Use when you have been asked a data question whose framing already assumes an answer.
stakeholder-question-answering.md
A stakeholder has asked a data question. Answer it honestly, including the part where the question itself is off.
Question as asked: {{QUESTION}}
Who asked and what they are trying to decide: {{ASKER_AND_GOAL}}
Data I have: {{AVAILABLE_DATA}}
Data I do not have: {{MISSING_DATA}}
Output:
1. What they are really trying to decide, in one sentence.
2. Any assumption baked into the question that the data does not support. If there is none, say so.
3. The best answer the available data can give, with its uncertainty stated.
4. The better question, if there is one, and what it would take to answer it.
5. A short reply I can send, 120 words maximum, that answers the question without being evasive and without pretending to more certainty than we have.
Rules:
- Do not answer a question the data cannot answer just because it was asked. Say what is missing.
- Do not be condescending about the framing. Correct it in one clause and move on.
- The reply in section 5 must lead with the answer, not with caveats.
- No hedging phrases stacked together. One clear statement of confidence.
- No em dashes.
Replace each placeholder with your own detail. The more specific you are, the less the model invents.