Best AI Prompts for Survey Question Generation with ChatGPT

TL;DR

• ChatGPT can generate comprehensive survey question banks covering multiple themes and formats
• MECE frameworks ensure questions are mutually exclusive and collectively exhaustive
• Persona injection techniques produce questions relevant to specific audience segments
• Iterative refinement through prompt chaining improves question quality progressively
• Rating scales and question formatting can be customized for optimal response quality
• Human review remains essential to validate question relevance and avoid bias

Introduction

Survey design is both an art and a science. The questions you ask directly determine the quality of insights you will extract. Poorly worded questions produce ambiguous responses. Leading questions introduce bias. Scales that are too coarse lose nuance; scales that are too fine confuse respondents.

ChatGPT offers powerful capabilities for generating survey questions at scale. When properly prompted, it can produce question banks that cover your topic comprehensively, use appropriate formats for different question types, and maintain consistency across questionnaire sections. The key lies in knowing how to direct its outputs effectively.

This guide provides battle-tested prompts for generating high-quality survey questions using ChatGPT. You will learn frameworks for comprehensive question generation, techniques for tailoring questions to specific audiences, and methods for ensuring the resulting questions meet rigorous quality standards.

Table of Contents

1. Foundation: Defining Survey Objectives
2. Building a Question Bank
3. MECE Framework Application
4. Audience Persona Techniques
5. Question Format Optimization
6. Scale and Rating Design
7. Question Refinement and Testing
8. Specialized Survey Types
9. FAQ

---

Foundation: Defining Survey Objectives

Clear Objective Statement

Before generating questions, establish clear survey objectives that will guide question generation.

Best Practice Prompt:

I need to create a customer satisfaction survey for [describe your product/service, e.g., "a B2B project management SaaS targeting small to mid-size marketing agencies"].

Please help me define specific, measurable survey objectives by asking targeted questions about:

1. PRIMARY GOALS: What are the 3-5 key decisions this survey will inform?
2. AUDIENCE: Who will complete this survey, and what do we already know about them?
3. KEY TOPICS: What specific aspects of the product/experience do we need feedback on?
4. SUCCESS METRICS: How will we know if this survey is successful?
5. CONSTRAINTS: Are there length limits, mandatory questions, or sensitive topics to avoid?

Please ask me 8-10 clarifying questions to ensure the survey objectives are well-defined before we begin question generation.

Topic Decomposition

Break down your survey topic into manageable components that can be systematically addressed through questions.

Best Practice Prompt:

For a survey about [TOPIC, e.g., "customer experience with our e-commerce checkout process"], please decompose the topic into 6-10 distinct sub-topics or dimensions that a comprehensive survey should cover.

For each sub-topic, provide:
1. A clear name for the sub-topic
2. 2-3 specific aspects or facets this sub-topic should address
3. Why this sub-topic matters for overall satisfaction
4. Example question formats that would work well for this sub-topic

Consider both:
- Transactional elements (ease, speed, clarity)
- Emotional elements (confidence, trust, satisfaction)
- Functional elements (features, reliability, accuracy)
- Support elements (help availability, resolution quality)

Present as a hierarchical topic map that shows relationships between sub-topics.

---

Building a Question Bank

Comprehensive Question Generation

Generate a large question bank first, then select the best questions.

Best Practice Prompt:

Generate a question bank of 40-50 survey questions for [SURVEY TOPIC, e.g., "employee satisfaction with remote work policies"].

GUIDELINES:
1. Cover multiple question formats: multiple choice, Likert scales, open-ended, ranking
2. Include both direct and indirect questioning approaches
3. Vary specificity from general satisfaction to specific feature feedback
4. Include behavioral questions (what do you do) alongside attitudinal questions (how do you feel)
5. Add demographic questions where appropriate for segmentation
6. Avoid leading questions, double-barreled questions, or assumptions

FORMAT FOR EACH QUESTION:
[Question Number] | [Question Text] | [Format] | [Sub-topic] | [Suggested Scale if applicable]

Please generate comprehensive coverage across these sub-topics:
[LIST SUB-TOPICS FROM PREVIOUS ANALYSIS]

Quality bar: Questions should be clear enough for a first-time respondent to answer without clarification.

