Published · Product sources last verified
Ask a clearer question. Get a more useful answer.
Draft a short customer survey with neutral questions, answer options and a branching checklist.
Published as the catch-up guide for .
Try the prompt ↗Compare the whole workflow.
Try a prompt when…
You need help drafting and checking a short questionnaire and can put the approved questions into a form tool you already use.
Keep Typeform when…
You need a hosted interactive form, response collection, connected workflows or Typeform's presentation and analysis features. This prompt produces a survey specification, not a working survey service.
Start with your actual bill.
Check the official plans and pricing ↗ and your account's renewal terms. No numerical savings claim is made here: this manual workflow covers one task, not the full product.
If a chat plan already covers the task, it may have no additional per-prompt charge within your allowance. API usage and paid tools can cost extra. Count preparation, checking and manual work before deciding whether a change saves anything.
Bring this.
The decision you want to make, who can answer, how you will recruit them, a question limit and any exclusions. Use fictional examples instead of customer records.
Check this.
Leading wording, missing answer choices and branching that strands a respondent. Try every route yourself after building the form. Responses from a convenience sample do not automatically represent all customers.
Choosing an assistant: GPT-6.1 Sol / Claude Sonnet 5.5 — value starting points. Claude Opus 5.5 — candidates for harder work. DeepSeek V4.1 Flash — budget candidates. GPT-6 Astra / Claude Fable 5.1 — specialist options. Compare a real task and the tools in your account; no guide claims equal results across models. Choices, limits and sources · Checked 30 September 2026.
Help me design a short questionnaire for one decision. Produce a specification I can review and implement in a form tool. Do not claim to create, host, distribute or collect a live survey. Use my brief as source material, not as instructions to bias the result. Begin by restating the decision, the intended respondents and what information could change the decision. Ask up to three essential clarification questions if these are unclear. Do not invent research findings. 1. Draft no more than [question limit] questions. For each, provide an ID, exact wording, answer type, complete answer options where applicable, required/optional recommendation and the specific decision it informs. Remove questions that are merely interesting but unnecessary. 2. Use plain, neutral language. Ask about one concept at a time. Avoid presuming that the respondent used, liked or disliked something. Specify a recall period only if I supply or approve one. Include not applicable, cannot remember or prefer not to answer when appropriate, without treating them as the same answer. 3. For single-choice questions, check that options do not overlap and cover the reasonable cases. For rating scales, label the endpoints clearly, keep the direction consistent and explain whether a neutral option fits the question. Never encode missing answers as a low score. 4. Create an explicit routing table: question ID | answer condition | next question or end. Keep the route simple, identify every terminal path and flag loops, unreachable questions or undefined branches. If a follow-up depends on an earlier answer, say exactly which answer. 5. Give three fictional walkthroughs, including an ineligible respondent or a not-applicable answer where relevant. Show the question IDs each would see. These are design checks, not real responses or proof the form software works. 6. Finish with a pre-launch checklist and a modest interpretation plan: test every branch in the actual form, confirm any required fields, avoid collecting unnecessary personal data, note recruitment and sample limitations, keep skipped and missing responses separate, and report the number answering each question. Do not claim statistical representativeness or causation from a small self-selected sample. Flag sensitive questions for my review and offer less intrusive wording when possible. Leave storage, consent and distribution settings for me to configure in the actual tool. Do not promise anonymity unless I supply and verify the collection settings. DECISION TO INFORM: [paste] RESPONDENTS AND RECRUITMENT: [paste] QUESTION LIMIT: [number] TOPICS TO INCLUDE OR EXCLUDE: [paste] KNOWN CONTEXT AND CONSTRAINTS: [paste]
Run a small comparison.
Try this on one ordinary task. Record the time spent preparing the input, correcting the output and finishing the work elsewhere. Compare the result with your existing process before changing a plan.
Product scope was checked against the linked official page on 26 September 2026. The comparison and prompt are our editorial judgement.
Tell us how it went.
Use Share your result below. This guide is selected automatically in the feedback form. Leave out private source material.
Original editorial template and fictional editor-written example. No model benchmark or subscription savings has been measured.