An empty text box can make a survey feel generous: here, at last, is somewhere to say what the prepared options missed. It can also leave someone staring at “Any other feedback?” with no idea where to begin. Giving people freedom to answer is useful only when the question gives them a reason, and enough direction, to use it.
Open-ended questions let participants formulate a response in their own words. In an interview, that response can become a conversation; in a questionnaire, the wording has to do more of the work because there may be nobody available to clarify it.
Make the invitation specific
“Tell us about your experience” is open, but it asks the participant to choose the subject, decide the level of detail and reconstruct an unspecified period of time. “What happened the last time you tried to share a report with someone outside your team?” gives the answer somewhere to start without supplying a verdict.
The distinction matters because open wording can still lead. “Why did you find the dashboard confusing?” assumes confusion, while “What, if anything, was difficult about using the dashboard?” allows the person to reject that premise. When the research concerns an actual event, ask about what happened before moving to an explanation. People can describe a sequence of actions more readily than they can always account for the motives behind it.
During user interviews, a brief follow-up such as “What happened next?” can keep an account moving without steering it towards the researcher’s preferred explanation. In a survey, a concise prompt asking for one example may work better than a bundle of several questions inside one text box.
Leave room for answers you did not anticipate
Open responses are particularly useful when the team does not yet understand the possible reasons for a behaviour. In a fictional cancellation study, a prepared list might cover price, missing features and poor performance while overlooking that the customer’s project has simply ended. An unprompted account can reveal that omission before it becomes a misleading chart.
The order of questions affects this opportunity. If respondents see a list of proposed explanations first, their later account may borrow from it. Ask the open question first when spontaneous language matters; ask it after a rating when the purpose is specifically to understand that rating. Those are different research aims, even if both use the same text field.
Plan for the answers you will receive
Free text takes work to produce and work to interpret. Pew Research Center’s analysis of open-question nonresponse examines how the demands of different questions affect whether people answer. A blank field is not necessarily evidence that someone has nothing to say.
Keep open questions purposeful, and make sensitive elaboration optional where appropriate. When analysing responses, retain enough context to distinguish a complaint about the product from a complaint about the situation in which it was used. Grouping both under “frustration” may be tidy, but it leaves the team without a useful explanation.
If you report how often a theme appeared, state the denominator and how responses were coded. Several themes may occur in one answer, and people who wrote a comment may differ from those who skipped it. A vivid quotation can illuminate a finding; it cannot, by itself, establish how common the experience is.