Section-Based Generation

Organize question generation by survey sections for better coherence.

Best Practice Prompt:

Create a structured survey with 4-6 sections, generating 8-12 questions per section.

SURVEY STRUCTURE:
Section 1: [Name and objective]
Section 2: [Name and objective]
Section 3: [Name and objective]
[Continue as needed]

FOR EACH SECTION, GENERATE:
1. Section introduction text (1-2 sentences setting context)
2. 8-12 questions in recommended formats:
   - 3-4 multiple choice or dichotomous questions
   - 2-3 Likert scale questions (5 or 7 point)
   - 1-2 ranking or constant sum questions
   - 1-2 open-ended questions (limit to 2 per section maximum)
3. Transition text to next section (if applicable)

QUALITY CHECKLIST FOR EACH QUESTION:
- Is the question clear and unambiguous?
- Does it avoid double negatives or complex sentence structures?
- Is the scale appropriate for the information sought?
- Does the question avoid leading the respondent?
- Is this question answerable by the target respondent?

Please flag any questions that need revision and provide revised versions.

---

MECE Framework Application

Mutually Exclusive Question Design

Ensure questions do not overlap or capture similar information.

Best Practice Prompt:

Review this draft question list and identify any questions that overlap in what they measure:

DRAFT QUESTIONS:
[LIST 20-30 QUESTIONS]

For each overlap identified, provide:
1. Questions involved in the overlap
2. Explanation of how they measure similar things
3. Recommended resolution:
   - Combine into single question
   - Rephrase one to capture different information
   - Remove one question
4. Revised question text if applicable

Goal: Ensure each question captures distinct information that no other question captures.

Collectively Exhaustive Coverage

Verify your question set covers all important aspects.

Best Practice Prompt:

Analyze this question list for coverage gaps:

CURRENT QUESTIONS BY SUB-TOPIC:
[LIST QUESTIONS GROUPED BY SUB-TOPIC]

SUB-TOPICS TO COVER:
[LIST TARGET SUB-TOPICS]

For each sub-topic, indicate:
1. Coverage status: Fully covered / Partially covered / Not covered
2. If partially covered: What aspects are missing?
3. If not covered: Why this gap might be acceptable or problematic
4. If gap exists: Suggest 1-2 questions to address

Also identify:
- Any sub-topics with too many questions (more than 8) that might need consolidation
- Any questions that do not fit clearly into any sub-topic
- Recommendations for achieving balanced coverage

---

Audience Persona Techniques

Persona Injection

Tailor questions to specific audience segments by defining personas.

Best Practice Prompt:

Generate survey questions tailored to this specific audience persona:

PERSONA DEFINITION:
- Industry: [e.g., Healthcare, Financial Services, Retail]
- Company Size: [e.g., 50-200 employees]
- Role: [e.g., Marketing Manager, IT Director]
- Experience Level: [e.g., New user, Power user, Decision maker]
- Primary Use Case: [Specific way they use the product/service]
- Key Challenges: [Top 3 challenges they face]
- Tech Savviness: [Low / Medium / High]

QUESTION TAILORING:
1. Use industry-appropriate terminology
2. Reference use cases relevant to this persona
3. Adjust question complexity to match their expertise level
4. Include questions about their specific pain points
5. Frame benefits in terms relevant to their role

Please generate 15-20 questions specifically optimized for this persona, with notes explaining how each question was tailored.

Multiple Persona Variations

Generate question variations for different audience segments.

