Write Voice-First Questions That Get Real Stories

In voice interviews, people cannot highlight text, scan back up, or re-read. Your question has to work the first time they hear it.

Survey research explains why this matters: people must understand the question, retrieve the right memory, decide what they believe, and then respond. In voice, there is an extra step; they also have to plan and articulate speech in real time. If your wording is vague or abstract, different people interpret it differently, and your results become mush.

The Voice Question Blueprint

Write each main question using this structure:

1. Time anchor: "Think about the last time…"
2. One clear ask: "What made you decide to…?"
3. Guardrail: "Give one example." / "Main reason." / "In a sentence or two."
4. Neutral follow-on (optional): "What happened next?"

This blueprint works because it reduces cognitive load at every step: the time anchor helps retrieval, the single ask prevents confusion, and the guardrail tells participants what kind of answer is helpful.

Time anchors are the easiest upgrade

Avoid "recently." Instead say "in the past week" or "the last time." Vague time periods mean different things to different people, and your data becomes incomparable.

• Bad: "How do you usually handle onboarding?"
• Better: "Think about the last time you onboarded a new tool. What was the first thing you did?"

Guardrails prevent rambling without leading

You are not telling participants what to say. You are helping them know what kind of answer is helpful.

• "One sentence is fine."
• "Start with the main reason."
• "Focus on the first moment you hesitated."

If your question has two verbs, it is usually two questions

"What did you like and what would you change?" becomes:

• "What worked well for you?"
• "What would you change?"

This is the "one question per turn" rule, arguably the most important rule in voice interview design. If you want two ideas, write two questions.

Keep it short

Write prompts that can be spoken in one or two breaths. Long prompts create high cognitive load, and participants lose track of what you are actually asking.

Bad-to-better examples for marketing studies

Messaging test

• Bad: "Do you like this tagline and why?"
• Better: "When you hear this tagline, what do you think it is promising? One sentence is fine."
• What improved: One task instead of two. Avoids "like" (which is vague). Focuses on meaning, which is more actionable.

Pricing and value

• Bad: "Is the price too high?"
• Better: "Think about the last time you saw the price. What did you do next, and what was going through your mind?"
• What improved: Time anchor plus behavior. Avoids leading with "too high."

Churn and switching

• Bad: "Why did you churn?"
• Better: "What was the main reason you stopped using it? If there were several reasons, start with the biggest one."
• What improved: Removes blame. Sets a "main reason" guardrail.

Competitive positioning

• Bad: "What is better about us than competitors?"
• Better: "When you compare options, what is the one thing you wish this product did better?"
• What improved: Avoids "us is better" bias. Elicits specific improvement targets.

Onboarding

• Bad: "Tell me everything about onboarding."
• Better: "Think about your last sign-up. What was the first moment you hesitated or almost gave up?"
• What improved: Episode-based. Focuses on a high-signal moment instead of a vague "everything."

Exercise 1: Rewrite three questions using the Blueprint

Take these weak questions and rewrite each one with a time anchor and a guardrail:

1. "Describe our product."
2. "Why did you not convert?"
3. "What do you think of our pricing?"

Exercise 2: Find the hidden double-barrels

Look through your current question list. Circle every "and." Any question with two "ands" gets split into separate questions.