twice. In a study run at Michigan, 118 students watched a videotaped interview with one college instructor. Half saw him warm and friendly, half saw him cold and distant; the man, the accent and the mannerisms were the same in both. The students who saw the warm version rated his appearance, mannerisms and accent as appealing, and the students who saw the cold version rated the same attributes as irritating.[1] Asked afterwards, they said their overall impression had not affected those ratings at all.
What that does to a feedback sheet. Hand an audience seven boxes after one talk and you are not collecting seven measurements. You are collecting one, copied out seven times with small variations. It is why a form that rates content, delivery, organisation, venue and refreshments so often comes back with every item within half a point of every other item.
So this form is ordered against the halo rather than around it. The concrete items go first, while nothing has been said about the person yet: could you hear it, did it start on time, could you follow the argument, was there time for questions. The two judgements of the speaker are the last things on the form. That will not remove the effect, and nothing will, but it stops the person-rating from being the frame that everything else is answered inside.
| What you ask an audience | Can they report it? | What the answer is really made of |
|---|---|---|
| Could you hear and see it | Yes. It is a fact about where they were sitting. | A fact. Treat a low score as a room problem and go and look at the room. |
| Did it start and finish on time | Yes. They watched the clock. | A fact, and one you can check against your own running order. |
| Could you follow the argument | Yes, for themselves. Not for anybody else. | A judgement about their own experience, which is the kind people are good at. |
| Can you use any of it | Partly. They can say what they intend to do. | An intention on the day. Question 11 is what makes it checkable. |
| Was the speaker good | They will answer, confidently, every time. | One overall impression, spread across whatever boxes you give them.[1] |
| Did you learn anything | No. Not reliably, about themselves. | A feeling of fluency. It is why a rating and a test disagree.[2] |
The column that matters is the third one. An audience is a reliable witness to its own experience and an unreliable witness to its own learning. Across ninety-seven studies matching teaching ratings against what students actually learned, the correlation was near zero once the researchers had no stake in the rating system.[2] Those were semester-long courses with exams at the end: if a full course cannot get a rating to track learning, an hour will not either.
What to do instead of giving up. Question 11 asks for one thing the respondent is taking away, and it is the only item that produces evidence rather than a score. When most of an audience can name something and the answers agree with each other, the talk landed. When the box is empty on half the replies and the ratings are still fours and fives, you ran an enjoyable hour.
Three things get called a seminar and want a different form. If the room worked through exercises rather than listened, the workshop feedback questions fit that session. If it ran online, the webinar feedback questions ask what the attendance figures cannot. If it had an agenda and sessions to choose between, that is a conference feedback survey and a different unit entirely.
The 14 seminar feedback questions, in the order they should be asked
Eight ratings on one 5-point agree scale, two questions about who is answering, two open answers, and the speaker held back to the end. Each item says what a low answer to it actually means. Copy one block or the whole set.
The seminar as a whole
Two items that survive on their own if somebody answers nothing else.
- Overall, this seminar was worth the time it took.The one answer to protect. A high score here beside a low content score is the pattern to watch: a room that enjoyed itself and left with nothing.
- I would come to another seminar run by these organisers.Intent is harder to be polite about than satisfaction. If it sits below the overall rating, something specific put people off and the open answers will name it.
The content
What was said, kept apart from who said it.
- The talk told me something I did not already know.The item that separates a good talk from a familiar one. An audience of specialists will mark this down on a talk pitched for beginners, which is a billing problem rather than a speaker problem.
- I could follow the argument from start to finish.Structure, not intelligence. A low answer here from people who scored the speaker highly points at the running order or the slides.
- There is something here I can use in the next month.The nearest thing to a use test that an audience can answer honestly on the day. Read it against question 11, where they say what the thing actually was.
The room
Whether people could take part at all. Nobody rates a talk they could not hear.
- I could hear and see clearly from where I was sitting.Sight lines, sound and captions in one construct. Anything under a 4 belongs next to the open answers, which is where the reason it failed somebody gets named.
- The seminar ran to the times advertised.The cheapest thing on the list to fix and the one people remember. When a talk overruns it is the time for questions that gets cut, so read this item beside question 8 rather than on its own.