Best Practice Prompt:

For a survey being sent to multiple audience segments, generate segment-specific question variations:

SEGMENT A: IT Administrators
- Focus: Technical implementation, integration capabilities, security features
- Technical sophistication: High
- Key concerns: Compatibility, uptime, admin controls

SEGMENT B: End Users
- Focus: Ease of use, daily workflow integration, learning curve
- Technical sophistication: Medium
- Key concerns: Productivity, accessibility, support

SEGMENT C: Executives/Decision Makers
- Focus: ROI, strategic value, vendor relationship
- Technical sophistication: Low
- Key concerns: Business outcomes, cost-benefit, risk

FOR EACH SEGMENT, GENERATE:
1. 5-6 questions tailored to their specific focus and sophistication level
2. Segment-specific terminology and framing
3. Questions that address their unique concerns
4. Open-ended questions that would reveal their perspective

Also generate 8-10 questions common to all segments for cross-segment comparison.

---

Question Format Optimization

Multiple Choice Excellence

Generate high-quality multiple choice questions.

Best Practice Prompt:

Generate multiple choice questions following best practices:

FORMAT REQUIREMENTS:
1. Question stem: Clear, complete question ending with question mark
2. Options: 4-6 mutually exclusive choices
3. Exhaustive: One "None of the above" or "Other" option where appropriate
4. Balanced: For agree/disagree questions, balanced positive and negative options
5. Non-overlapping: No option should be subsumed by another

STYLE GUIDELINES:
- Keep all options similar in length and structure
- Avoid "all of the above" and "none of the above" in same question
- Use parallel construction across options
- Ensure no double-barreled options (options combining two ideas)

QUESTION TOPICS:
[LIST 8-10 TOPICS]

Please generate 2 multiple choice questions per topic, totaling 16-20 questions.

Open-Ended Question Design

Craft open-ended questions that yield rich, actionable responses.

Best Practice Prompt:

Design open-ended questions that will generate detailed, actionable responses:

OPEN-ENDED BEST PRACTICES:
1. Be specific about the type of response you want
2. Limit to one clear topic per question
3. Use "what" and "how" questions rather than "why" when possible
4. Provide context about why you are asking
5. Avoid leading or assuming

POOR EXAMPLE:
"Why do you like our product?" (Assumes they do like it, vague)

BETTER EXAMPLE:
"What specific aspects of our product or service have met your expectations, and what has fallen short?" (Specific, allows positive/negative, clear scope)

Please generate 8-10 open-ended questions for [TOPIC] following these guidelines.

For each question, provide:
1. The question text
2. Why this question will yield valuable information
3. How you would categorize/analyze the responses
4. Any instructions to include for respondents

---

Scale and Rating Design

Likert Scale Optimization

Create consistent, meaningful Likert scales.

Best Practice Prompt:

Design a set of 5-point Likert scales for this survey:

SCALE CONSTRUCTION RULES:
1. Labels should be symmetric around neutral point
2. Scale points should be equidistant in meaning
3. Include clear labels for each point (not just numbers)
4. Avoid middle "neutral" option when you want forced choice
5. Ensure labels are unambiguous

SUGGESTED SCALE TYPES TO INCLUDE:
- Agreement scale (Strongly Disagree to Strongly Agree)
- Frequency scale (Never to Always)
- Satisfaction scale (Very Dissatisfied to Very Satisfied)
- Importance scale (Not at All Important to Extremely Important)
- Likelihood scale (Very Unlikely to Very Likely)

FOR EACH SCALE TYPE NEEDED:
1. Exact wording for each scale point (1-5)
2. Whether to include N/A option
3. Example question using this scale
4. Common pitfalls to avoid

TOPICS REQUIRING SCALES:
[LIST 8-10 TOPICS]

Please provide complete scale definitions ready for immediate use in your survey.

Custom Scale Development

Create specialized scales for unique measurement needs.

Best Practice Prompt:

I need a custom scale to measure [SPECIFIC CONSTRUCT, e.g., "perceived value for money" in a B2B SaaS context].

