Start with the decision.
Before adding a question, ask how its answer will change the next step. “What prevents attendance?” may influence delivery. “How useful was the explanation?” may influence teaching. Collecting a detail merely because it is available adds work and may add data protection obligations.
Use one topic, a clear time frame and a concrete experience. “Recently” can mean different periods to different people; “in the last 30 days” gives a shared reference.
Rewrite common problems.
| Problem | Better starting point |
|---|---|
| “How clear and useful was the trainer?” | Ask separately about clarity of explanation and usefulness of the content. |
| “How helpful was our excellent programme?” | “How useful was the programme for the task you wanted to do?” |
| “Do you often use spreadsheets?” | “On how many days in the last 30 days did you use a spreadsheet?” |
| A delivery rating shown to everyone | Ask whether anything was delivered, then show the rating to the relevant customers. |
Give the options the same care.
Use non-overlapping categories and label the scale. Include “not applicable”, “do not know” or a refusal option when they serve different meanings. Do not turn these into the midpoint or a zero score.
For multiple-choice questions, explain whether “none” is exclusive. Check for selections that contradict each other. For an open question, avoid requesting identifying details the research does not need.
Choose the answer format for the evidence.
Single choice identifies one category. Multiple choice captures several applicable options. A rating asks about degree.
An open text or voice answer lets someone explain an experience in their own words. A scored test item provides different evidence from self-reported confidence.
Nuvopoll’s editor supports the available question formats and conditional paths. AI can prepare a draft, but the researcher checks the wording, options and method.
Pilot with people, including in every language.
- Ask someone to explain what they think the question is asking.
- Check whether a likely answer is missing from the options.
- Walk through conditional paths and the mobile experience.
- Review translated concepts with local partners.
- Remove questions that do not support a research decision.