- There was enough time for questions from the floor.At a seminar this is the only moment the audience gets to speak. It is regularly the lowest item on the sheet and almost never on the agenda.
Who is answering
Two taps that decide how much weight any other answer carries.
- How much of the seminar were you there for?Somebody who caught half of it is rating half of it. Report the full-attendance group separately before you average anything.
- Was this your first seminar with these organisers?First-timers and regulars are rating different things. Regulars mark novelty down because they have heard the groundwork before.
In their own words
Two open answers, placed before the end, while people still have a reason to type.
- What is the one thing you will take away from this seminar?The most useful box on the form. If most people cannot name one thing, the ratings above are measuring an enjoyable hour.
- What would have made this seminar better?One thing, not any comments. A single question gets a specific answer where an open invitation gets a shrug.
The speaker, asked last
Held to the end so the concrete answers are given before the judgement of a person is made.
- The speaker knew the subject.Command of the material. It moves independently of the next item, which is the whole reason for splitting them.
- The speaker held the room.Delivery. High here with a low score on question 3 is the pattern to watch: a room that was held and taught nothing.
Nothing here asks for a name. Two of the seminar templates you will find elsewhere open with name, email and phone, which is three things to fill in before anyone has been asked what they thought. If you need a contact detail, ask for it in the middle and make it optional. The last thing on this form is a rating, because a rating is the cheapest question to answer once somebody is nearly finished.
One scale, not five. Every rating is the same 5-point agree scale running the same direction, so a respondent learns it once and the eight answers lie side by side without conversion. Mixing an Excellent-to-Poor scale, a satisfaction scale and a yes-or-no inside one short form costs you the only comparison worth having: your own items ranked against each other.
Two questions that change how the rest is read. Questions 9 and 10 cost one tap each and they do more work than any rating. Somebody who was there for less than half of it is rating less than half of it, and a regular is marking novelty against four previous seminars while a first-timer is not. Split on both before you average anything.
Getting it in front of people. A seminar ends and the room stands up at once, so the only reliable moment is the one before that: a short link or a printed code on the closing slide, while people are still sitting down. All fourteen questions are already written, so open them in the free online survey tool, change any wording that does not sound like you, and have the link ready before you start.
Two jobs this form deliberately does not do. Questions 13 and 14 rate the speakers as a bill and cannot tell you which of two was the problem; when you need a scorecard per named person, and the confidentiality that goes with one, use the speaker feedback questions instead. And if you want to know what the room hoped to hear rather than what it thought afterwards, the pre event questionnaire ask that before the day.
How many replies before the average means anything
Forty people came, eleven answered, the mean is 4.2. Nobody on this search will tell you what that number is worth, so here is the honest answer: at eleven replies it is not a measurement, and it should not be reported as one.
| Replies you have | What the average is | What to do with it |
|---|---|---|
| Under 10 | A list of opinions wearing a decimal point. | Read them one by one and quote them. Do not report a mean, and do not compare it with anything. |
| About 10 | Unstable. In the study behind this table, dependability at ten responses was 0.778.[3] | Report the split, not the mean: how many chose 4 or 5. Say how many people answered, every time. |
| 20 to 30 | Usable for teaching-style items. Around twenty raters on a dozen items reached 0.87.[4] | Rank your own items against each other and fix the lowest. Still no decimal places. |
| 100 or more | Stable. Dependability reached 0.938 at a hundred responses.[3] | Trend it against your own last seminar, same questions, same wording. |
Where those numbers come from. One Ghanaian university put its teaching-evaluation system through a generalizability analysis: 2,553 students across 145 courses, at an overall response rate of 48.4 per cent. Dependability sat at 0.938 when a hundred students answered and fell to 0.778 when ten did.[3] A second analysis of the same programme found that adding raters bought far more dependability than adding items, and that twenty raters on a dozen items was already enough.[4] These are university courses, not seminars, so take the shape rather than the rate. The shape decides the form: keep it short and spend the effort on getting more people to finish it.
