These prompts help UX researchers use AI tools throughout the survey lifecycle — from reviewing draft questions for bias to coding open-ended responses and generating stakeholder reports. Replace [bracketed placeholders] with your project-specific details before pasting into an LLM.
Prompt 1: Review survey questions for bias and clarity
I am designing a UX survey and need you to review the following questions for quality issues.
Survey context:
- Product: [product name and description]
- Target respondents: [who they are]
- Research questions this survey should answer: [list]
Here are the draft survey questions:
[Paste all questions with their response options]
For each question, check:
1. Is it leading or loaded? (Does it push toward a particular answer?)
2. Is it double-barreled? (Does it ask about two things at once?)
3. Are the response options exhaustive and mutually exclusive?
4. Is there a "Not applicable" or "I don't know" option where needed?
5. Is the language clear at a 6th-grade reading level?
6. Does the question measure what it claims to measure?
For each issue found, provide the original question, explain the problem, and suggest a corrected version.Prompt 2: Code open-ended survey responses into themes
I have [number] open-ended responses from a UX survey. The question asked was:
"[paste the exact survey question]"
Here are the responses:
[Paste all responses, one per line, or attach as CSV]
Please:
1. Read all responses and identify the main themes (aim for 5-12 themes, not more)
2. Create a codebook: for each theme, provide a name, a description, and 2-3 example responses
3. Assign each response to one or more themes (a response can belong to multiple themes)
4. Count the frequency of each theme and rank from most to least common
5. Write a narrative summary (3-5 paragraphs) of what respondents said, organized by theme, including representative quotes
Flag any responses that seem ambiguous, contradictory, or potentially spam.Prompt 3: Design a survey from research questions
I need to create a UX survey for [product name/type].
Context:
- Research questions: [list the specific questions the survey should answer]
- Target audience: [who will take the survey]
- Distribution method: [in-product / email / panel]
- Target completion time: [5-10 minutes]
- Any standardized instruments to include: [NPS / CSAT / SUS / CES / none]
Create a complete survey including:
1. Introduction text (who is running it, purpose, estimated time, data usage)
2. Screening questions (if any segments should be filtered)
3. Core questions organized by topic, with question type and response options for each
4. Standardized instrument questions in their correct format
5. 1-2 open-ended questions placed strategically
6. Closing/thank-you text
For each question, note: the research question it maps to, the question type (single-select, multi-select, Likert, ranking, open-ended), and whether answer randomization is recommended.
Total should not exceed 25 questions.Prompt 4: Analyze survey results and generate a report
I have completed a UX survey with [number] respondents. Here is the data:
[Paste or attach the dataset — columns should include question names and response values]
Survey context: [product description, what prompted the survey, any prior survey data for comparison]
Please produce:
1. For each closed-ended question: response distribution (counts and percentages), mean score (for scales), and visualization recommendation
2. Cross-tabulation by [segment variable, e.g., user role, device type]: identify statistically significant differences between segments
3. For each open-ended question: theme coding with frequencies (see prompt 2 format)
4. Key findings summary: the 5-7 most important insights from the data
5. For each finding, use the "What / So What / Now What" framework
6. Recommended actions prioritized by impact and feasibility
Flag any data quality issues (low response rates for specific questions, suspicious patterns, segments with too few respondents for reliable comparison).