Survey questions from interviews: turning qualitative hypotheses into questions
Write the hypothesis without quoting anyone, define an observable indicator, choose a neutral question and then draft the response options. Each question should test one idea.
How to turn interview observations into a survey
Write the hypothesis without quoting anyone, define an observable indicator, choose a neutral question and then draft the response options. Each question should test one idea.
A transcript is a verifiable source of wording. It preserves context, qualifiers and the order of the reasoning. The author still needs to decide on the final wording and test it with participants.
Build the survey step by step
Collect hypotheses
Identify recurring problems and alternative explanations. Keep a timecode beside each one.
Define an indicator
Decide what behavior, frequency or choice respondents can report.
Write neutrally
Remove cues about the preferred answer, double-barreled questions, jargon and loaded words.
Run a pilot
Give the survey to a few people, ask them to explain what the questions mean and fix ambiguities.
Example: a survey based on interviews about document approval
This is a fictional training example. The names, circumstances and outcomes are included solely to explain the method.
“Sometimes I open five chats to find out who’s already reviewed the contract.”
Hypothesis: participants lose track of the approval status. This does not show how common the problem is. A testable question is “Where do you usually check a document’s current status?” with mutually exclusive options and an “Other” response.
Avoid “Are you also annoyed by endless chats and lost documents?” This suggests an emotion and combines two problems. Use a separate question with a time window to measure frequency, such as “In the past 30 days, how many times…?”
| Element | Weak version | Working version |
|---|---|---|
| Problem | People lose everything | People look for the document status in several channels |
| Period | Usually | In the past 30 days |
| Action | Do you like it? | Where did you check the status last time? |
| Options | Yes / no | List of channels + Other + Did not check |
Survey question template
Copy the template and fill it in using a specific recording. Label guesses as hypotheses.
Hypothesis from the interview: Observable indicator: Recall period: Question wording: Response options: How we tested understanding:
Four checks before sending
- The question tests one idea.
- The wording does not suggest a preferred answer.
- Options do not overlap and include “Other” where appropriate.
- Questions about behavior name a period or a specific situation.
A closer look
Can I use a quote in the survey?
Sometimes, as a response option. Remove details specific to the person first, then check whether others understand the wording.
How many questions should I ask?
As many as you need to make the decision. Remove questions whose answers cannot change what the team does.