Report the split, not the mean. Say how many of your respondents chose 4 or 5, and say how many people answered, in the same sentence. A share moves when opinion moves and it does not hide a room that was divided; a mean of 3.5 looks identical whether everybody was lukewarm or half the audience loved it and half walked out. And name the denominator: eleven of forty is a different finding from eleven of twelve.
What to compare it with. Your own last seminar: same questions, same wording, asked at the same point. There is no published norm for seminar feedback, and any vendor number you find describes that vendor's customers rather than your audience. Inside one form, rank the eight ratings against each other and fix the lowest. If the seminar was really a course that ran over weeks, the course evaluation survey template has the items for that instead.
Seminar feedback form FAQ
What questions should a seminar feedback form ask?
Fourteen here: two on the seminar overall, three on the content, three on the room, two on who is answering, two open answers, and two on the speaker. Where they sit matters as much as what they say. The concrete items go first and the judgement of the speaker goes last, because a general impression of a person quietly rewrites the ratings of everything else about them.
How many questions should a seminar feedback form have?
Fewer than you think, because getting more people to answer is worth more than getting more answers from each of them. In a generalizability study of teaching evaluation, adding raters improved dependability far more than adding items did, and about twenty raters on a dozen items was enough.[4] A short form finished by thirty people beats a long one finished by eight.
Should a seminar feedback form be anonymous?
It asks for no name and no email, so nobody can be identified from the answers alone. Name the reader before people answer, out loud from the front: whoever booked the room is the person who will see this. Then promise nothing beyond that, because the promise is easy to break by accident. In an audience of fifteen, one extra question about which department somebody came from will undo it.
What is the difference between a seminar feedback form and a seminar evaluation form?
Only the noun. A sheet handed round the room is usually called an evaluation form and a link is usually called a feedback form, and the questions are the same either way. This one is a link that works on a phone while people are still in their seats, which is the job the paper sheet used to do.
Does a high seminar rating mean people learned something?
No, and it is worth being clear about that before you report one. Across ninety-seven studies matching teaching ratings to what students actually learned, the relationship was close to nothing once the authors had no stake in the rating system.[2] A rating tells you whether an audience would come back. Question 11, where they name what they are taking away, is the cheapest evidence you have that anything landed.
Methods and sources
What this template is based on
Two arguments, four measurements. The ordering of the form comes from the halo experiment on a videotaped lecturer[1] and from the meta-analysis of teaching ratings against learning.[2] The advice on how many replies you need comes from two generalizability analyses of an institutional teaching-evaluation dataset.[3][4]
Question licensing
Seminar feedback has no standard or licensed instrument, so every question here is ours to copy, edit and translate without asking anyone. Nothing on this page reproduces or paraphrases a proprietary training-evaluation model, and no named framework is used to structure the form.
Limits and disclosure
None of the four sources measured seminar attendees: three concern university teaching evaluation and one is a laboratory experiment with 118 students watching a videotape. Each is cited for direction rather than as a rate, and the dependability figures come from one institution. There is no published benchmark for seminar feedback and this page does not invent one.
SuperSurvey makes the editor this template opens in. Everything on this page is free to use with it or without it.
References
- Nisbett, R. E. and Wilson, T. D. "The halo effect: Evidence for unconscious alteration of judgments." Journal of Personality and Social Psychology 35, no. 4 (1977): 250-256. Summary and abstract: explorable.com
- Uttl, B., Cnudde, K. and White, C. A. "Conflict of interest explains the size of student evaluation of teaching and learning correlations in multisection studies: a meta-analysis." PeerJ 7 (2019): e7225. pmc.ncbi.nlm.nih.gov
- Dzakadzie, Y. and Quansah, F. "Modeling unit non-response and validity of online teaching evaluation in higher education using generalizability theory approach." Frontiers in Psychology 14 (2023): 1202896. frontiersin.org
- Dzakadzie, Y. "Identifying optimal number of raters and items for dependable student evaluation of teaching: evidence from generalizability theory." Frontiers in Education 11 (2026): 1814494. frontiersin.org
Have the link on the closing slide
Fourteen questions, about four minutes. Open it in the editor, change any wording, and print the code on the last slide before you start.
Use this template