Conference Feedback Form: The Questions That Change Next Year

TL;DR

• Most conference feedback forms measure mood, not decisions. Ask about choices you will make again: session length, running order, room format, who to book.
• Ask in the room before people leave. Web surveys sit about 12 percentage points below other modes on response rate.
• Ten questions is the ceiling. That is roughly five minutes, and completion drops past seven or eight.
• A 1 to 5 score on overall satisfaction is unusable on its own, and one open question beats three.

I have hosted more than 800 events, and I can usually guess an organiser’s feedback form before I see it. Nine questions, eight of them rating scales, one box marked “any other comments”. It comes back with a 4.2, everyone is relieved, and nothing about next year changes. A conference feedback form is only worth the room’s time if the answers point at a decision.

The test for each question is blunt. If the answer came back tomorrow, what would you do differently? If you cannot name the action, cut it.

What should a conference feedback form actually ask?

Ask about the decisions you will make again: session length, running order, the plenary and breakout balance, room format, which topics to repeat, and who to put back on stage.

Those are the levers an organiser controls. “Did you enjoy the day” is not a lever. “Were the afternoon breakouts the right length” is, because the answer sends you to 30 minutes or keeps you at 45.

Build the form backwards from next year’s agenda. Write down the five decisions you are dreading, then the question that settles each one. You will land on six or seven, about the right length anyway.

Why does a 1 to 5 overall satisfaction score tell you nothing?

Because it compresses a whole day into one number. That number almost always lands between 4 and 4.5, and you cannot act on it.

Two things make it worse. People judge an experience by its most intense moment and its ending: a 2022 meta-analysis of 174 effect sizes found the peak-end effect was large (r = 0.581), while the length of the experience barely registered. A strong closing speaker therefore lifts the score for the whole day, including the session that quietly died at 2pm.

I have watched rooms rise and fall through the day for twenty years. One overall score flattens all of that into whatever people remember last.

The scale is doing less than you think, too. Three points is too few, with reliability problems documented by Jeff Sauro in August 2019, while five, seven and eleven points perform much alike. The UK Government Analysis Function’s questionnaire design guidance, published 14 March 2023, asks for at least five points and a neutral midpoint.

Keep one overall score if the board wants a trend line. Just do not pretend it told you anything.

When should you send a conference feedback form?

Ask in the room, before people pick up their coats. Anything by email afterwards is a smaller, slower, politer version of the same survey.

A 2020 meta-analysis of 114 experimental comparisons found web surveys ran 12 percentage points below other modes on response rate. Explori, an event research firm writing in March 2021, puts a good post-event response for conferences at 10 to 20 per cent, with two reminders inside seven days.

There is a trade. In the room you get volume, but a warmer answer, because the person in the next seat can see your screen. By email you get fewer replies, skewed to the delighted and the furious, but franker ones.

So split it. Three or four questions in the room while the stage is warm, then one short email next morning with the open question. Do not send the same form twice.

How many questions do you get away with?

Ten, and seven is better. Ten questions take about five minutes.

SurveyMonkey’s analysis of 100,000 surveys deployed between January 2009 and September 2010 found drop-off rose fastest across the first 15 questions. Its completion times work shows completion falling by 5 to 20 per cent past seven or eight minutes, and people spending roughly half as long per question on long surveys. That second finding should worry you most: a 25-question form gives the same data, answered with less care.

The rewrites that make a weak question useful

Most bad questions are not badly worded. They ask about a feeling where they could ask about a decision.

Two patterns there. Split anything asking about two things at once: the Analysis Function guidance says double-barrelled questions should be separated, and “venue and catering” is two questions wearing one coat. Then swap a rating for a behaviour: “Name one thing you will do differently” cannot be answered on autopilot.

Avoid leading questions, too. “How much did you enjoy the keynote” has decided that you did.

How do you ask about individual speakers without a popularity contest?

Ask what the session was for, not whether the speaker was likeable. Warmth is what a rating scale picks up, and warmth is not value.

Higher education has been round this loop. A meta-analysis in Studies in Educational Evaluation, volume 54, September 2017, found student evaluation ratings and actual learning not related. Treat a 4.8 for your funniest speaker with the same caution.

Three questions better than a star rating:

• Was the session pitched at the right level for you (too basic, about right, too advanced)?
• What is the one thing from this session you will use?
• Would you want this topic again next year, with a different speaker?

That last one separates the subject from the person, which is what a programme decision needs. For a comparative read, ask delegates to pick the two sessions they would keep rather than scoring all eight. Forced choice beats a grid of scales, and it is kinder to the speakers who miss out.

The one open question that earns its place

You get one. Put it at the end, make it specific, and accept that plenty will skip it.

Open questions cost you answers. Pew Research Center reported in October 2021 that item nonresponse on its panel averages 18 per cent for open-ended questions against 1 to 2 per cent for closed ones. They also change what you hear. In Pew’s survey questions guidance, 58 per cent picked the economy from a closed list where only 35 per cent volunteered it unprompted.

So “any other comments” wastes the one slot you have. Use something that points at the decision: “If you were running this day, what would you change first?” It is specific, it gives permission to criticise, and the answers arrive sorted into things you can fix.

How can the host prompt completion without nagging?

From the stage, once, with a reason and a deadline the room can see. Not three reminders and a prize draw.

This is the part organisers hand to a host and never brief. What works is naming the decision out loud: “There are two questions on your phone. The first decides whether next year’s breakouts are 30 minutes or 45. I would like it before you stand up.” That is a different ask from “please fill in our survey”.

A few things that help:

• Put it where attention already is. If the room has used audience response systems or live Q&A all day, the feedback form belongs on that screen, not in a new email. I take live questions through my own software, RoomReader, so the last question of the day lands where people have already been.
• Do it before the final applause. Once the closing remarks land, the room is standing up.
• Keep it to 60 seconds of stage time, and never apologise for asking. A host who sounds embarrassed teaches the room the form does not matter.
• Report back next year. “You told us the breakouts were too short, so they are 45 minutes” beats any incentive.

People take part when the ask is worth their time and answering costs them nothing in front of others. That holds for a feedback form as much as for audience engagement in the main room.

What to do with the answers

Read the open answers first, before you average anything. They tell you which numbers matter.

Then pick three changes and put a name against each. Most feedback exercises die not because the data was bad but because nobody owned the next step.

Keep the form stable year on year, too. The comparison is where the value sits. Reading the room tells you more than any survey, but the survey is what you hand to a budget holder.