Conduct a deep research project on the question below. Build a defensible evidence base rather than a persuasive answer. If you cannot access current sources, say so clearly and provide a research plan instead of implying that you verified facts. Research question: [RESEARCH QUESTION] Decision this will inform: [DECISION] Scope and definitions: [SCOPE] Audience and prior knowledge: [AUDIENCE] Geography and time period: [GEOGRAPHY AND PERIOD]
Source constraints: [SOURCE REQUIREMENTS]
Ask up to 3 clarifying questions only if a required input is missing. Then produce:
1. A research plan with subquestions, key terms, inclusion and exclusion rules, and the evidence needed to answer each subquestion.
2. An executive brief of 500 to 800 words: direct answer first, then the strongest supporting evidence, material disagreement, and decision implications.
3. An evidence table with claim, source, publication date, source type, direct support, limitations, and confidence level. Use linked citations where available.
4. A section titled "What remains uncertain" listing gaps, conflicting findings, stale evidence, and the exact next source or method needed.
5. A short recommendation framed as options and tradeoffs, not certainty.
Prefer primary sources, official datasets, peer-reviewed research, filings, standards, and original reporting. Separate facts from your interpretation. Do not manufacture citations, numbers, quotations, access, or consensus. Before answering, check that every material claim has a cited source or is explicitly labeled as inference. Also check source dates and flag evidence that may be outdated for the decision.Fill in
| Placeholder | What to enter | Example |
|---|---|---|
| [RESEARCH QUESTION] | State the precise question you need answered in one or two sentences. | What evidence supports four-day workweeks for customer support teams without reducing service quality? |
| [DECISION] | Explain the decision, recommendation, or action the research will support. | Decide whether to run a six-month pilot for a 45-person support organization. |
| [SCOPE] | Define important terms, exclusions, comparison groups, and the depth you need. | Compare reduced-hour models, not compressed 40-hour schedules; include productivity, burnout, service level, and staffing effects. |
| [AUDIENCE] | Describe who will read the work and how much subject knowledge they have. | Operations leaders who understand support metrics but do not have time to read every study. |
| [GEOGRAPHY AND PERIOD] | State the countries, regions, and date range that the research must cover. | United States, United Kingdom, and New Zealand; research from 2018 through the present. |
| [SOURCE REQUIREMENTS] | List required or prohibited source types, citation style, and recency requirements. | Prioritize peer-reviewed studies, pilot reports with methods, and official labor data; use links and flag non-peer-reviewed sources. |
How to use
- Write a decision question, not just a topic; "Should we pilot X under Y constraints?" is easier to research than "Tell me about X."
- Define what counts as evidence before asking for conclusions, especially terms such as effectiveness, risk, adoption, or cost.
- Open the cited sources for every claim that could change a costly, legal, medical, or strategic decision.
- Follow up with: "Audit the evidence table for selection bias, missing comparison groups, and claims supported only by secondary sources."
Variations
Source audit
Use when you already have a draft or list of sources and need to test its reliability.
Audit this research draft and source list: [DRAFT OR SOURCES]. The decision at stake is [DECISION]. Evaluation standards: [STANDARDS]. Return a table for each material claim with its supporting source, source quality, date, directness of support, conflicts, and needed correction. Then give a prioritized list of evidence gaps. Do not replace missing citations with plausible ones. Check for circular citations, unsupported causal claims, and sources that cite one another without primary evidence.
Competitive landscape
Use when researching companies or products for a strategic decision.
Research the competitive landscape for [MARKET] in [GEOGRAPHY] as of [DATE]. Decision: [DECISION]. Define competitors and segments as [SCOPE]. Return a market map, comparison table, evidence-backed differentiators, recent moves, and uncertainties. Cite sources for all material claims and distinguish company marketing from independent evidence. Do not infer revenue, customers, or product capabilities without support. Check source recency and flag private-company information that cannot be verified.
Literature review
Use when synthesizing research studies rather than news or market sources.
Prepare a focused literature review on [QUESTION] for [AUDIENCE]. Include studies from [DATE RANGE] and these populations or settings: [SCOPE]. Return search terms, selection criteria, a study table with design, sample, outcome, limitations, and findings, then a synthesis of agreement and disagreement. Do not treat correlation as causation or invent study details. Check for publication bias, small samples, and whether results generalize to the stated population.
Tips
- Put definitions in the request, because disagreements about a term often explain more than the headline conclusion.
- Ask for an evidence table before a recommendation; it makes weak links and stale sources visible.
- Weight a source by its direct relevance, method, incentives, and date, not just by the reputation of its publisher.
- For high-stakes decisions, use the output to locate and read primary material or consult a qualified professional; it is not a substitute for either.
FAQ
What makes research "deep" rather than a long answer?
It has a defined scope, a transparent source method, traceable claims, and explicit uncertainty. Length alone does not make research reliable.
Can I trust citations generated by AI?
Treat them as leads until you open them. Models can misstate, misread, or fabricate citations, especially when they lack source access.
How current should sources be?
That depends on the decision. Technology, regulation, pricing, and company information may need very recent sources, while foundational research may remain useful for years.