SCALE REQUIREMENTS:
1. What is the construct you are measuring? [Provide definition]
2. What are the extremes of this construct? [Low/High definitions]
3. What behaviors or attitudes indicate each level?
4. Is this a single dimension or multi-faceted construct?

DEVELOPMENT STEPS:
1. Generate 6-8 candidate items (statements that would be rated)
2. Test items for clarity and redundancy
3. Select final 4-5 items for scale
4. Define scale labels (1-5 or 1-7)
5. Provide usage instructions

Example item formats:
- "The pricing is justified by the value I receive" (Agreement format)
- "How would you rate the value you receive relative to price paid?" (Rating format)

Please develop a complete, validated-scale-ready measurement instrument.

---

Question Refinement and Testing

Cognitive Testing Prompts

Use ChatGPT to identify potential respondent confusion.

Best Practice Prompt:

Review these survey questions and identify potential comprehension issues:

QUESTIONS TO REVIEW:
[LIST 15-20 QUESTIONS]

COGNITIVE INTERVIEWING FRAMEWORK:
For each question, evaluate:

1. COMprehension: Can a respondent clearly understand what is being asked?
   - Are there ambiguous words or phrases?
   - Does the question require knowledge the respondent may not have?
   - Are there complex grammatical structures?

2. Retrieval: Can respondents retrieve the information needed?
   - Does the question ask about memorable experiences or vague impressions?
   - Is the timeframe clear and appropriate?
   - Might response bias affect accuracy?

3. Judgment: Can respondents form a rating or answer?
   - Are scales clear and interpretable?
   - Do respondents have the information to make the requested judgment?
   - Is there a clear reference point?

4. Response: Can respondents formulate and communicate their answer?
   - Are options exhaustive?
   - Might social desirability bias affect answers?
   - Would respondents be comfortable answering honestly?

Please flag questions needing revision and provide specific suggestions.

Pilot Test Analysis

Analyze potential pilot test results before fielding.

Best Practice Prompt:

We plan to field this survey to [SAMPLE SIZE, e.g., 500 customers] with a [COMPLETION RATE, e.g., 35%] expected response rate.

Please analyze potential issues:

1. EXPECTED COMPLETION RATE CONCERNS:
   - Survey length: [NUMBER OF QUESTIONS]
   - Which questions might cause drop-off?
   - How could length be reduced without losing critical data?

2. RESPONSE QUALITY CONCERNS:
   - Which open-ended questions might get low-quality responses?
   - Which scales might show low variance (everyone answers same)?
   - Are any questions potentially offensive or uncomfortable?

3. BIAS CONCERNS:
   - Which questions might have social desirability bias?
   - Are certain segments less likely to complete the survey?
   - Could question order affect responses (primacy/recency effects)?

4. VALIDITY CONCERNS:
   - Will the data collected answer our research questions?
   - What analyses can we run to validate findings?
   - Are there alternative explanations for expected patterns?

Please provide recommendations to address each concern identified.

---

Specialized Survey Types

NPS and Customer Satisfaction Surveys

Generate specialized questions for common survey types.

Best Practice Prompt:

Create a comprehensive NPS (Net Promoter Score) survey with supporting questions:

NPS CORE QUESTION:
"Please indicate how likely you are to recommend [Company/Product] to a friend or colleague" on 0-10 scale.

SUPPLEMENTING QUESTIONS TO INCLUDE:
1. Three driver questions: What specific aspects drive your likelihood to recommend?
2. Detractor probe: What would need to change for you to become a promoter?
3. Promoter confirmation: What specifically did we do well?
4. Segment questions: [Appropriate demographic/firmographic]

QUESTION BANK (12-15 questions total):
- 1 NPS core question
- 3-4 driver/detractor questions
- 3-4 satisfaction questions on key features
- 2-3 demographic/segmentation questions
- 2-3 follow-up questions based on NPS score

Please structure as a complete, field-ready survey with logic skips.

