A questionnaire can look perfectly reasonable to the team that wrote it and still be difficult to answer. “How often do you share reports?” sounds straightforward until one person counts sending a link, another counts presenting a slide deck, and a third wonders whether forwarding someone else’s report qualifies. The resulting numbers describe several different activities under one heading.
Questionnaire design is the work of finding those ambiguities before they reach the sample. It covers wording, answer options, sequence and presentation, whether people complete the questions themselves or an interviewer administers them. A good instrument makes the intended question clear while leaving room for an answer the team may not expect.
Work backwards from the decision
Imagine a fictional team deciding whether to improve report sharing or build more export formats. Asking customers to rate both ideas might establish a preference among the people who reply, but it would reveal little about the difficulty they encounter today. Questions about the last report they shared, its recipient and the steps involved would give the team something more useful to investigate.
Before writing the form, sketch the analysis you expect to do. If you need to compare occasional and frequent users, you need a defensible way to distinguish those groups. If an answer would make no difference to the decision, consider leaving the question out; a survey is rarely improved by collecting information simply because it might become interesting later.
Give each question an answer people can honestly choose
Ask one thing at a time. Someone may find a product fast but unreliable, so “How satisfied are you with its speed and reliability?” leaves them negotiating a compromise that the researcher cannot interpret. A specific reference period also helps: “In the past seven days” is easier to interpret consistently than “usually”, provided the activity is something people can reasonably remember over that period.
Answer options deserve the same attention as the wording. Overlapping ranges make classification uncertain, and a missing “I haven’t used this” option can turn inexperience into an apparent negative opinion. “Neither satisfied nor dissatisfied”, “don’t know” and “not applicable” express different circumstances; combining them conceals distinctions that may matter later. The Government Analysis Function’s questionnaire design guidance provides a useful reference for reviewing wording, response categories and testing.
Read the questionnaire as a sequence
Each question arrives in the context created by those before it. A long section about failures may influence a later overall rating, while an early list of possible benefits may supply the vocabulary for an ostensibly unprompted answer. Put questions that need an unprompted response before related lists, and keep the sequence stable when measuring change over time.
Routing should spare people questions they cannot answer. Test every route, including skipped and optional answers, and read the survey on a phone as well as a large screen. A grid that looks economical in the editor can become a laborious exercise in scrolling on the device a participant actually uses.
Test understanding before collecting the full sample
A pilot should examine how people arrive at an answer, as well as how long completion takes. Ask a few people from the intended audience to explain what they understood a difficult question to mean. Then inspect their responses as if the main study had finished: can you actually perform the planned analysis?
This is where open-ended questions can expose missing categories and where apparently harmless repetition can reveal survey fatigue. Once the instrument works, Ballpark’s surveys provide a way to collect the responses. The quality of the evidence will still depend on the questions, the recruitment and the care taken when interpreting who answered.