# Design a survey that produces press coverage

> Use before you commission a survey, when you want the results to be pitchable rather than merely interesting.

## Fill in before running

- `{{TOPIC}}`
- `{{RESPONDENT_PROFILE}}`
- `{{SAMPLE_SIZE}}`
- `{{TARGET_PUBLICATIONS}}`
- `{{HOPED_FINDINGS}}`

## Prompt

```
You are a research director who designs surveys with press outcomes in mind, without leading respondents.

Topic: {{TOPIC}}
Audience to survey: {{RESPONDENT_PROFILE}}
Budget or sample size: {{SAMPLE_SIZE}}
Publications we want to reach: {{TARGET_PUBLICATIONS}}
Angles we hope the data supports: {{HOPED_FINDINGS}}

Produce:
1. Twelve to fifteen questions, each with: the exact wording, answer options, and the headline it could produce if the result goes one way or the other. Include at least two demographic or regional splits, because splits create local stories.
2. A flag on any question that is leading, double-barrelled, or would produce an indefensible statistic. Rewrite each flagged question.
3. A note on which of my hoped findings the survey cannot honestly test, and why.
4. The methodology paragraph to publish alongside results: sample, fielding dates placeholder, provider placeholder, weighting.

Rules:
- Do not write questions engineered to produce a predetermined answer. If a hoped finding can only be reached by a leading question, say so and refuse to write it.
- Do not invent benchmark figures from other studies.
- No em dashes in output.
```

## Getting a better result

- Regional and age splits are what turn one national story into thirty local ones; do not cut them to save budget.
- If it refuses to write a question because it is leading, that question would have been picked apart by any competent editor.
- Write the methodology paragraph before fielding, not after; it forces you to fix sample problems while you still can.

---

From the QuQi prompt library - https://www.quqi.io/prompts/survey-design-for-press