Research [RESEARCH QUESTION] for [DECISION OR AUDIENCE]. My current context and assumptions are: [BACKGROUND]. Limit the work to [SCOPE, GEOGRAPHY, AND DATE RANGE]. Prefer these source types or exclude these sources: [SOURCE RULES]. Start by restating the question as 3–6 answerable subquestions. Then conduct the research and produce: 1. Executive answer: 150–250 words that answers the question directly, separating well-supported findings from informed interpretation. 2. Findings table: one row per material claim, with the claim, supporting evidence, publication date, source link, source type, and confidence level (high, medium, or low). 3. Competing evidence: explain credible disagreement, differences in definitions, samples, incentives, geography, or time period that could change the conclusion. 4. Practical implications: 3–5 actions or decisions relevant to [DECISION OR AUDIENCE], each tied to a finding. 5. Research gaps: what cannot be concluded from the available evidence and the next best sources or data to check. 6. Source list: direct links, author or publisher, title, and date. Prefer primary sources, official datasets, original studies, filings, and direct reporting. Use secondary summaries only when the original is unavailable or to show interpretation. Do not manufacture citations, statistics, publication dates, quotations, or access to paywalled material. If browsing is unavailable, say so before making claims and distinguish general knowledge from verified research. Do not treat search-result snippets as evidence. For time-sensitive facts, verify the current date and label the date checked. Before answering, self-check that every factual claim in the executive answer appears in the findings table with a source, and flag any source that is outdated, promotional, methodologically weak, or not directly relevant. Ask up to 3 clarifying questions only if a required input is missing.
Fill in
| Placeholder | What to enter | Example |
|---|---|---|
| [RESEARCH QUESTION] | Enter the specific question you need answered. | How are mid-market U.S. software companies using usage-based pricing, and what are the retention effects? |
| [DECISION OR AUDIENCE] | Enter who will use the research and the decision or purpose it supports. | Product leadership deciding whether to test a hybrid pricing model |
| [BACKGROUND] | Enter relevant facts, hypotheses, and constraints already known. | We sell workflow software to 200–2,000 employee companies and currently charge per seat. |
| [SCOPE, GEOGRAPHY, AND DATE RANGE] | Enter the topic boundaries, locations, and period the research should cover. | B2B SaaS in the United States and Canada, evidence from January 2023 onward |
| [SOURCE RULES] | Enter preferred source types, required sources, or sources to avoid. | Prioritize company filings, pricing pages, investor materials, reputable surveys with methods, and original interviews; avoid vendor blogs unless clearly labeled. |
How to use
- Replace every bracketed input, especially the decision the research must support.
- Paste the prompt into a model with web access when current sources matter.
- Open several cited sources and check that the strongest claims reflect what those sources actually say.
- Follow up with: “Turn the practical implications into a one-page decision memo and retain citations after each material claim.”
Variations
Literature review
Use this for academic or technical evidence rather than a business brief.
Create a focused literature review on [TOPIC] for [AUDIENCE OR PROJECT]. Cover studies published from [DATE RANGE] and prioritize [DATABASES OR SOURCE TYPES]. State the inclusion and exclusion criteria, then organize findings by theme rather than study-by-study summary. For each major conclusion, name the study design, population, sample size when reported, result, and limitation. Separate correlation from causation. End with unresolved questions and a bibliography with DOI or direct link. Do not invent papers or findings; flag citations you cannot verify. Ask up to 3 questions only if a required input is missing.
Competitor research
Use this when comparing companies, products, or market positions.
Research [COMPETITORS] for a team considering [DECISION]. Use public evidence current as of [DATE]. Produce a comparison table covering target customer, positioning, product capabilities, pricing model, distribution, notable proof points, and recent changes. Link every factual row to a primary source where possible, such as product documentation, pricing pages, filings, or customer announcements. Separate observed facts from your inference. Identify three gaps in our understanding, not guesses. Do not claim private revenue, customers, or strategy without evidence. Ask up to 3 questions only if a required input is missing.
Research writing
Use this after sources have been collected and you need a supported draft.
Write a [LENGTH]-word research section answering [QUESTION] using only [SOURCE NOTES]. The audience is [AUDIENCE]. Build a clear claim-evidence-reasoning structure: make a bounded claim, cite the relevant note immediately, explain why it supports the claim, and acknowledge material limitations or counterevidence. Use [CITATION STYLE]. Do not add facts, quotes, figures, or citations absent from my notes. Mark any missing evidence as [SOURCE NEEDED]. Before answering, check that every citation maps to a supplied note and that the conclusion does not overstate the evidence. Ask up to 3 questions only if required input is missing.
Tips
- Turn a broad topic into a decision question; “Should we enter Germany in 2027?” produces more useful research than “German market trends.”
- Give a date range because older evidence can be useful context while being wrong for a current operating decision.
- Ask for source-level confidence, not one confidence score for the whole brief, because the weakest claim can distort a recommendation.
- Treat a model’s source list as a starting bibliography and inspect primary evidence before relying on a consequential conclusion.
FAQ
Can AI research replace reading the sources?
No. It can organize and compare material quickly, but you should inspect the original sources behind important claims, numbers, and recommendations.
What sources should I request?
For factual claims, request primary sources first: official datasets, filings, original studies, regulations, documentation, and direct statements. Use expert analysis to add context, not as the only evidence.
How do I reduce made-up citations?
Require direct links and ask the model to flag unverified citations instead of filling gaps. Do not accept a source merely because its title sounds plausible.