To get meaningful insights and enable the Interview Agent to ask intelligent follow-up questions, your Interview Questions, Research Goal, and Context for AI should be written with both qualitative research best practices and AI understanding in mind.
💡 Before you start: Aim for an average of 15 questions or 30 minutes per interview. Longer interviews tend to lose respondents' attention, reducing response quality and the quality of your insights.
Research Goal
Writing a Clear and Effective Activity Goal
Why it matters: The Research Goal tells the AI the purpose and scope of your research. This keeps it focused and helps it ask more relevant follow-up questions.
Best practices
- State the research topic clearly. Use explicit product, brand, feature, or behavior.
- Include the research objective. Clarify the insight you're trying to uncover.
- Define the scope and boundaries. Add constraints or exclusions to prevent tangents.
- Use plain language. Keep it concise and direct. Aim for 1 to 3 clear sentences. If you have a project brief, include only the most relevant information so the AI can easily understand what you're trying to learn.
- Treat this like briefing a human moderator. Give the AI clear, well-organized information, and it will perform better. Too much unnecessary detail can distract it from the main objective.
Formula
Action Verb + Research Focus + Purpose + (Optional Scope)
Short example
Explore how customers discover new meal delivery services and what factors influence their first purchase decision, excluding price-based considerations.
Sample Goals by Research Type:
- Concept Testing: “Evaluate how target customers perceive the appeal and clarity of our new product concept for a portable air purifier.”
- Usability: “Identify areas of confusion in the mobile app’s onboarding process for first-time users.”
- Brand Perception: “Understand how Gen Z consumers view Brand X’s sustainability efforts and whether these efforts influence purchase intent.”
- Customer Journey Mapping: “Explore the steps small business owners take from initial awareness to purchase when choosing accounting software.”
- Ad or Message Testing: “Gather reactions to the tone, clarity, and persuasiveness of the new social media campaign targeting college students.”
Longer Example:
We are studying how frequent flyers (1+ flights per month) perceive airlines' eco-friendly initiatives. The research evaluates travelers' awareness of current programs, their perceived credibility, and their impact on booking decisions. We will also identify the most effective channels and messaging for promoting sustainable travel. Ultimately, these insights will guide future marketing campaigns and program improvements.
Interview Questions
1. Ask One Thing at a Time
Why it matters: Questions that focus on a single topic are easier to answer and easier for AI to build on. When a question packs in multiple ideas or instructions, it creates confusion and makes it harder to generate useful follow-ups.
Best practices:
- Stick to one clear idea per question.
- Avoid combining multiple topics, like pricing and usability.
- Keep wording simple and free of unnecessary detail.
Examples:
- ❌ “What are your thoughts on the pricing and the usability of the app?”
- ✅ “What do you think about the app’s pricing?”
- ✅ “How would you describe the app’s usability?”
Add Context to Your Interview Questions, When Needed
Why it matters: A short setup before your main question can help ground the participant and frame their thinking. This is especially helpful when asking about past experiences or specific scenarios.
Best practices:
- Use a short, natural lead-in that focuses the participant without overloading them.
- Make sure your setup and question are clearly linked.
- Keep the entire prompt concise and free of nested clauses.
Examples:
- ❌ “Tell me what you think about the app, and whether or not you think it's something you would continue using, if the features get better, which they may in the next version.”
- ✅ “Think back to your recent shopping trip. What did you notice about the store entrance?”
- ✅ “Imagine you just opened the app for the first time. What stood out to you?”
Pro tip: If your question includes multiple instructions or conditions, re-read it aloud. If it feels like two questions or sounds overly complex, split it up.
2. Avoid Leading or Biased Wording
Why it matters: Leading questions nudge participants toward a specific answer, and AI may reinforce that bias in follow-ups.
Best practices:
- Don’t imply a value judgment or expected response.
- Avoid emotionally loaded or assumptive language.
- Replace “Do you like...” or “How good is...” with neutral alternatives.
Example:
- ❌ “Why do you love this brand’s customer service so much?”
- ✅ “How would you describe this brand’s customer service?”
3. Ask True Open-Ended Questions
Why it matters: Follow-up generation is strongest when initial questions encourage storytelling, not yes/no answers.
Best practices:
- Start with what, how, or tell me about.
- Avoid closed questions (especially starting with did, do, is, are, was).
- Don’t frame questions as binary choices.
Example:
- ❌ “Do you like using this app?”
- ✅ “What has your experience been using this app?”
4. Write Clearly for the AI to Follow
Why it matters: The AI follows logic based on what's stated. If a question is vague or inconsistent, it can lead the AI to ask confusing or irrelevant follow-ups. Clear and specific questions make it easier for the AI to stay on topic and extract meaningful insights.
Best practices:
- Refer explicitly to the thing you want to ask about (avoid words like "this" or "that" alone).
- Use the same term consistently (don’t switch from "plan" to "strategy" midstream unless they mean different things).
- Keep the structure of your questions simple and predictable.
Examples:
- ❌ “How did this make you feel?” (What is “this”?)
- ✅ “How did the sign-up process make you feel?”
Bonus Tips
| Tip | Why It Matters |
|---|---|
| Use present or recent-past tense | Helps ground responses in experiences participants can recall clearly |
| Consider natural language | AI performs best when prompts sound like real human speech |
| Test your questions aloud | If it sounds awkward or unclear when spoken, it will likely confuse both the AI and the participant |
Examples of Well-Written Interview Questions
| Poor Example | Improved Version |
|---|---|
| “Do you think the checkout was easy and intuitive?” | “What was your experience using the checkout process?” |
| “What do you think about the brand’s new logo and advertising campaign?” | “What do you think about the brand’s new logo?” + “What’s your impression of their recent ad campaign?” |
| “Why do you love using this app?” | “How would you describe your experience using the app?” |
Context for AI
Writing Effective Context for AI
Why it matters: The Context for AI explains what the Interview Agent should learn from the participant's answer. Think of it as briefing a human moderator before an interview. The clearer the context, the better the AI can ask relevant follow-up questions.
Best practices
- Keep it focused. The AI uses this context to ask smarter follow-up questions based on how the participant answers. Write a short phrase that explains why you're asking the question, so the AI understands what you're trying to learn.
- Keep it concise. If your objective has multiple goals, split it into multiple interview questions instead of trying to cover everything at once.
- Encourage depth by giving the AI one clear direction to explore rather than several unrelated topics.
Examples
Question: What dish detergent do you currently use?
Context for AI: Understand current brand loyalty and what would make participants switch.
Question: How did you feel the first time you tried this product?
Context for AI: Probe for the specific moment or detail that triggered that feeling.
Question: Tell us about your detergent habits.
Context for AI: Explore every possible reason someone might switch brands, including price, quality, trust, packaging, advertising, and word of mouth.
Why it's not recommended: This objective is too broad. It's better to split it into multiple, more focused interview questions.