Design a Complete Survey Question Set from Scratch

Work backward from research goals to create targeted questions and response options.

Prompt · 3 variables

You are a professional survey researcher who designs questionnaires and analyzes the resulting data. You always clarify the core research goals before drafting any questions.

I am planning to conduct a survey targeting {{respondent target}}. The goal of this survey is to "{{survey purpose}}", and the target length is roughly {{item count}}.

Do not start by writing the survey questions right away. First, break down the core objective into 3–5 specific "core inquiry questions." Then, write survey questions designed to answer each core inquiry question. Do not include any questions that do not tie back to these core inquiries.

Present the questionnaire in a table format with the following columns: "Number | Question | Response Type | Options / Scale | Target Core Inquiry Question". Choose response types from single choice, multiple choice, 5-point Likert scale, and open-ended text.

Avoid leading questions such as "Don't you agree that...". Do not ask double-barreled questions (two ideas in one question), and ensure that multiple-choice options are mutually exclusive and collectively exhaustive (MECE). Limit open-ended questions to a maximum of 2, and do not ask for personal identifiers like names or contact information.

After creating the table, provide a self-review section addressing these 3 checks below: ① Are there any core inquiry questions left without corresponding survey items? ② Are there any survey items not tied to a core inquiry? ③ Can respondents realistically complete this survey within 5 minutes? If you identify any issues, provide the revised final table.

Copy, then paste here · ChatGPT and Claude open with the prompt filled in Open in ChatGPT ↗Open in Claude ↗Open in Gemini ↗ Edit in builder Download classroom card

Why it is written this way

Role
You are a professional survey researcher who designs questionnaires and analyzes the resulting data. You always clarify the core research goals before drafting any questions.
Context
I am planning to conduct a survey targeting {{respondent target}}. The goal of this survey is to "{{survey purpose}}", and the target length is roughly {{item count}}.
Task
Do not start by writing the survey questions right away. First, break down the core objective into 3–5 specific "core inquiry questions." Then, write survey questions designed to answer each core inquiry question. Do not include any questions that do not tie back to these core inquiries.
Format
Present the questionnaire in a table format with the following columns: "Number | Question | Response Type | Options / Scale | Target Core Inquiry Question". Choose response types from single choice, multiple choice, 5-point Likert scale, and open-ended text.
Constraints
Avoid leading questions such as "Don't you agree that...". Do not ask double-barreled questions (two ideas in one question), and ensure that multiple-choice options are mutually exclusive and collectively exhaustive (MECE). Limit open-ended questions to a maximum of 2, and do not ask for personal identifiers like names or contact information.
Self-check
After creating the table, provide a self-review section addressing these 3 checks below: ① Are there any core inquiry questions left without corresponding survey items? ② Are there any survey items not tied to a core inquiry? ③ Can respondents realistically complete this survey within 5 minutes? If you identify any issues, provide the revised final table.

If you simply ask an AI to write survey questions, it will immediately generate a generic list. It is common to paste ten AI-generated questions into Google Forms, run the survey, and only realize afterward that none of the responses actually answer what you needed to know. Because you cannot easily re-run a live survey, this mistake is especially costly.

Setting the role as a researcher who "designs questionnaires and analyzes the data" is deliberate. Only those who analyze results understand which questions end up useless once data is collected. The third paragraph is the core of this prompt: breaking the objective down into core inquiries first ensures that irrelevant filler questions are eliminated.

In the format, the final column—"Target Core Inquiry Question"—makes the logical link explicit. If any row is missing this link, it can be deleted immediately. The constraints paragraph proactively prevents three common survey design flaws: leading questions, double-barreled questions, and overlapping/incomplete answer options. Excluding personal info avoids unneeded PII collection and protects completion rates.

The self-review in the final paragraph adds a layer of quality control. Prompting the AI to evaluate its own draft against three checks leads it to catch missing angles and adjust the table automatically. This verification is crucial even for familiar formats like satisfaction surveys, where generic flaws often slip through unnoticed.

Unfamiliar terms? See Aha AI: output-format, hallucination

Compared with a bad example

Common bad example

Make a 10-question satisfaction survey for pilates members so I can put it in Google Forms.

This produces a generic survey asking for 5-point ratings on facilities, instructors, pricing, and cleanliness. The issue is that even after running it, you still won't know why members hesitate to renew. Satisfaction scores are often generally high, and even if an item scores lower, you won't know if that is the actual reason members leave. You would have to run another survey from scratch.

Variations

When questions are ready and only ordering is needed

When questions are ready and only ordering is needed

I have prepared a draft set of survey questions targeting {{respondent target}}. The research goal is "{{survey purpose}}".

Please organize the optimal sequence for these questions. Start with easy-to-answer items, place sensitive or high-effort questions toward the end, and make sure earlier questions do not bias responses to later ones. For each question whose order you change, provide a one-line explanation of why you placed it there.

Question sequencing alone heavily influences response rates and data quality. Requiring rationale for each placement lets you make informed adjustments.

Pre-defining output charts before drafting questions

Pre-defining output charts before drafting questions

I want to conduct a {{item count}} survey targeting {{respondent target}}. The goal is "{{survey purpose}}".

Before drafting the questions, outline the specific summary tables and cross-tabulation comparisons I should expect to see once all responses are collected. Then, work backward to specify the exact survey questions needed to populate those tables.

Visualizing the final output dashboard first exposes useless questions immediately and prevents collecting data that cannot be effectively analyzed.

Model notes

Different LLMs may use slightly varying phrasing for 5-point Likert scale options. After receiving the table, adding a quick follow-up prompt like "Standardize the 5-point scale labels consistently across all rating items" makes it much easier to copy and paste into Google Forms or Typeform.

Related prompts

Last updated 2026-09-02 · Found a mistake? Let us know