How to write a feedback form

To write a feedback form, decide the one thing the answers will change, ask for an overall rating first because it is the easiest question to answer, add a small number of open questions where they will be most honest, leave names and emails to an optional last question, and keep the whole form short enough to finish in about two minutes. Then tell people, on the form, who reads what they write. That last sentence does more for honesty than any wording trick.

One goal

Write the goal down before any question: “the answers will decide X”. The event feedback form for a fictional one day marketing conference puts its goal on the welcome screen, where respondents can see it: the session ratings decide next year’s programme, and the free text is read by the two people who planned the day.

Everything on the form serves that goal. Session ratings feed the programme. The open questions feed the planners. “Would you come next year?” is the question the organisers actually face. Nothing asks about the sponsor’s booth or the hashtag, because nobody would change anything based on the answer.

When two teams want different things, split the form. The course evaluation form from a fictional training institute shows the same discipline inside one form: three separate rating grids for the teaching, the materials and the outcome, so each average lands with the person who can fix it.

Rating first

Open with the easiest question somebody can answer. An overall rating has no wrong answer, takes one tap, and gets people started. The event form’s first screen asks “Overall, how was the day?” with four labelled options from excellent to poor.

Rating scales come in two broad families. Nielsen Norman Group describes Likert items, which measure agreement with a statement, and semantic differentials, which place an answer between two opposite adjectives. It warns that agreement scales are vulnerable to acquiescence bias, the tendency to agree, and it recommends sticking to established labels such as strongly agree through strongly disagree rather than inventing new ones.

When several things share a scale, use a grid rather than a stack of separate questions, and give every row an honest exit. The event form’s session grid offers Skip, and its description tells people to use it for anything they missed, so a talk somebody missed is recorded as missed rather than as a polite middle score. The course evaluation’s materials grid does the same with “Did not use”. Without that exit, the portal nobody opened scores poor and gets scrapped for the wrong reason.

Open question placement

Open questions produce the answers you could not have listed, and they cost effort. Nielsen Norman Group notes that a survey with many open questions will usually have a lower completion rate than one with more closed questions. So use few, and place them on purpose.

Placement changes what you get. Pew Research Center reports that when closed questions on a topic come before an open question, respondents are much more likely to mention the ideas those earlier questions raised. That gives you a clear choice. If you want the reason behind a score, ask the open question straight after the rating, which is what the customer feedback survey does with “What is the main reason for your score?”. If you want people’s own priorities, unprompted, ask the open question before the grids.

The event form pairs its open questions: “What was the best part?” then “What would you change?”, both required, side by side. The organiser reads a keep list and a change list of equal weight, instead of one comments box where praise crowds out the complaint that matters. The difference between the two question types is set out in the open against closed questions note.

Anonymity

People write more candidly when nothing has asked who they are. Nielsen Norman Group points to research showing that asking for names and other identifying information increases social desirability bias, so the cheapest honesty gain on a feedback form is to leave identity out, or put it last.

The event form asks for a name and email only on its final screen, labelled as being for anyone who wants a reply. A response stores the answers, the form version and the time it was submitted, so without that last field nothing records who answered.

Anonymity can still leak through small groups. The employee pulse survey asks only which team somebody is in, says it is the only segment recorded, and promises that a team with fewer than five responses is never reported on its own. If your feedback form segments by team, session or class, make the same promise and keep it. And remember that email alerts and webhooks carry every answer to wherever you send them.

Length

Short is not a style choice for feedback; it is the response rate. The event form is eight questions with a two minute estimate. The course evaluation is nine questions in about three minutes. The customer feedback survey is six questions in ninety seconds. The pulse survey is six questions in under two minutes.

A conversational form helps here, because one question per screen makes eight questions feel like eight small asks. It also prints a short description under each question, which is where the event form says to use Skip for sessions somebody missed.

For more question ideas without the length problem, the workshop feedback question bank holds twenty questions in four sections and says plainly that about ten is what a short workshop should send. Treat a bank like that as a menu. Pick the questions that serve your one goal and leave the rest for another round.

A template you can copy

The table at the end of this article sets out the event form as a template: all eight questions, the component each uses, whether it is required and why it is on the form. Swap the session grid for whatever your event or service has several of, keep the pair of open questions, keep identity last, and resist adding a ninth question until you can name the decision it serves.

Common mistakes

No exit on a grid. Every rating grid needs a Skip, not applicable or did not use column, because a matrix is required or optional as a whole.

Asking for a name at the start. It costs candour on every answer that follows.

One comments box for everything. Split it into what to keep and what to change.

A new scale every round. If you compare this term with last term, the statements and the points must match. The pulse survey keeps the same six statements every month for that reason.

Wording faults. Leading and double barrelled questions ruin feedback as quickly as they ruin any survey. The survey question writing post covers those faults with examples, so they are not repeated here.

Build it

The rating matrix page shows the grid component: rows of statements against shared columns, with the selection type set to one answer per row or several. Around it, a feedback form usually needs radio buttons for the overall rating, long text for the open questions, a dropdown for a segment and a short text for an optional email. All of those are available on both classic and conversational forms.

When answers arrive, the tutorial on viewing, exporting and managing responses covers the table and the CSV. The export has one column per field and joins a whole matrix into one column as statement and answer pairs, so plan to split that column before averaging. For a live feed, the email alerts add-on sends each response to one address, and a webhook posts it once, with no retry.