Employee Engagement Surveys

Design questions for measuring employee satisfaction and engagement.

Best Practice Prompt:

Generate a comprehensive employee engagement survey question bank:

FRAMEWORK: Gallup Q12 adapted context
CATEGORIES TO COVER:
1. Role clarity and expectations
2. Resources and support
3. Recognition and appreciation
4. Growth and development opportunities
5. Manager relationship and feedback
6. Team collaboration and cohesion
7. Company direction and leadership
8. Work-life balance
9. Compensation and benefits perception
10. Overall satisfaction and retention intent

FOR EACH CATEGORY:
- 2-3 Likert scale questions (5-point agreement scale)
- 1 open-ended follow-up question
- Category introduction text

TOTAL: 20-25 scale questions + 8-10 open-ended questions

Include:
- Anonymity assurances language
- Completion estimate (5-7 minutes)
- Company-specific customization points

---

FAQ

How many questions should a survey include?

Optimal survey length depends on audience and topic complexity. For customer satisfaction surveys, 15-25 questions work well with 35-50% completion rates. Employee surveys can be slightly longer (25-40 questions) given engaged populations. The key is to prioritize questions that directly inform decisions and minimize those that are “nice to know” rather than essential.

How do I avoid leading questions in AI-generated surveys?

Leading questions subtly suggest a preferred answer. To avoid them, explicitly instruct ChatGPT to generate balanced questions. Include examples of leading questions to avoid, such as “How much did you love our excellent customer service?” which assumes the service was excellent. Better phrasing: “How would you rate our customer service?” with neutral scale options.

Should I use 5-point or 7-point Likert scales?

5-point scales work well for most survey contexts, offering simplicity and sufficient variation. Use 7-point scales when you need finer discrimination, such as in academic research or when comparing subtle differences. The key is consistency within your survey and appropriate labeling of each point.

How do I handle sensitive topics in survey questions?

For sensitive topics, use indirect questioning techniques rather than direct questions. Instead of “Have you experienced discrimination?”, ask about perceptions of workplace fairness. Include anonymity assurances, use neutral language, and consider whether a third-party administrator would increase honest responses.

Can ChatGPT generate questions for B2B surveys?

Yes, with appropriate context. B2B surveys require industry-specific terminology, longer time horizons (purchase decisions take months, not minutes), and personas representing different decision-makers. Provide ChatGPT with industry context, company size considerations, and the specific B2B buyer journey stage you are researching.

How do I validate that AI-generated questions are high quality?

Pilot test with a small sample (20-50 respondents) before full deployment. Monitor completion rates, response variance, and open-ended response quality. Conduct cognitive interviews with 5-10 respondents to identify comprehension issues. Track which questions produce unusable data (straight-lining, gibberish, N/A responses) and refine or replace them.

Can I use ChatGPT to translate surveys for international audiences?

Yes. ChatGPT has strong multilingual capabilities. Generate questions in your source language, then prompt for translation to target languages. Have translations back-translated to verify accuracy. Be aware that some survey concepts do not translate directly across cultures and may require adaptation rather than literal translation.

---

Conclusion

ChatGPT transforms survey question generation from a time-consuming manual process into an efficient collaborative exercise. The key to success lies in providing clear context about your audience, objectives, and topic structure, then iteratively refining outputs until they meet your quality standards.

Key Takeaways:

• Define clear survey objectives before generating questions
• Build comprehensive question banks, then select the best questions
• Apply MECE frameworks to ensure mutually exclusive, collectively exhaustive coverage
• Tailor questions to specific audience personas
• Use appropriate question formats for different information types
• Test and refine questions through cognitive testing and pilot studies
• Maintain consistency in scale construction and labeling

Your next step is to define your survey objectives clearly and begin generating questions using the prompts in this guide. Start with topic decomposition, build your question bank, then refine iteratively until you have a field-ready instrument.