The prompt
Score 10 of 10, Excellent
Use it when You have a topic that is too big and need a few focused questions, with the evidence each would need and whether you can get it in time.
- {{topic}}
- {{purpose_and_reader}}
- {{time_and_resources}}
- {{access}}
- {{deliverable}}
Full prompt and why it works
Prompt
Broad topic: {{topic}} Why I am researching it and who it is for: {{purpose_and_reader}} Time and resources I have: {{time_and_resources}} Data or access I already have: {{access}} What I must produce: {{deliverable}} Task: Help me turn the topic above into a research question I can answer with the time and access I have. A good question is narrow enough to answer with evidence, specific about who, where and when, and still matters to the people who will read the result. Format: 1. Why the topic is too broad: two or three sentences naming what makes it hard to answer. 2. Five candidate questions, each narrowed along a different line (for example one population, one place, one time period, one outcome, or a comparison). For each, a short block: Question: one sentence. What would answer it: the kind of evidence or data needed. Can I get it: Yes, Maybe or No, given my access, with the reason. Risk: what could make it unanswerable or trivial. 3. Recommendation: the one question that best fits my time, access and readers, in improved wording. 4. Out of scope: what this question deliberately leaves out, as a short list I can put in my plan. Example of one candidate's form only, on an invented topic: Question: How did average wait times at our three city clinics change after online booking began in 2024? What would answer it: booking records from before and after. Can I get it: Yes, I have access to the records. Risk: other changes in 2024 may also have affected waits. Requirements: - Under 450 words. - Do not state facts about the topic, such as statistics or what research already shows. Where knowing the existing research would change the advice, write "Check whether this has been studied" instead. - Avoid questions that can only be answered yes or no, or that assume their own answer. If my time, access or deliverable is unclear, give your best candidates and list what you would need to know to choose between them.
Variables
- {{topic}}
- The topic as you would describe it now, for example "AI in customer service".
- {{purpose_and_reader}}
- Why you are doing it and who reads the result, for example "MBA capstone, read by faculty and my employer".
- {{time_and_resources}}
- For example "six weeks, evenings only, no budget".
- {{access}}
- Data, people or systems you can get at, for example "our ticket data for 2023 to 2025". Write "none" if you have none.
- {{deliverable}}
- For example "a 3,000-word report" or "a ten-minute presentation".
Why it works
- What good looks like. Every candidate is tested against the evidence it needs and whether you can get it, so the recommended question is one you can actually finish.
- Length. Five candidates, each narrowed along a different line, show the real choices instead of five rewordings of the same question.
- What to avoid. It does not claim what existing research shows; where that matters, it tells you to check, so your plan does not rest on an invented gap.